oe INTERN CT EDITION / Wl a 2 ACOURSEINFUZZY PI 1 ee a Ce eT")A Course in Fuzzy Systems and Control Li-Xin Wang Prentice-Hall International, Inc.Contents Preface 1 Introduetion 1.1 Why Fuzzy Systems? 1.2 What Are Fuzzy Systems? 1.3. Where Are Fuzzy Systems Used and How? 1.3.1 Fuzzy Washing Machines 1.3.2 Digital Image Stabilizer 133° Fuzy Systems in Cars 1.3.4 Fuzzy Control of a Cement Kiln 1.3.5 Fuzzy Control of Subway Train 14 What Are the Major Research Fields in Fuzzy Theory? A Brief History of Fuzzy Theory and Applications 1.5.1 The 1960s: The Beginning of Fuzzy Theory 1.5.2 The 1970s: Theory Continued to Grow and Real Applications Appeared 1.5.3 The 1980s: Massive Applications Made a Difference 1.5.44 The 1900s: More Challenges Remain 1.6 Summary and Further Readings LT Exercises ‘The Mathematics of Fuzzy Systems and Control Fuzzy Sets and Basic Operations on Fuzzy Sets 2.1 From Classical Sets to Fuzzy Sets 2.2 Basie Concepts Associated with Fuzzy Set 13 ul 1% re 16 19 20 2023. Operations on Fuzzy Sets 2A Summary and Further Readings 25. Exercises Further Operations on Fuzzy Sets 3.1 Fuzzy Complement 32. Fuzzy Union—The $-Norms 3.3 Fuzty Intersection—The T-Norms 3A Averaging Operators 3.5 Summary and Further Readings 3.6 Exercises ‘Fuzzy Relations and the Extension Principle 4.1 From Classical Relations to Fuzzy Relations 4.1.1 Relations 4.1.2 Projections and Cylindrie Extensions 4.2. Compositions of Fuzzy Relations 4.3. The Extension Principle 44 Summary and Further Readings 45° Exercises Linguistic Variables and Fuzry IF-THEN Rules 5.1 From Numerical Variables to Linguistic Variables 5.2 Linguistic Hedges 5.3 Fuzzy IF-THEN Rules 53.1. Furzy Propositions 5.3.2 Interpretations of Fuzzy IF-THEN Rules 54 Summary and Further Readings 5.5 Exercises Fuzzy Logic and Approximate Reasoning 6.1 From Classical Logic to Fuzzy Logie 6.11 Short Primer on Classical Logic 6.1.2 Basic Principles in Fuzzy Logic 62. The Compositional Rule of Inference 6.3. Properties of the Iinplication Rules 6.3.1 Generalized Modus Ponens Contents 32 32 34 35 35 aL ry 46 46 30 61 2 e m 73. 3 3 B 78 a a1CONTENTS 6.3.3 Generalized Hypothetical Syllogism 64 Summary and Further Readings 6.5 Bxercises II Fuzzy Systems and Their Properties 7 Fuzzy Rule Base and Fuzzy Inference Engine TA Fuzzy Rule Base TAL Structure of Fuzzy Rule Base 7.1.2 Properties of Set of Rules 72. Fwzay Inference Engine 7.2.1 Composition Based Inference 7.2.2 Individual-Rule Based Inference 7.2.8 The Details of Some Inference Engines 7.3. Summary and Further Readings TA. Exercises 8 Fuzzifiers and Defuzzifiers 8.1 Fuzziliers 82. Defuaeifiers 8.2.1 center of gravity Defuzzifier 82.2 Center Average Defuzzificr 823 Maximum Defuzzifier 8.2.4 Comparison of the Defuzzfiers 83. Summary and Further Readings 84 Exercises 9 Fuzzy Systems as Nonlinear Mappings 9.1 The Formulas of Some Classes of Fuzzy Systems 9.1.1 Fuzzy Systems with Center Average Defuzifier 9.1.2 Fuzzy Systems with Maximum Defuzitier 9.2. Fuzzy Systems As Universal Approximators 9.3 Summary and Further Readings 94 Exercises 10 Approximation Properties of Fuzzy Systems I 10.1 Preliminary Concepts 10.2 Design of Fuzzy System vil SER 89 m1 1 92 97 104 104 105 105 108 109 10 m2 a2 us 16 8 us us 122 4 137 wr 128 129 asiContents Approximation Accuracy of the Fuzzy System 133 10.4 Summary and Further Readings 138 10.5 Beercises 138 11 Approximation Properties of Fuzzy Systems II 140 11.1 Fuzzy Systems with Second-Order Approximation Accuracy 140 11.2 Approximation Accuracy of Fuzzy Systems with Maximum Defuzzifier145 11.3 Summary and Further Readings 49 114 Exercises 149 III Design of Fuzzy Systems from Input-Output Data 151. 12 Design of Fuzzy Systems Using A Table Look-Up Scheme 153 12.1 A Table Look-Up Scheme for Designing Fuzzy Systems from Input- Output Pairs 153, 122 Application to Truck Backer-Upper Control 187 123 Application to Time Series Prediction 161 124 Summary and Further Readings 166 125 Exercises and Projects 166 13 Design of Fuzzy Systems Using Gradient Descent Training 168 18.1 Choosing the Structure of Fuzzy Systems 168 13.2 Designing the Parameters by Gradient Descent 169 183 Application to Nonlinear Dynamic System Identification 1m 13.3.1 Design of the Identifier in 13.32 Initial Parameter Choosing 174 133 Simulations 175 194 Summary and Further Readings 176 135 Exercises and Projects 178 14 Design of Fuzzy Systems Using Recursive Least Squares 180 M1 Design of the Fuzzy System 180 142 Derivation of the Recursive Least Squares Algorithm 182 143 Application to Equalization of Nonlinear Communication Channels 183, 14.3.1 The Equalization Problem and Its Geometric Formulation 183 14.32 Application of the Fuzzy System to the Equalization Problem 186 M4 Summary and Further Readings 190 14.5 Exercises and Projects, 190CONTENTS 15 Design of Fuzzy Systems Using Clustering 15.1 An Optimal Fuzzy System 15.2 Design of Fuzzy Systems By Clustering 15.3 Application to Adaptive Control of Nonlinear Systems 154 Summary and Further Readings 15.5 Exercises and Projects IV Nonadaptive Fuzzy Control 16 The Trial-and-Brror Approach to Fuzzy Controller Design 16.1 Fuzzy Control Versus Conventional Control 16.2 The Trial-and-Error Approach to Fuzzy Controller Design 16.3 Case Study I: Fuzzy Control of Cement Kiln 163.1 The Cement Kiln Process 16.32 Fuzzy Controller Design for the Cement Kiln Process 16.33 Implementation 164 Case Study Il: Fuzzy Control of Wastewater Treatment Process 16.4.1 The Activated Shudge Wastewater Treatment Process 16.4.2 Design of the Fuzzy Controller 16.5 Summary and Further Readings 16.6 Exercises 17 Fuzzy Control of Linear Systems I: Stable Controllers ITA Stable Fuzzy Control of Single-Input-Single-Output Systems 17.1.1 Exponential Stability of Fuzzy Control Systems 17.1.2 Input-Output Stability of Fuzzy Control Systems 17.2 Stable Fuzzy Control of Multi-Input-Multi-Output Systems 17.2.1. Exponential Stability 17.22 Input-Output Stability 17.3 Summary and Further Readings 174 Exercises 18 Fuzzy Control of Linear Systems trollers 18.1 Optimal Fuzzy Control 18.11 The Pontryagin Minimum Principle 18.1.2 Design of Optimal Fuzzy Controller 18.1.8 Application to the Ball-and-Beam System 192 192 193 199 203 203 205 206 206 208 208 208 210 212 24 214 215 aut ar 219 219 221 223 225 225 227 228 28 Optimal and Robust Con- 280 230 231 231 234x contents 18.2 Robust Fuzzy Control 235 18.3 Summary and Further Readings 236 184 Exercises 237 19 Fuzzy Control of Nonlinear Systems I: Sliding Control 238 19.1 Fuzzy Control As Sliding Control: Analysis 238 19.1.1 Basie Principles of Sliding Control 238 19.1.2. Analysis of Fuzzy Controllers Based on Sliding Control Principle241 19.2 Fuzzy Control As Sliding Control: Design 2a 192.1 Continuous Approximation of Sliding Control Law 2a 19.22 Design of Fuzzy Controller Based on the Smooth Sliding Con- trol Law 4 19.3 Summary and Further Readings m7 19.4 Exercises 7 20 Fuzzy Control of Nonlinear Systems II: Supervisory Control 249 20.1 Multi-level Control Involving Fuzzy Systems 49 20.2 Stable Fuzzy Control Using Nonfuzzy Supervisor 251 20.2.1. Design of the Supervisory Controller 251 20.2.2 Application to Tnverted Pendulum Balancing, 254 20.3 Gain Scheduling of PID Controller Using Fuzzy Systems 257 20.3.1 The PID Controller 257 203.2 A Fuzzy System for Turning the PID Gains 258 204 Summary and Further Readings 263 20.5 Exercises 264 21 Fuzzy Control of Fuzzy System Models 265 211 The Takagi-Sugono-Kang Fuzzy System 265 21.2 Closed-Loop Dynamics of Fuzzy Model with Furey Controller 266 21.3 Stability Analysis of the Dynamic TSK Fuzzy System 269 214 Design of Stable Fuzzy Controllers for the Fuzzy Model 273 21.5 Summary and Further Readings 275 21.6 Exercises 276 22 Qualitative Analysis of Furry Control and Hierarchical any Sys tems 2% 22.1 Phase Plane Analysis of Fuzzy Control Systems 217 22.2 Robustness Indices for Stability 280 22.2.1 The One-Dimensional Case 281CONTENTS 22.2.2 The n-Dimensional Case 22.3 Hierarchical Fuzzy Control 22.3.1 ‘The Curse of Dimensionality 22.3.2 Construction of the Hierarchical Fuzzy System 22.3.3. Properties of the Hierarchical Fuzzy System 22.4 Summary and Further Readings 22.5. Exercises V_ Adaptive Fuzzy Control 28 Basic Adaptive Fuzzy Controllers T 23.1 Classification of Adaptive Fuzzy Controllers 23.2 Design of the Indirect Adaptive Fuzzy Controller 23.2.1 Problem 25.22 Design of the Fuzzy Controller 25.2.3 Design of Adaptation Law 28:3 Application to Inverted Pendulum Tracking Control 284 Summary and Further Readings 23.5 Exercises 24 Basie Adaptive Fuzzy Controllers IL 24.1 Design of the Direct Adaptive Fuzzy Controller 24.1.1 Problem Specification 24.1.2 Design of the Fuzzy Controller 24.1.3 Design of Adaptation Law 24.14 Simulations 24.2 Design of the Combined Direct/Indirect Adaptive Fuzzy Controller 24.2.1 Problem Specification 24.2.2 Design of the Fuzzy Controller 24.2.3 Design of Adaptation Law 24.2.4 Convergence Analysis 243 Summary and Further Readings 244 Exercises 25 Advanced Adaptive Fuzzy Controllers I 25.1 State Boundedness By Supervisory Control 2.1.1 For Indirect Adaptive Fuzzy Control System 282 284 284 285 286 287 288 289 291 BeR8e8 302 302 304 304 305 si 31 312 313 315 315 si 317 3iT25.1.2 For Direct Adaptive Fuzzy Control System 25,2 Parameter Boundedness By Projection 25.2.1 For Indirect Adaptive Fuzzy Control System 25.2.2 For Direct Adaptive Fuzzy Control System 25.3 Stable Direct Adaptive Fuzzy Control System 25.3.1 Stability and Convergence Analysis 25.2 Simulations 254 Summary and Further Readings 25.5 Exercises 26 Advanced Adaptive Fuzzy Controllers I 26.1 Stable Indirect Adaptive Fuzzy Control System 26.1.1 Stability and Convergence Analysis 26.1.2 Nonlinear Parameterization 26.2 Adaptive Fuzzy Control of General Nonlinear Systems 26.2.1 Intuitive Concepts of Input-Output Linearization contents 319 320 320 322 323 323 325 327 327 328 328 328 320 331 392 26.2.2 Design of Adaptive Fuzzy Controllers Based on Input-Output inearization 26.2.8. Application to the Ball-and-Beam System 26.3 Summary and Further Readings 26.4 Exercises VI Miscellaneous Topics 27 The Fuzzy C-Means Algorithm 27.1 Why Fuzzy Models for Pattern Recognition? 27.2 Hard and Fuzzy c-Partitions 27.3 Hard and Fuzzy c-Means Algorithms 34 335 339 339 341 342 342 27.3.1 Objective Function Clustering and Hard Means Algorithm 345 27.3.2 The Fweay c-Means Algorithm 274 Convergence of the Fuzzy c-Means Algorithm 27.5 Summary and Further Readings 27.6 Bxercises 28 Fuzzy Relation Equations 28.1 Introduction 28.2 Solving the Fuzzy Relation Equations 28.8 Solvability Indices of the Fuzzy Relation Equations MT 350 352 352 54 354 34 358CONTENTS 28.3.1 Equality Indices of Two Fuzzy Sets 28.3.2 The Solvability Indices 284 Approximate Solution—A Neural Network Approach 28.5 Summary and Further Readings 28.6 Exercises 29 Fuzzy Arithmetic 29.1 Fuzey Numbers and the Decomposition Theorem 29.2 Addition and Subtraction of Fuzzy Numbers 29.2.1 The a-Cut Method 29.2.2 The Extension Principle Method 29.3 Multiplication and Division of Fuzay Numbers 29.3.1 The a-Cut Method 29.3.2 The Extension Principle Method 294 Fuzzy Equations 29.5 Fuzzy Ranking 29.6 Summary and Further Readings 29.7 Exercises 80 Fuzzy Linear Programming 30.1 Classification of Fuzzy Linear Programming Problems 30.2 Linear Programming with Fuzzy Resources 30.3 Linear Programming with Fuzzy Objective Coefficients 30.4 Linear Programming with Fuzzy Constraint Coefficients 30.5 Comparison of Stochastic and Fuzzy Linear Programming 30.6 Summary and Further Readings 30.7 Exercises 31 Possibility Theory 31.41 Introduction 31.2 The Intuitive Approach to Possibility 31.2.1 Possibility Distaibutions and Possibility Measures 31.2.2 Marginal Possibility Distribution and Noninteractiveness 31.23 Conditional Possibility Distribution 31.3 The Axiomatic Approach to Possibility 31.3.1 Plansibility and Belief Measures 31.3.2 Possibility and Necessity Measures 81.4 Possibility versus Probability 368 309 370 372 372 373 374 376 378 379 381 381 384 385 387 388 300 300 393, 303 304 304 396 397 307 308xiv contents 314.1 The Endless Debate 400 31.4.2 Major Differences between the Two Theories 401 3143 How to View the Debate from an Engineer's Perspective 402 31.5 Summary and Further Readings 403 316 Bxercises 403 Bibliography 405 Index 419Preface ‘The fold of fuzzy systems and control has been making rapid progress in recent years. Motivated by the practical success of fuzzy control in consumer products ‘and industrial process control, there has been an increasing amount of work on the rigorous theoretical studies of fuzzy systems and furry control. Researchers are trying to explain why the practical results are good, systematize the existing approaches, and develop more powerful ones. As a result of these efforts, the whole Picture of fuzzy systems and fuzzy control theory is becoming clearer. Although there are many books on fuzzy theory, most of them are either research monographs that concentrate on special topies, oF collections of papers, or books on fuzzy math- ‘ematics. We desperately need a real textbook on fuzzy systems and control that provides the skeleton of the field and surumatizes the fundamentals ‘This book, which is based on a course developed at the Hong Kong University of Science and Technology, i intended as a textbook for graduate and senior students, ‘and as a self-study book for practicing engineers. When writing this book, we required that it be: ‘¢ Well-Structured: This book is not intended as a collection of existing results ‘on fuzzy systems and fuzzy control; rather, we first establish the structure that a reasonable theory of fuzzy systems and fuzzy control should follow, and then fil in the details. For example, when studying fuzzy control systems, we should consider the stability, optimality, and robustness of the systems, and classify the approaches according to whether the plant is linear, nonlinear, or ‘modeled by fuzzy systems. Fortunately, the major existing results fit very well into this structure and therefore are covered in detail in this book. Because the field is not mature, as compared with other mainstream fie, there are holes in the structure for which no results exist. For these topics, we either provide our preliminary approaches, or point out that the problems are open. * Clear and Precise: Clear and logical presentation is crucial for any book, especially for book associated with the word “fuzzy.” Fuzzy theory itself is precise; the “fuzziness” appears in the phenomena that fuzzy theory triesxvi Preface to study. Once a fuzzy description (for example, “hot day”) is formulated in terms of fuzzy theory, nothing will be fuzzy anymore. We pay special attention to the use of precise language to introduce the concepts, to develop the approaches, and to justify the conclusions. ‘+ Practical: We recall that the driving force for fuzzy systems and contro! is practical applications. Most. approaches in this book are tested for prob- Joms that have practical significance. In fact, a main objective of the book is to teach students and practicing engineers how to use the fuzzy systems approach to solving engineering problems in control, signal processing, and. ‘communications ‘* Rich and Rigorous: This book should be intelligently challenging for stu- dents, In addition to the emphasis on practicality, many theoretical results are given (which, of course, have practical relevance and importance). All the theorems and lemmas are proven in a mathematically rigorous fashion, and some effort may have to be taken for an average student to comprehend the details ‘# Easy to Use as Textbook: To facilitate its use as a textbook, this book is ‘written in such a style that each chapter is designed for a one and one-half hour lecture. Sometimes, three chapters may be covered by two lectures, or vice versa, depending upon the emphasis ofthe instructor and the background of the students. Bach chapter contains some exercises and mini-projects that {form an integrated part of the text. ‘The book is divided into six parts. Part I (Chapters 2-6) introduces the fun- damental concepts and principles in the general feld of fuzzy theory that are par- ticularly useful in fuzzy systems and fuzzy control. Part IK (Chapters 7-11) studies the fuzzy systems in detail. The operations inside the fuzzy systems are carcfully analyzed and certain properties of the fuzzy systems (for example, approximation capability and accuracy) are studied. Part IIT (Chapters 12-15) introduces four methods for designing fuzzy systems from sensory measurements, and all these methods are tested for a number of control, signal processing, or communication problems. Part IV (Chapters 16-22) and Part V (Chapters 23-26) parts concentrate ‘n fuzzy control, where Part IV studies nonadaptive fuzzy control and Part V stud- jes adaptive fuzzy control. Finally, Part VI (Chapters 27-31) reviews a number of topies that are not included in the main structure of the book, but are important and strongly relevant to fuzzy systems and fuzzy control. ‘The book can be studied in many ways, according to the particular interests of the instructor or the reader. Chapters 1-15 cover the general materials that can be applied to a variety of engineering problems. Chapters 16-26 are more specialized in control problems. If the course is not intended as a control course, then some ‘materials in Chapters 16-26 may be omitted, and the time saved may be used for ‘a more detailed coverage of Chapters 1-15 and 27-81. On the other hand, if itPreface xvii is a control course, then Chapters 16-26 should be studied in detail. The book also can be used, together with a book on neural networks, for a course on neural networks and fuzay systems. In this case, Chapters 1-15 and selected topies from Chapters 16-31 may be used for the fuzzy system half of the course. If a practicing ‘engineer wants to learn fuzzy systems and fuzzy control quickly, then the proofs of the theorems and lemmas may be skipped. ‘This book has benefited from the review of many colleagues, students, and friends. First of all, T would like thank my advisors, Lotfi Zadeh and Jerry Mendel, for their continued encouragement. I would like to thank Karl Astrém for sending. his student, Mikael Johansson, to help me prepare the manuscript during the sum- ‘mer of 1995. Discussions with Kevin Passino, Frank Lewis, Jyh-Shing Jang, Hua ‘Wang, Hideyuki Takagi, and other researchers in fuzzy theory have helped the or- ganization of the materials. The book also benefited from the input of the students ‘who took the course at HKUST. Support for the author from the Hong Kong Research Grants Council was greatly appreciated, Finally, I would like to express my gratitude to my department at HKUST for providing the excellent research and teaching environment. Especially, I would like to thank my colleagues Xiren Cao, Zexiang Li, Li Qiu, Erwei Bai, Justin Chuang, Philip Chan, and Kwan-Fai Cheung for their collaboration and critical remarks on various topics in fuzzy theory. LiXin Wang ‘The Hong Kong University of Science and TechnologyPrefaceChapter 1 Introduction 1.1 Why Fuzzy Systems? According to the Oxford English Dictionary, the word “fuzzy" is defined as “blurred, indistinct; imprecisely defined; confused, vigue.” We ask the reader to disregard this definition and view the word “fuzzy” asa technical adjective. Specifically, fuzzy systems are systems to be precisely defined, and fuzzy contr i a special kind of nonlinear control that also will be precisely defined. ‘This is analogous to linear systems and control where the word “Linear” isa technical adjective used to specify “ystems and control” the same is true for the word “fuzzy.” Essentially, what we want to emphasize is that although the phenomena that the fuzzy systems theory characterizes may be fuzzy, th theory itself i precise. In the literature, there are two kinds of justifcati for fuzzy systems theory: ‘© The real world is too complicated for precise descriptions to be obtained, ‘therefore approximation (or fuzziness) must be introduced in order to obtain a reasonable, yet trackable, model ‘+ As we move into the information era, human knowledge becomes increasingly important. We need a theory to formulate human knowledge in a systematic ‘manner and put it into engineering systems, together with other information like mathematical models and sensory measurements. ‘The first justification is correct, but does not characterize the unique nature of fuzzy systems theory. In fact, almost all theories in engineering characterize the real world in an approximate manner. For example, most real systems are nonlinear, bbut we put a great deal of effort in the study of linear systems, A good engineering. theory should be precise to the extent that it characterizes the key features of the real world and, at the same time, is trackable for mathematical analysis. In this aspect, fuzzy systems theory does not differ from other engineering theories, ‘The second justification characterizes the unique feature of fuzzy systems theory land justifies the existence of fuzzy systems theory as an independent. branch in 12 Ineroduction Ch. cengincoring. As a general principle, a good engineering theory should be eapable of making use of all available information effectively. For many practical systems, important information comes from two sources: one source is human experts who describe their knowledge about the system in natural languages; the other is sensory ‘measurements and mathematical models that are derived according to physical laws. An important task, therefore, is to combine these two types of information into system designs. To achieve this combination, a key question is how to formulate ‘numan knowledge into a similar framework used to formulate sensory measurements and mathematical models. In other words, the key question is how to transform ‘a human knowledge base into a mathematical formula. Bssentially, what a fuzzy system does is to perform this transformation. In order to understand how thi transformation is done, we must first know what fuzzy systems are. 1.2. What Are Fuzzy Systems? Fuzzy systems are knowledge-based or rule-based systems. ‘The heart of a fuzzy system isa knowledge base consisting of the so-called fuzzy IF-THEN rules. A fuzay IP-THEN rule is an IF-THEN statement in which some words are characterized by continuous membership functions. For example, the following is « fuzzy IF-THEN rule IF the speed of a car is high, THEN apply less force to the accelerator (1.1) where the words “high” and “less” are characterized by the membership functions shown in Figs.L.1 and 1.2, respectively.) A fuzzy system is constructed from a collection of fuzzy IF-THEN rules. Let us consider two examples. Example 1.1, Suppose we want to design a controller to automatically control the speed of a car. Conceptually, there are two approaches to designing such a controller: the first approach is to use conventional control theory, for example, designing a PID controller; the second approach is to emulate human drivers, that is, converting the rules used by human drivers into an automatic controller. We now ‘consider the second approach. Roughly speaking, human drivers use the following three types of rules to drive a car in normal situations IF speed is low, THEN apply more force to the accelerator (1.2) IP speed is medium, THEN apply normal force to the accelerator (1.3) IF speed ia high, THEN apply less force to the accelerator (1.4) ‘where the words “low.” “more,” “medium,” “normal,” high,” and “less” are char- acterized by membership functions similar to those in Figs..1-1.2. Of course, more rules are needed in real situations. We can construct a fuzzy system based on these TA detaled definition and anal of membership functions willbe given in Chapter 2. At this point, an intuitive understanding of the membersip functions in Figs. 1.1 and 1.2 swufient.See.1.2._ What Are Furry Systems? 3 ‘membership funetion for “high” speed (mp) 4555 6 Figure 1.1. Membersip funtion for high,” where the horizontal ax presents the speed ofthe ca ad the ver ical ais reprewnte the membership vale fr “igh” rules. Because the fuzzy system is used as a controller, it also is called a fuzzy controller. 3 Example 1.2. In Example 1.1, the rules are control instructions, that is, they represent what a human driver does in typical situations. Another type of human knowledge s descriptions about the system. Suppose a person pumping up a balloon wished to know how much air he could add before it burst, then the relationship among some key variables would be very useful. With the balloon there are three key variables: the air inside the balloon, the amount it increases, and the surface tension. We can describe the relationship among these variables in the following fuzzy IF-THEN rules: IF the amount of air is small and it is increased slightly, THEN the sur face tension will increase slightly as JP the amount of air is small andi is increased substantially, (4 6) THEN the surface tension will increase substantially : JP the amount of ai i large and itis increased slightly, (> THEN the surface tension will increase moderately » LF the amount of ai is large and it is increased substentially, (yg) THEN the sur face tension will increase very substantially where the words “small,” “slightly,” “substantially,” etc., are characterized by mem- bership functions similar to those in Figs.1.1 and 1.2. Combining these rules into @ fuzzy system, we obtain & model for the balloon. ©4 Ineroduction Ch. 1 membership function for “loss” \ foros vo acceleratoe igure 1.2. Mombership function for “los” where the hortontal asi representa the force applied to the accler- stor andthe vertical ais repreeats the membership value for “tes” In summary, the starting point of constructing a fuzzy system is to obtain a collection of fuzzy IF-THEN rules from human experts or based on domain know! edge. The next step is to combine these rules into a single system. Dillerent fuzzy systems use different principles for this combination. So the question is: what are the commonly used fuzzy systems? ‘There are three types of fuzzy systems that are commonly used in the literature: (i) pure fuzzy systems, (i) Takagi-Sugeno-Kang (TSK) fuzzy systems, and (it) fuzzy systems with fuzzifier and defuzzfier. We now briefly describe these three types of, fuzzy systems. ‘The basic configuration of a pure fuzzy system is shown in Fig. 1.3. The fuzsy rule base represents the collection of fuzzy IF-THEN rules. For examples, for the car controller in Example 1.1, the fuzzy rule base consists of the three rules (1.2)-(1-4), land for the balloon model of Example 1.2, the fuzzy rule base consists of the four rules (1.5)-(1.8). The fussy inference engine combines these fuzzy IF-THEN rules ‘nto a mapping from fuzzy sets? in the input space U CR" to fuzzy sets in the output space VC R based on fuzzy logic principles. If the dashed feedback line in Fig. 1.8 exists, the aystem becomes the so-called fuzzy dynamic system. ‘The main problem with the pure fuzzy system is that its inputs and outputs are i proke deliniion of funy wot it given in Chapter 2. At tis point i is suicient to view a foney ses asa word Uke, fr example, “high? which ls characterized by the membership function shown in FigSec. 1.2. What Ave Fury Syrtems? 5 Fuzzy Rule Base 1 ge! Fuzzy intorence - fuzzysets | Engine fuzzy sols nu. Inv Figure 1.3. Basic confgusnton of pave furzy eyen, fuzzy sets (that is, words in natural languages), whereas in engincering systems the ‘inputs and outputs are real-valued variables. To solve this problem, Tala’, Sugeno, ‘and Kang (Takagi and Sugeno [1985] and Sugeno and Kang [1988 proposed another fuzzy system whose inputs and outputs are real-valued vatiables, Instead of considering the fuxzy IF-THEN rules in the form of (1.1), the Takagi- Sugeno-Kang (TSK) system uses rules in the following forma: IF the speed x of a car is high, as) THEN the force to the accelerator is y = cr : where the word “high” has the same meaning as in (J.1), and ¢ is @ constant, Comparing (1.9) and (1.1) we see that the THEN part of the rule changes from a description using words in natural languages into a simple mathematical formula. This change makes it easier to combine the rules. In fact, the Takagi-Sugeno-Keng. fuazy system is a weighted average of the values in the THEN parts of the rules ‘The basic configuration of the Talagi-Sugeno-Kang fuzzy system is shown in Fig, 14, ‘The main problems with the ‘Takagi-Sugeno-Kang fuzzy system are: (i) its ‘THEN part is a mathematical formula and therefore may not provide a natural framework to represent human knowledge, and (i) there is not much freedom left to apply different principles in fuzzy logic, so that the versatility of fuzzy systems is not well-represented in this framework. ‘To solve these problems, we use the third ‘ype of fuzzy systems—fuzzy systems with fuzzifier and defuzzifer.6 Introsution Ch. 1 Fuzzy Rule Base Weighted Average xin yin Figure 1.4. Basic configuration of Takag-Suyeno Kang In order to use pure fuzzy systems in engineering systems, a simple method is to add a fuzzifier, which transforms a real-valued variable into a fuzzy set, to the input, and a defuzzifier, which transforms a fuzzy set into a real-valued variable, to the output. The result is the fuzzy system with fuzzifier and defuzaifier, shown in Fig. 15. This fuzzy system overcomes the disadvantages of the pure fuzzy systems and the Takngi-Sugeno-Kang fuzzy systems. Unless otherwise specified, from now on ‘when we refer fuzzy systems we mean fuzzy systems with fuzzifier and defuzztie. ‘To conclude this section, we would like to emphasize a distinguished feature of farzy systems: on one hand, fuzzy systems are multi-inpat-single-output mappings from a real-valued vector to a real-valued sealar (a multi-output mapping can be decomposed into a collection of single-output mappings), and the precise mathemat- ical formulas of these mappings can be obtained (soe Chapter 9 for details); on the other hand, fuzay systems are knowledge-based systems constructed from human ‘knowledge in the form of fuzzy IF-THEN rules, An important contribution of fuzzy systems theory is that it provides a systematic procedure for transforming a knowl- fe base into a nonlinear mapping. Because of this transformation, we are able to use knowledge-based systems (Fuzzy systems) in engineering applications (control, signal processing, or communications systems, etc.) in the Same manner as we use ‘mathematical models and sensory measurements. Consequently, the analysis and design of the resulting combined systems can be performed in a mathematically rigorous fashion. ‘The goal of this text is to show how this transformation is done, ‘aud how the analysis and design are performed.Sec_13._ Where Ave Fury Systims Used and How? Fuzzy Rule Base Defuzzifier yinv Fuzzy Inference Engine fuzzy sets 2 fuzzy sets inv inv Figure 1.5. Basic configuration of fusay systems with furier and defor 1.3 Where Are Fuzzy Systems Used and How? Fuzzy systems have been applied to a wide variety of fields ranging from control, signal processing, communications, integrated circuit manufacturing, and expert systems to business, medicine, psychology, etc. However, the most significant ap- plications have concentrated on control problems. Therefore, instead of listing the applications of fuzzy systems in the different fields, we concentrate on a number of control problems where fuzzy systems play a major role. Fuzzy systems, as shown in Fig. 1.5, can be used either as open-loop controllers or closed-loop controllers, as shown in Figs. 1.6 and 1.7, respectively. When tsed as ‘an open-loop controller, the fuzzy system usually sets up some control parameters and then the system operates according to these control parameters, Many applica- tions of fuzzy systems in consiumer electronics belong to this eategory. When used 18s a closed-loop controller, the fuzzy system measures the outputs of the process and takes control actions on the process continuously. Applications of fuzzy systems in industrial processes belong to this eategory. We now briefly describe how fuzzy systems are used in a number of consumer products and industrial systems,—-| Fuzzy L,| system Process. Figure 1.6, Puaty system as open-loop controler Process: = Fuzzy system |-+— Figure 1.7. Pasay system as closed loop controller 1.3.1 Fuzzy Washing Machines ‘The fuzzy washing machines were the first major consumer products to use fuzzy systems. They were produced by Matsushita Electric Industrial Company in Japan around 1990. ‘They use a fuzzy system to automatically set the proper cycle ac- cording to the kind and amount of dirt and the size of the load. More specifically, the fuzzy system used is a three-input-one-output system, where the three inputsSec. 13. Where Are Fury Systems Used and How? are measurements of drtines, type of dirt, and load size, and the outputs the correct cycle. Sensors supply the fuzzy system with the inputs. The optical sensor sends a beam of light through the water and measures how much of it reaches the other side. ‘The dirtier the water, the less light exosss. The optical sensor also can tell whether the dire is muddy or oily. Muddy dit dissolves faster. So, if the light readings reach minimum quickly, the dit is maddy. Ifthe downswing is slower, it And ifthe curve slopes somewhere in between, the dir i mixed, ‘The ma- Chine also has a load sensor that registers the volume of clothes. Cleary, the more volume of the clothes, the more washing time is needed. The heuristics above were summarized in a number of fuay TF-THEN rules that were then used to construct the fuzzy system 1.3.2 Digital Image Stabilizer Anyone who has ever used a camcorder realizes that it i very dificult for a human hhand to hold the cameorder without shaking slightly and imparting an irksome quiver to the tape. Smoothing out this jitfer would produce a new generation of camoorders and would have tremendous commercial value. Matsushita introduced ‘what it calls a digital image stabilizer, based on fuzzy systems, which stabilizes the picture when the hand is shaking. The digital image stabilizer is a fuzzy system that is constructed based on the following heuristics: LP all the points in the picture are moving inthe sme direction, (yo) THEN the hand is shaking I only some points in the picture are moving, THEN the hand ie not shaking ‘More specifically, the stabilizer compares each current frame with the previous images in memory. Ifthe whole appears to have shifted, then according to (1.10) the hhand is shaking and the fuzzy system adjusts the frame to compensate. Otherwise, it leaves it alone. Thus, if a car crosses the field, only a portion of the image will change, so the camcorder does not try to compensate. In this way the picture remains steady, although the hand is shaking. (aa) 1.3.3. Fuzzy Systems in Cars An automobile is a collection of many systems~—engine, transmission, brake, sus- pension, steering, and more—and fuzzy systems have been applied to alinost all of them. For example, Nissan has patented a fuzzy automatic transmission that saves fuel by 12 to 17 percent. Tt is based on the following observation. A normal transmission shifts whenever the car passes a certain speod, it therefore changes quite often and each shift consumes gas. However, human drivers not only shift less frequently, but also consider nonspeed factors. For example, if accelerating up10 Introduction Ch. 1 hill, they may delay the shift. Nissan's fuzzy automatic transmission device sum= marized these heuristies into a collection of fuzzy IF-THEN rules that were then used to construct a fuzzy system to guide the changes of gears. Nissan also developed a fuzzy antilock braking system. The challenge here is to apply the greatest amount of pressure to the brake without causing it to lock. The ‘Nissan system considers a number of heuristics, for example, IF the car slows down very rapidly, THEN the system assumes brake ~ lock and eases up on pressure (11?) In April 1992, Mitsubishi announced a fuzzy omnibus system that controls a car's automatic transmission, suspension, traction, four-wheel steering, four-wheel drive, and air conditioner. The fuzzy transmission downshifts on curves and also keeps the car from upshifting inappropriately on bends or when the driver releases the accelerator. The fuzzy suspension contains sensors in the front of the car that register vibration and height changes in the road and adjusts the suspension for a smoother ride, Fuzzy traction prevents excess speed on corners and improves the grip on slick roads by deciding whether they are level or sloped. Finally, fuzzy steering adjusts the response angle of the rear wheels according to road conditions ‘and the car's speed, and fuzzy air conditioning monitors sunlight, temperature, and bbumnidity to enhance the environment inside the car. 1.3.4 Fuzzy Control of a Cement Kiln Cement is manufactured by Snegeinding of cement clinker. ‘The clinkers ate pro- duced in the cement Kiln by heating a mixture of lineston, clay, and sand compo- nents. Because cement kilns exbibit time-varying nonlinear behavior and relatively few measurements are available, they are diffenlt to control using, conventional ‘control theory Tin the late 1970s, Holmblad and Ostergaard of Denmark developed a fuzzy system to control the cement kiln. ‘The fuzzy system (fuzzy controller) had four inputs and two outputs (which ean be viewed as two fuzzy systems in the form of Fig. 1.5, which share the same inputs). The four inputs are: (2) oxygen percentage in exhausted gases, (i) temperature of exhaust gases, (i) kiln drive torque, and (iv) litre weight of clinker (Indicating temperature level in the burning zone and quality of clinker). ‘The two outputs are: ()) coal fend rate and (i) air flow. A Collection of fuzay IF-THEN rules were constructed that describe how the outputs Should be related to the inputs. Foe example, the following two rules were used: LP the oxygen percentage is high and the temperature is low, 4 1) THEN increase air flow z IF the oxygen percentage is high and the temperature is high, THEN reduce the coal feed rate slightly Ce‘Sec_1.4. What Are the Major Research Fields in Fussy Theory? uw ‘The fuzzy controller was constructed by combining these rules into fuzay systems, In June 1978, the fuzzy controller ran for six days in the cement kiln of F.L. Smith & Company in Denmark—the first suecessful test of fuzzy control on a full-scale industrial process. The fuzzy controller showed a slight improvement over the results of the human operator and also cut fuel consumption. We will show more details about this system in Chapter 16. 1.3.5. Fuzzy Control of Subway Te ‘The most significant application of fuzzy systems to date may be the fuzzy control system for the Sendai subway in Japan. On a single north-south route of 13.6, kilometers and 16 stations, the train runs along very smoothly. ‘The fuzzy control system considers four performance criteria simutaneously: safety, riding comfort, traceability to target speed, and accuracy of stopping gap. The fuzzy control system consists of two parts: the constant speed controller (it starts the train and keeps the speed below the safety limit), and the automatic stopping controller (it regulates the tain speed in order to stop at the target position). The constant speed controller ‘was constructed from rules such as: For safety, IF the speed of train is approaching the lint speed, 15) THEN select the maximum brake notch For riding comfort; IF the speed is in the allowed range, THEN do not change the control notch More rules were used in the real system for traceability and other factors. ‘The ‘automatic stopping controller was constructed from the rules like (116) For riding comfort; TF the train will stop inthe allowed zone, THEN do not change the control notch aan For riding comfort and safety IF the train isin the allowed zone, (4 9) THEN change the control notch from acceleration to slight braking Again, more rules were used in the real system to take care of the accuracy of stopping gap and other factors. By 1901, the Sendai subway had carried passengers for four years and was still one of the most advanced subway systems. 1.4. What Are the Major Research id Fuzzy Theory? By fuzey theory we mean all the theories that use the basic concept of fuzzy set oF continuous membership fuuction. Fuzzy theory ean be roughly classified according to Fig.1.8. ‘There are five major branches: (i) fuzzy mathematics, where classi- cal mathematical concepts are extended by replacing classical sets with fuzzy sets; (ii) fuzzy logic and artificial intelligence, where approximations to classical logicFuzzy Theory Fuzzy Fuzzy | [Fuzzy Decision] [Uncertainty &| | Fuzzy Logic Mathematics} |systems| aking Information | | & Al Tay 5 TS cred. | | muttertra optmzatiod | [fuzzy loot princiles fuzzy meee) | | fuzzy mathomateal approxmate reasoning fuzzy analyse || | programming itzzy expen systoms fuzzy topology ; tuzy conics | wazysigna] [eormencatod posstity to oes froasures v t T lncentlnty controler desi ‘equalization Stabiyanalys| | pater recognition | channel Phage processing] assignment igure 1.8, Clnssifcation of furey theory. ‘are introduced and expert systems are developed based on fuzzy information and ‘approximate reasoning; (il) fuzzy systems, which include fuzzy control and fuzzy approaches in signal processing and communications; (iv) uncertainty and infor- mation, where different kinds of uncertainties are analyzed; and (v) fuzzy decision ‘making, which considers optimalization problems with soft constraints Of course, these five branches are not independent and there are strong inter- connections among them. For example, fuzzy control uses concepts from fuzzy ‘mathematies and fuzzy logic From a practical point of view, the majority of applications of fuzzy theory has concentrated on fuzzy systems, especially fuzzy control, as we could see from the examples in Section 1.3. There also are some fuzzy expert systems that performSee. 1.5. A Brief History of Fuzzy Theory and Applications 13 ‘medical diagnoses and decision support (Terano, Asai and Sugeno [1994]). Because fuzzy theory is still in its infancy from both theoretical and practical points of view, wwe expect that more solid practical applications will appear as the field matures, From Fig. 1.8 we see that fuzzy theory is a huge field that comprises a variety of research topics. In this text, we concentrate on fuzzy systems and fuzzy control We first will study the basic concepts in fuzzy mathematics and fuzzy logic that are useful in fuzzy systems and control (Chapters 2-6), then we will study fuzzy systems and control in great detail (Chapters 7-26), and finally we will briefly review some topics in other fields of fuzzy theory (Chapters 27-31), 1.5. A Brief History of Fuzzy Theory and Applications 1.5.1 The 1960s: The Beginning of Fuzzy Theory Fuzzy theory was initiated by Lotfi A. Zadeh in 1965 with his seminal paper “Fuzzy Sets" (Zadeh [1965]). Before working on fuzzy theory, Zadeh was a well-respected scholar in control theory. He developed the concept of “state,” which forms the basis for modern control theory. In the garly "60s, he thought that classical control theory had put too much emphasis on’precision and therefore could not handle the complex systems. As early as 1962, he wrote that to handle biological systems “we need a radically different kind of mathematics, the mathematics of fuzzy ot cloudy quantities which are not deseribable in terms of probability distributions” (Zadeh [1962)). Later, he formalized the ideas into the paper *Fuzay Sets.” Since its birth, fuzzy theory has been sparking Controversy. Some scholars, like Richard Bellman, endorsed the idea and began to work in this new field. Other scholars objected to the idea and viewed “fuszification” as against basic scientific principles. The biggest challenge, however, came from mathematicians in statistics and probability who claimed that probability is sufficient to characterize uncer tainty and any problems that fuzzy theory can solve can be solved equally well or better by probability theory (see Chapter 31). Because there were no real practical applications of fuzzy theory in the beginning, it was difficult to defend the field from a purely philosophical point of view. Almost all major research institutes in the world failed to view fuzzy theory as a serious research field. Although fuzzy theory did not fall into the mainstream, there were still many researchers around the world dedicating themselves to this new fel. In the late 1960s, many new fuzzy methods like fuzzy algorithms, fuzzy decision making, etc., were proposed. 1.5.2. The 1970s: Theory Continued to Grow and Real Applications Appeared It is fair to say that the establishment of fuzzy theory as an independent field is largely due to the dedication and outstanding work of Zadeh, Most of the funda-4 Inwroduction Ch. rental concepts in fuzzy theory were proposed by Zadeh in the late “60s and early 7s. After the introduetion of fuzzy sets in 1965, he proposed the concepts of fuzzy algorithms in 1968 (Zadeh (1968)), fuzzy decision making in 1970 (Bellman and Zadeh (1970)), and fuzzy ordering in 1971 (Zadeh [1971b]). In 1973, he published ‘another seminal paper, “Outline of a new approach to the analysis of complex sys- tems and decision processes” (Zadeh [1973]), which established the foundation for fuzzy control. In this paper, he introduced the concept of linguistic variables and proposed to use fuzzy IF-THEN rules to formulate human knowledge. A big event in the '70s was the birth of fuzzy controllers for real systems. In 1975, Mamdani and Assilian established the basic framework of fuzzy controller (hich is essentially the fuzzy system in Fig.1.5) and applied the fuzzy controller to control a steam engine. Their results were published in another seminal paper in fuzzy theory “An experiment in linguistic synthesis with a fuzzy logic controller” (Mamdani and Assilian (1975}). They found that the fuzzy controller was very easy to construct and worked remarkably well. Later in 1978, Holmblad and Ostergaard Aeveloped the first fuzzy controller for a full-scale industrial process—the fuzzy cement kiln controller (see Section 1.3). Generally speaking, the foundations of fuzzy theory were established in the 1970s. With the introduction of many new concepts, the picture of fuzzy theory as ‘a now field was becoming clear. Initial applications like the fuzzy steam engine con- troller and the fazzy cement kiln controller also showed that the field was promising, Usually, the field should be founded by major resources and major research inst- tutes should put some manpower on the topic. Unfortunately, his never happened. On the contrary, inthe late 70s and early 80s, many researchers in fuzzy theory had to change their field because they could not find support to continue their work. ‘This was especially true in the Unived States 1.5.3 The 1980s: Massive Applications Made a Difference In the easly °805, this feld, from a theoretical point of view, progressed very slowly. Few new concepts and approaches were proposed during this period, simply because very few people were still working in the field. It was the application of fuzzy control that saved the field ‘Japanese engineers, with their sensitivity to new technology, quickly found that fuzzy controllers were very easy’ to design and worked very well for many problems. ‘Because fuzzy control does not require a mathematical model of the process, it could be applied to many systems where conventional control theory could not be used due toa lack of mathematical models. In 1980, Sugeno began to create Japan's first fuzzy application--control of Fuji Hleetric water purification plait. In 1983, he began the pioneer work on a fuzzy robot, a seléparking car that was controlied by calling out commands (Sugeno and Nishida [1985)). In the early 1980s, Yasunobu ‘and Miyamoto from Hitachi began to develop a fuzzy control system for the Sandai‘Sec. 1.6, Summary nd Further Readings 15 subway. They finished the project in 1987 and created the most advanced subway system on arth, This very impressive aplication of fuzzy cots made a very big difference, yeu In July 1987, the Second Annual International Fuzzy Systems Association Con- ference was held in Tokyo. The conference began three days after the Sendai subway began operation, and attendees were amused with its dreamy ride. Also, in the con- ference Hirota displayed a fuzzy robot arm that played two-dimensional Ping-Pong in real time (Hirota, Arai and Hachisn [1989)), and Yamakawa demonstrated a fuzzy system that balanced an inverted pendiluth’ (Yamakawa [1989]). Prior to this event, fuzzy theory was not well-known in Japan. After it, a wave of pro-fuzzy sentiment swept through the engincering, government, and business communities. By the early 1990s, a large number of fuzzy consumer products appeared in the market (see Section 1.3 for examples). 1.5.4 The 1990s: More Challenges Remain ‘The success of fuzzy systems in Japan surprised the mainstream researchers in the United States and in Europe. Some still criticize fuzzy theory, but many others have been changing their minds and giving fuzzy theory a chance to be taken seriously. In February 1992, the first IEEE International Conference on Fuzzy Systems was hheld in San Diego. ‘This event symbolized the acceptance of fuzzy theory by the largest engineering organization—-IEEE. In 1993, the IEEE Transactions on Fuzzy’ Systems was inaugurated. From a theoretical point of view, fuzzy systems and control has advanced rapidly in the Inte 1980s and early 1990s. Although it is hard to say there is any break- through, solid progress has been made on some fundamental problems in fuzzy systems and control. For examples, neural network techniques have been used to determine membership functions in a systematic manner, and rigorais stability analysis of fuzzy control systems has appeared. Although the whole picture of fuzzy systems and control theory is becoming clearer, much work remains to be done. Most approaches and analyses are preliminary in nature. We believe that only when the top research institutes begin to put some serious man power on the research of fuzzy theory can the field make major progress. 1.6 Summary and Further Readings In this chapter we have demonstrated the following: ‘© The goal of using fuzzy systems is to put human knowledge into engineering systems in a systematic, efficient, and analyzable order. ‘© The basic architectures of the commonly used fuzey systems,16 Introduction Ch. ‘© The fuzzy IF-THEN rules used in certain industrial processes and consumer products. ‘© Classification and brief history of fuzzy theory and applications, AA very good non-technical introduction to fuzzy theory and applications is Me- Neill and Fretberger (1993). It contains many interviews and describes the major events. Some historical remarks were made in Kruse, Gebhardt, and Klawonn (1994). Klir and Yuan [1995] is perhaps the mast. comprehensive book on fuzzy sets nd fuzzy logic. Earlier applications of fuzzy control were collected in Sugeno (1985) and more recent applications (mainly in Japan) were summarized in Terano, Asai, and Sugeno [194] 1.7 Exercises Exercise 1.1. Is the fuzzy washing machine an open-loop control system or ‘closed-loop control system? What about the fuzzy cement kiln control system? Explain your answer. Exercise 1.2, List four to six applications of fuzzy theory to practical problems father than those in Section 1.3. Point out the referenoes where you find these applications.Sec. 1.7. Gwecies 7 Exereise 1.3. Suppose we want to design a fuzzy system to balance the inverted pendulum shown in Fig. 1.9. Let the angle @ and its derivation 6 be the inputs to the fuzzy system and the force u applied to the cart be its output. (@) Determine three to five fuzzy IF-THEN rules based on the common sense of how to balance the inverted pendulum. (b) Suppose that the rules in (a) can successfully control a particular inverted pendulum system, Now if we want to use the rules to control another inverted pendulum system with different values of me,m, and 1, what parts of the rules should change and what parts may remain the same. Figure 1.9. The inverted pendulum system.18 Inerodution cnPart | The Mathematics of Fuzzy Systems and Control Fuzzy mathematics provide the starting point and basic language for fuzzy sys- ‘tems and fuzzy control. Fuzzy mathematics by itself is a huge field, where fuzzy mathematical principles are developed by replacing the sets in classical mathemati- cal theory with fuzzy sets. In this way, all the classical mathematical branches may be “fuzzified.” We have seen the birth of fuzzy measure theory, fuzzy topology, fuzzy algebra, fuzzy analysis, etc. Understandably, only a small portion of fuzzy mathematics has found applications in engineering. In the next five chapters, we will study those concepts and principles in fuzzy mathematics that are useful in fuzzy systems and fuzzy control. In Chapter 2, we will introduce the most fundamental concept in fuzzy theory —the concept of fuzzy set. In Chapter 3, set-theoretical operations on fuzzy sets such as complement, union, and intersection will be studied in detail. Chapter 4 will study fuzzy relations and introduce an important principle in fuzzy theory— the extension principle. Linguistic variables and fuzzy IF-THEN rules, which are ‘essential to fuzzy systems and fuzzy contzol, will be precisely defined and studied in Chapter 5. Finally, Chapter 6 will focus on three basie principles in fuzzy logie that are useful in the fuzzy inference engine of fuzzy systems. 19Chapter 2 Fuzzy Sets and Basic Operations on Fuzzy Sets 2.1 From Classical Sets to Fuzzy Sets Let Ube the universe of discourse, or universal set, which contains all the possible clements of concern in each particular context or application. Recall that a classical (crisp) set A, or simply a set A, in the universe of discourse U can be defined by listing all ofits members (the list method) or by specifying the properties that must bbe satisfied by the members of the set (the rule method). The list method can be used only for finite sets and is therefore of limited use. ‘The rule method is more general. In the rule method, a set A is represented as A= {2 €Ulz meets some conditions} ay ‘There is yet a third method to define a set A—the membership method, which Introduces a zero-one membership function (also called characteristic function, dis- crimination function, or indicator function) for A, denoted by y14(2), such that Lh @2) ‘The set A is mathematically equivalent to its membership function a(x) in the sense that knowing j14(2) is the same as knowing A itself, Example 2.1. Consider the set of all cars in Berkeley; this is the universe of discourse U. We can define different sets in U according to the properties of cars Fig. 2.1 shows two types of properties that can be used to define sets in U: (a) US ‘ears or non-US cars, and (b) number of cylinders. For example, we can define a set Aas all cars in U that have 4 eylinders, that is, A= (26 Ulz has 4 cylinders} a) 20Sec. 2.1. From Classical Sets to Funny Sets 21 4 Cylinder US cars, 6 Cylinder 8 Cylinder Non-US car Others Figure 2.1. Partitioning ofthe st ofall ears in Berkley ino aabset by: (a) US ca or non-US cas, and (0) number of eindrs. 1 if, 2€U and has 4 cylinders mat={ 6 tf Seu ands dana ne tatinders 20 If we want to define a set in U according to whether the car is a US ear or a non-US. car, we face a difficulty. One perspective is that a car is a US ear if it carries the name of a USA auto manufacturer; otherwise it is a non-US ear. However, many people feel that the distinction between a US ear and a non-US ear is not as crisp as it once was, because many of the components for what wo consider to be US ears (for examples, Fords, GM's, Chryslers) are produced outside of the United States Additionally, some “non-US” cars are manufactured in the USA. How to deal with, this kind of problems? © Essentially, the difficulty in Example 2.1 shows that some sets do not have clear boundaries. Classical set theory requires that a set must have a well-defined property, therefore it is unable to define the set like “all US cars in Berkeley.” ‘To overcome this limitation of classical set theory, the concept of fuzay set was introduced. Tt turns out that this limitation is fundamental and a new theory is needed—this is the fuzzy set theory. Definition 2.1. A fuzzy set in a universe of discourse U is characterized by a ‘membership function jia(z) that takes values in the interval [0 1) ‘Therefore, a fuzzy set is « generalization of a classical set by allowing the mem-22 Fussy Set and Bate Operations on Furry Sets_Ch.2 bership function to take any values in the interval [0,1]. In other words, the mem- bership function of a classical set can only take two values—zero and one, whereas ‘the membership function of a fuzzy set is a continuous function with range (0,1) We see from the definition that there is nothing “fuzzy” about a fuzzy set; it is simply a set with a continuous membership function. A fuzzy set A in U may be roprosonted as a set of ordered pairs of a generic clement 2 and its membership value, that is, A= {(2,nalz))]e €U} (2.5) ‘When U is continuous (for example, A= [mnie 26) A), Als commonly written as ‘where the integral sign does not denote integration; it denotes the collection of all, points x € U with the associated membership function y14(2). When U is discrete, ‘Ais commonly written as A=Duale/e 27) where the summation sign does not represent arithmetic addition; it denotes the collection ofall points = € U with the asociated membership function (2) ‘We now return to Example 2.1 and see how to use the concept of fuzzy st to define US and non-US cars Example 2.1 (Cont'd). We can define the set “US cars in Berkeley,” denoted by D, asa fuzzy set according tothe percentage of the car's parts made in the USA. Specialy, D is defined by the membership function Bo(z) = plz) (2.8) where p(2) isthe percentage ofthe parts of car x made in the USA and it takes ‘alues from 0% to 100%. For example, if particular cat zp has 60% ofits pars Inde in the USA, then we say thatthe car‘z9 belongs to the fury set D to the degree of 0.6 Similarly, we can define the set: “non-US cars in Berkeley,” denoted by F, as a fuzzy set with the membership function P(z) (2.9) ‘where p(z) is the same as in (2.8). Thus, if a particular car zy has 60% of its parts ‘made in the USA, then we say the car 2» belongs to the fuzzy set F to the degree of 1-0.6=0.4. Fig. 2.2 shows (2.8) and (2.9). Clearly, an element can belong to different fuzzy sets to the same or diferent degrees. (1 ‘We now consider another example of fuzzy sets and from it draw some remarks. wrt° 100% — p(x) Figure 2.2. Membership functions for US (yp) and non- US up) cars ased on the percentage of parts of the car rade in the USA (p(2)). Example 2.2. Let Z be a fuzzy set named “numbers close to zero.” Then a possible membership function for Z is hala) =e (2.10) where ¢ € R. This is a Gaussian function with mean equal to 2ero and standard derivation equal to one. According to this membership function, the numbers 0 and 2 belong to the fuzzy set Z to the degroes of e? = 1 and e~*, respectively. ‘We also may define the membership function for Z as 0 if rca rel if ~1s2<0 lnz if Ose<1 0 if ise According to this membership function, the numbers 0 and 2 belong to the fuzzy set Z to the degrees of 1 and 0, respectively. (2.10) and (2.11) are plotted graphically in Figs. 2.3 and 2.4, respectively. We can choose many other membership functions to characterize “numbers close to zero.” 0 wala) (2a) From Example 2.2 we can draw three important remarks on fuzzy sets: ‘© The properties that a fuzzy set is used to characterize are usually fuzzy, for ‘example, “numbers close to zero” is not a precise description. Therefore, weuna Sets and Basic Operations on Fuzzy ets_Ch. 2 Figure 2.3. \ possible membership function to character ine "numbers cae to ero may use different membership functions to characterize the same description. However, the membership functions themselves are not fuzzy—they are precise mathematical functions. Once a fuzzy property is represented by a member- ship function, for example, once “numbers close to zero" is represented by the membership function (2.10) or (2.11), nothing will be fuzzy anymore. Thus, by characterizing a fuzzy description with 2 membership function, we essen- tially defuzzify the fuzzy description. A common misunderstanding of fuzzy set theory is that fuzzy set theory tries to fuzzify the world. We see, on the contrary, that fuzzy sets are used to defuzzify the world Following the previous remark is an important question: how to determine the ‘membership functions? Because there are a variety of choices of membership functions, how to choose one from these alternatives? Conceptually, there are ‘two approaches to determining a membership function. ‘The frst approach is to use the knowledge of human experts, that is, ask the domain experts to specify the membership functions. Because fuzzy sets are often used to formulate human knowledge, the membership functions represent a part of ‘human knowledge. Usually, this approach can only give a rough formula of the ‘membership function; fine-tuning is required. In the second approach, we use data collected from various sensors to determine the membership functions. Specifically, we first specify the structures of the membership functions and ‘then fine-tune the parameters of the membership functions based on the data, Both approaches, especially the second approach, will be studied in detail in‘Se. 21,_From Classical Sets to Fuzzy Ses 25 Figure 2.4. Another posible membership function to characterise “numbers lowe 40 er.” later chapters. ‘« Finally, it should be emphasized that although (2.10) and (2.11) are used to characterize the same description “nurabers close to zero.” they are different fuzzy sets. Hence, rigorously speaking, we should use different labels to rep- resent the fuzzy sets (2.10) and (2.11); for example, we should use j1z, (2) in (2.10) and j1z,(z) in (2.11). A fuzzy set has a one-to-one correspondence with its membership function. That is, when we say a fuzzy set, there must be a unique membership funetion associated with it; conversely, when we give a membership function, it represents a fuzzy set. Fuzzy sets and their membership functions are equivalent in this sense. Let us consider two more examples of fuzzy sets, one in continuous domain and ‘the other in discrete domain; they are classical examples from Zadeh’s seminal paper (Zadeh (1965) Example 2.8. Let U be the interval [0,100] representing the age of ordinary Jumans. Then we may define fuzzy sets “young” and “old” as (using the integral notation (2.6)) (212) (213)26 Fuzzy Sets and Basic Operations on Fuzzy Sets_Ch.2 1 ung, old age Figure 2.5. Disgrammatic representation of “young” and Sola” See Fig. 2.5. 0 Example 2.4. Let U be the integers from 1 to 10, that is, U = {1,2,..,10), ‘Then the fuzzy set “several” may be defined as (using the summation notation an) several = 0.5/3+ 08/4 + 1/5+1/6+08/7 +0.5/8 (2.14) ‘That is, 5 and 6 belong to the fuzzy set “several” with degree 1, and 7 with degree 0.8, 3 and 8 with degree 0.5, and 1,2,9 and 10 with degree 0. See Fig, 2.6. 0 2.2 Basic Concepts Associated with Fuzzy Set ‘We now introduce some basic concepts and terminology associated with a fuzzy set Many of them are extensions of the basic concepts of a classical (crisp) set, but some are unique to the fuzzy set framework. Definition 2.2. The concepts of support, fuzzy singleton, center, crossover point, height, normal fuzzy set, a-cut, convex fizay set, and projections are defined as follows, ‘The support of a fuzzy set A in the universe of discourse U is a crisp set that contains all the elements of Uthat have nonzero membership values in A, that is, supp(A) = {2 € Uluale) > 0} (218)Sec. 2.2. Basie Concepts Associated with Furry Set a Hseverat®) og integer x 12345678910 Figure 2.6. Memborshlp function for fuzy st “several” where supp(A) denotes the support of fuzzy set A. For example, the support of fuzzy set “several” in Fig. 2.6 is the set of integers {3,4,5,6,7,8). If the support of a fuzzy set is empty, itis called an empty fussy set. A fuzzy singleton is a furzy set whose support is a single point in U. ‘The center of a furzy set is defined as follows: if the mean value of all points at which the membership function of the fuzzy set achieves its maximum value is finite, then define this mean value as the center of the fuzzy set; if the mean value equals positive (negative) infinite, then the center is defined as the smallest (largest) ‘among all points that achieve the maximum membership value. Fig. 2.7 shows the centers of some typical fuzzy sets. The crossover point of a fuzzy set isthe point in U whose membership value in A equals 0.5. The height ofa furzy setis the largest membership value attained by any point. For example, the heights of all the fuzzy sets in Figs 2.2-24 equal one. If the height of a fuzzy set equals one, itis called a normal fuzzy set. All the fuzzy sets in Figs. 2.2-24 are therefore normal fuzzy sets. ‘An a-cut of a fuzzy set A is a crisp set Aq that contains all the elements in U that have membership values in A greater than or equal to a, that is, Ag = {2 € Ulpalz) 20) (2.16) For example, for a = 0.3, the a-cut of the fuzey set (2.11) (Fig. 2.4) is the crisp set [-0.7,0.7}, and for a = 0.9, itis (-0.1,0.1) ‘When the universe of discourse (is the n-dimensional Euclidean space FR", the28 Fuzzy Sets and Basic Operations on Furry Sets_Ch.2 A Ay IA u center center center center ofA of Ay of Ag of Ag Figure 2.7. Centers of some typical fury sets ‘concept of set convexity can be generalized to fuzzy set. A fuzzy set is said to be conver if and only ifits a-cut Aa isa convex set for any a inthe interval (0,1) ‘The following lemma gives an equivalent definition of a convex fuzzy set. Lemma 2.1. A furzy set Ain R” is convex if and only if abday + (1 22a] 2 min{ua(er),ta(aa)] a7) for all 21,2 € RM and all € (0,1) Proof: First, suppose that A is convex and we prove the truth of (2.17). Let ‘and 2 be arbitrary points in R” and without loss of generality we assume ja(t) < wa(z2). Hf x4(z1) =O, then (2.17) is trivially true, so we let ra(z1) = a > 0. Since by assumption the a-cut Ay is convex and 21,22 € Ay (since a(ta) > ha(ei) a), we have Aty + (1 —A)rz € Aq for all \ € (0, 1]. Hence, eazy + (1 — )za] > = pala) = minlj al) aoa) Conversely, suppose (2.17) is true and we prove that A is convex. Let a be ‘an arbitrary point in (0,1). If Ag is empty, then it is convex (empty sets are convex by definition). If Ag is nonempty, then there exists 2 € R* such that paler) = (by the defisition of Ay). Let 2 be an arbitrary clement in Ag, then dualea) > a= jea(2y). Since (2.17) is true by assumption, we have yaPXzi + (1 — A)ra] > minia(z1),4a(z2)] = wa(z1) = a for all 4 € [0,1], which means that dar + (1 A)za € Aq. So Ay is a convex set. Since ais an arbitrary point in (0, 1], the convexity of Ay implies the convexity of A.See.23. Operations on Fury Sets 29 Lot A be a fuany set in R* with membership function pa(z) = jo4(ty, 025) and H be a hyperplane in RY defined by H = {z € R°\z1 = 0) (for notational implicity, we consider this special case of hyperplane; generalization to general hyperplanes is straightforward). The projection of A on His a furry set Avr in RP defined by Hag Eason) = SUP HalBts Fn) (218) sehere sup, cp HCP, ~n) denotes the maximum value ofthe function 4(21, 52) when 2 takes values in R. 2.3 Operations on Fuzzy Sets ‘The basic concepts introduced in Sections 2.1 and 2.2 concern only a single fuzzy set. In this section, we study the basic operations on fuzzy sets. In the sequel, we assume that A and B are fuzzy sets defined in the same universe of discourse U. Definition 2.8. The equality, containment, complement, union, and intersee- tion of two fuzzy sets A and B are defined as follows. We say A and B are equal if and only if .4(2) = wp(e) for all x € U. We say B contains A, denoted by AC B, if and only if 1a(2) < pp(2) for all x € U. The ‘complement of A is a fuzzy set Ain J whose membership function is defined as wala) (2.19) ‘The union of A and B is a fuzzy set in U, denoted by AU B, whose membership function is defined as wala) Haun (2) = maz[pa(z), Ha (2)) (2.20) ‘The intersection of A and B is a fuzzy set AM B in U with membership function pana (2) = min{ya(2),40(2)) (2.21) ‘The reader may wonder why we use “max” for union and “min” for intersection; ‘we now give an intuitive explanation. An intuitively appealing way of defining the ‘union is the following: the union of A and B is the smallest fuzzy set containing both A and B. More precisely, if C is any fuzzy set that contains both A and B, then it ‘also contains the union of A and B. ‘To show that this intuitively appealing definition is equivalent to (2.20), we note, frst, that AUB as defined by (2.20) contains both A and B because maz{iia, yi] > Ha and mazijea,sp] > jis. Furthermore, if C is ‘any fuzzy set containing both A and B, then uc > pia and fic > pip. Therefore, hc & maz|ia, 5] = wavs, which means that AU B as defined by (2.20) is the smallest fuzzy set containing both A and B. The intersection as defined by (2.21) can be justified in the same manner.Fuzzy Ses and Basic Operations on Furzy Sets_Ch.2 ° 100% poo igure 2.8. The membership functions for P and F: ‘Example 2.5. Consider the two fuzzy sots D and F defined by (2.8) and (2.9) (cee also Fig. 2.2). The complement of F, Fis the fuzzy set defined by Hp (@) = ~ v2) 222) ‘which is shown in Fig. 2.8. Comparing (2.22) with (2.9) we see that # = D. This rakes sense because if a car is not a non-US car (which is what the complement of F means intuitively), then it should be a US ear; or more accurately, the less a car is a non-US car, the more the car i a US ca. The union of F and D is the fuzzy set FUD defined by ~ ela) = ur(z) if OS plz) <05 {id ip seme 229 which is plotted in Fig. 29, ‘The intersection of F and D is the furay set FD dined by prun(2) = mazipr, pp} { p2e if O np and ja < pp. If ig > pay theo 1jig < 1g and 1~maeljea, pop] = bps = in| pa, 1—pp) whichis (2.27). Iyea < yp, then 1— 4 > 1p and 1 maria, pox] = 1— pn rmin(L—p14,1 nol, which is again (2.27). Henee, (2.27) is true. From the definitions (2.19)-(2.21) and the definition of the equality of two fuzzy sets, we see that (2.27) implies (2.25). © 2.4 Summary and Further Readings In this chapter we have demonstrated the following: *# The definitions of fuzzy set, basic concepts associated with a fuzzy set (sup- port, a-cut, convexity, ete.) and basic operations (complement, union, inter- section, etc.) of fuzny sets ‘¢ The intuitive meaning of membership functions and how to determine intu- itively appealing membership functions for specific fuzzy descriptions. ‘+ Performing operations on specific examples of fuzzy sets and proving basic properties concerning fuzzy sets and their operations. adel’ original paper (Zadeh [1965}) is still the best source to learn fuzzy set and related concepts. The paper was extremely well-written and the reader is ‘encouraged to read it. The basic operations and concepts associated with a fuzzy sot were also introduced in Zadeh (1965) 2.5 Exercises Exercise 2.1. Determine reasonable membership functions for “short persons,” cnet \ ee rudents, (b) top students, and (c) smart students, Bxercise 2.3. Consider the fuzzy sets F,G and H defined in the interval U (0,10) by the membership functions 2 1 ze HO) = 2% ale) = Te aE (228) pe(@) Determine the mathematical formulas and graphs of membership functions of each of the following fuzzy” sets @ FGA 0) FUG,FUH,GUHSec.25. Exercises 33 (© PNG.FaHGAH @ FuGUHFOGOR (FOR, GHHFOR 3 for: (a) a= 0.2, (b) a =05, (c) a =09, and (d) a= 1 Exercise 2.5. Let fuzzy set A be defined in the closed plane U = [—1,1]{~8,3] ‘with membership function Ng Exercise 2.4. Determine the a-cuts of the fuzny sets F,G and HT in Exercise 2. (eto) (2.29) alr, 2) = Determine the projections of A on the hyperplanes Hy = {z € Ulzi = 0} and Hy = {x € Ulza = 0}, respectively. \g Exercise 2.6. Show that the law of the exciuded middle, F UF = U, is not ue if F is a fuzzy set, Exereise 2.7. Prove the identity (2.26) in Lemma 2.2. Exercise 2.8, Show that the intersection of two convex fuzzy sets is also a convex fuzzy set. What about the union?Chapter 3 Further Operations on Fuzzy Sets In Chapter 2 we introduced the following basic operators for complement, union, ‘and intersection of fuzzy sets: ha(z) = 1— pale) (3.1) Have (2) = maz|pa(z), o(2)] (32) sana (2) = miniua(),40(2)] 3) We explained that the fuzzy set AU B defined by (3.2) is the smallest fuzzy set containing both A and B, and the fuzzy set ANB defined by (3.3) is the largest fuzzy set contained by both A and B. Therefore, (3.1)-(3.3) define only one type of operations on fuzzy sets, Other possibilities exist, For example, we may define AUB as any fuzzy set containing both A and B (not necessarily the smallest fuzzy set). In this chapter, we study other types of operators for complement, union, and intersection of fuzzy sets Why do we need other types of operators? The main reason is that the operators (3.1)-(3.8) may not be satisfactory in some situations. For example, when we take the intersection of two fuzzy sets, we may want the larger fuzzy set to have an Impact on the result. But if we use the rin operator of (3.3), the larger fuzzy set will have no impact. Another reason is that from a theoretical point of view itis interesting to explore what types of operators are possible for fuzzy sets. We know ‘that for nonfuzzy sets only one type of operation is possible for complement, union, fr intersection. For fuzzy sets there are other possibilities. But. what are they? What are the properties of these new operators? These are the questions we will ‘ny to answer in this chapter. ‘The new operators will be proposed on axiomatic bases. That is, we will start with a few axioms that complement, union, or intersection should satisfy in order +o be qualified as these operations. Then, we will list some particular formulas that satisfy these axioms. 34‘Sec. 3:._ Fuzzy Complement 35 3.1 Fuzzy Complement Let e : [0,1] > [0,1] be a mapping that transforms the membership function of fuzzy set A into the membership function of the complement of A, that is, elua(z)] = wale) (4) In the ease of (8.1), 4a(z)] = 1—sa(2). In order forthe function cto be qualified as a complement, it should satisfy at least the following two requirements: Axiom e1, o(0) = 1 and e(1) = 0 (boundary condition, Acxiom 2. For all a,b € [0,1], if a e(0) (noninereasing con- dition), where (and throughout this chapter) a and b denote membership functions ‘of some fuzay sets, say, a = pa(z) and b= wp(z). Axiom el shows that if an element belongs to a fuzay set to degree zero (one), then it should belong to the complement of this fuzzy set to degree one (zero) ‘Axiom c2 requires that an increase in membership value must result ina decrease or no change in membership value for the complement. Clearly, any violation of these ‘wo requirements will result in an operator that is unacceptable as complement. Definition 3.1. Any function ¢: [0,1] + [0,1] that satisies Axioms cl and c2 is called a fuzzy complement ‘One class of fuzzy complements is the Sugeno class (Sugeno (1977) defined by ea THe where \ € (~1,00). For each value of the parameter A, we obtain a particular fuzzy ‘complement. It is a simple matter to check that the complement defined by (3.5) satisfies Axioms cl and c2. Fig. 3.1 illustrates this class of fuzzy complements for dlifferont values of A. Note that when A= 0 it becomes the basic fuzzy complement (a). Another type of fuzzy complement is the Yager class (Vager [1980]) defined by co(a) = (1-a¥)'/* (36) ‘where w € (0,00). For each value of w, we obtain a particular fuzzy complement. It is easy to verify that (8.6) satisfies Axioms cl and 2. Fig. 32 illustrates the Yager class of fuzzy complements for different values of w. When w = 1, (3.6) becomes (32), ex@) = (a5) 3.2. Fuzzy Union—The S-Norms Let 8: [0,1] x(0, 1] > [0,1] be a mapping that transforms the membership functions of fuzzy sets A and B into the membership function of the union of A and B, that36 Further Operations on Fuzzy Sts_Ch. 3 Figure 3.2. Yager class of fuzzy complements cu(a) for diferent values of w s|Ha(z),Ho(2)] = pave (2) 7) In the case of (8.2), sa(2), a ()] = maz{y.4(2),45(2)]- In order for the function ‘¢ to be qualified at an union, it must satisfied at least the following four require ‘ments:37. ,8(0;a) = e(a,0) = a (boundary condition). Axiom s2. s(a,8) = (ba) (commutative condition). Axiom 88, Ifa #(0,0) =a (a7)See.32._Fueey The Figure 8.3. Membership function of DUF using the Yager norm (810) with w= 3, Figure 3.4. Membership funtion of DU F ws setae sum (3.3). Furthermore, the commutative condition Axiom s2 gives (a,b) = #(b,a) > 9(0,0) =b Combining (8.17) and (3.18) we have a(a,) > maz(a,). ‘Next we prove s(a,b) < s4s(a,6). IF6=0, then from Axiom s1 we have a(a,) = 39 (3.18)40 Further Operations on Fuzzy Sets_Ch.3 ‘s(a,0) = a, thus 6(a,8) = e4s(a,8). By the commutative condition Axiom s2 we have s(a,) = sas(a,6) ifa=0. Ifa #0 and 60, we have sas(a,b) =12 s(a,0) (3.19) ‘Thus 9(0,8) < s4s(a,t) for all a,b € [0,1]. 2 Finally, we prove an interesting property of the Dombi s-norm s,(a,b) (3.8): x(a,b) converges to the basic fuzzy union maz(a,2) as the parameter goes to infinity and converges to the drastic sum (a,b) as A goes to zero. Therefore, the Dombi s-norm covers the whole spectrum of s-norms. Lemma 3.1: Let #4(a,8) be dofined as in (3.8) and sye(a,b) be defined as in (3.11), then fim, s,(a,2) = maz(a,b) (3.20) Him £4(a,8) = ses(a,0) (21) Proof: We irst prove (3.20). Ifa = b # 0, then from (3.8) we have lim sae 83(0,6) = Him). oof1/(142-/A(2=1))] wa2(a,8). Ia = b= 0, then lims+50 6x(0,6) = Himsa 1/(1 + 0-4) ax(a,b). Tf a # b, then without loss of generality (due to Axiom s2) we assume a —o0 and a> 20, respectively.Chapter 4 Fuzzy Relations and the Extension Principle 4.1. From Classical Relations to Fuzzy Relations 4.1.1 Relations Let U and V be two arbitrary classical (nonfuzzy, crisp) sets. The Cartesian product of U' and V, denoted by U x V, is the nonfuzzy set of all ordered pairs (u,) such that we U and v € V; that is, UxV={(uv)ueU andveV} (4) Note that the order in which U and V appears is important; that is, if U # V, then UxV #VxU. In general, the Cartesian product of arbitrary n nonfuzzy sets Ui,Ury Ung denoted by Ui x Uz x ++ x Up, Is the nonfuzzy set of all n-tuples (u;,tay mtn) such that uy € Uy for # € {1,2,..,}5 that is, Uy x Ua 2° Un = {Ct ta, stent € Urs ti € Ua52-y tn Un) (4.2) A (nonfuzzy) relation among (nonfuzey) sets Us,Ua,..Un is a subset of the Cartesian product Uy x U: x+--% Uy that is, if we use Q(U,,.-,Un) to denote a relation among Us, Ua5,Up, then QW Wass Un) Cr x Up x0 x Ty (43) Asa special ease, a binary relation between (nonfuzzy) sets U and V is a subset of the Cartesian product U > V. Example 4.1. Let U = {1,2,3) and V = {2,3,4). Thon the cartesian product of ana Vis the set UV = {(1,2), (1,3), (14), (2,2); (2,3) (24), (8,2) (3,3). (8,4)} ‘A relation botween U and V is subset of U x V. For example, let Q(U; V) be @ relation named “the fst element is no smaller than the second element,” then QUW,¥) = {(2,2),(8,2).(8,3)) (4a) 4849 ‘Se. 4. From Classical Relations to Fussy 0 Because a relation is itself a set, all of the basic set operations can be applied to it without modifcation. Also, we can use the following membership function to represent a relation: { Lif (Urtay ttn) EQ Vay Un) (4.5) tay ty ee For binary relation Q(U, V) defined over Ux V which contains finite elements, we often collect the values of the membership function g1g into a relational matrix; see the following example. Example 4.1 (Cont'd). The relation Q(U, V) of (4.4) can be represented by the following relational matrix: v 234 1000 (46) v2i08)d0 Bi a ‘A classical relation represents a crisp relationship among sets, that is either there is such a relationship or not. For certain relationships, however, itis dificult to give a zero-one assessment; see the following example. Example 4.2. Let U = {SanFrancisco, HongKong, Tokyo} and V = {Boston, HongKong}. We want to define the relational concept “very far" between these two sets of cities. Cleary, classical relations are not useful because the concept “very far" is not well-defined in the framework of classical sets and relations. How- ever, “very fas" does mean something and we should find a numerical system to characterize it. If we use a number in the interval [0,1] to represent the degree of “very far” then the concept “very far” may be represented by the following (fuzzy) relational matrix: v Boston HK SF 03 09 (a7) voHK 1 oO Tokyo 095 01 0 Example 42 shows that we need to generalize the concept of classical relation in order to formulate more relationships in the real world. The concept of fuzzy ‘elation was thus introduced, Definition 4.1. A fuzzy relation isa fuzzy set defined in the Cartesian product of crisp sets Uj, Up,..,Up. With the representation scheme (2.5), a fuzzy relation50 Qin Uy x Up x +++ Up is dofined as the fuzzy set Q= {uss where yg : Us x Ug x +++ Uy 0,1) As a special case, a binary fuzzy relation isa fuzzy set defined in the Cartesian product of two crisp sets. A binary relation on a finite Cartesian product is usually represented by a fuzsy relational matrit, that is, a matrix whose elements are the ‘membership values of the corresponding pairs belonging to the fuzzy relation. For ‘example, (4.7) is a fuzzy relational matrix representing the fuzzy relation named “very far” between the two groups of cities. Example 4.8. Let U and V be the set of real numbers, that is, U=V = R.A fuzay relation “r is approximately equal to y," denoted by AE, may be defined by ‘the membership function un) (ass tn) (nay tn) € Ur Xa Xo Un) (48) Bag(ayy) =e us) Similarly, a fuzzy relation “x is much larger than y,” denoted by ML, may be defined by the membership function (4.10) Of course, other membership functions may be used to represent these fuzzy rela- tions. 4.1.2. Projections and Cylindric Extensions Because a crisp relation is defined in the product space of two or more sets, the ‘concepts of projection and eylindric extension were proposed. For example, conser the set A= {(x,y) € F2|(2~1)? + (y—1)? <1} which is aration in Ux V = Fe Then the projection of A on U is Ay = [0,1] C U, and the projection of A on V (0,1) C V; see Fig. 4.1. The cylindric extension of Ay to Ux V = Fis (0, 1] x (-00,00) C R?. These concepts ean be extended to fuzzy relations. Definition 4.2. Let Q be a fuzzy relation in Uy x ---% Uy and fir, --sin} be a subsequence of {1,2,...,.n}, then the projection of Q on Uj, x --- x Ui, is a fuzzy relation Qp in U;, X=" Ui, defined by the membership function Har ()= ctg o BB etyeey Matsa) (4D) where (uj Ujena} the complement of {aug} with espect to {u1,- ). Asa special ease i Q is a binary fuzay relation in U x V, then the projection of Q on U, denoted by Qy, isa fuzzy sot in U defined by Ha, (2) = masnotew) (42)Sec. 4.1, From Clamical Relations to Fury Relations BL J Ate igure 4.1. Projections and eylndsic extensions of a re Iaton Note that (4.12) is still valid if Q is a crisp relation. For example, if Q is the crisp. relation A in Fig. 41, then its projection Qy defined by (4.12) is equal tothe Ay in Fig. 4.1. Hence, the projection of fuzzy relation defined by (4.11) is a natural extension of the projection of crisp relation. Example 4.4. According to (4.12), the projection of fuzzy relation (4.7) on U and V are the fuzry sets Qs =0.9/SF + 1/HK +0.95/Tokyo (4.13) and Qo =1/Boston + 0.9/HK (4.4) respectively. Sinilaly, the projections of A defied by (4.9) on U and V are the fuzzy sets AE, = | mage" fe = fe 5) [wee I f 1: (415) and ABa= [mage y= f afy (416) respectively. Note that AE; equals the crisp set and ABs equals the crisp set V. ° ‘The projection constrains a fuzzy relation to a subspace; conversely, the cylindric extension extends a fuzzy relation (or fuzzy set) from subspace to the whole space. Formally, we have the following definition,52 Fuzzy Relations andthe Extension incple_Chit Definition 4.3. Let Qp be a furry relation in Ui, X= x Uig and fin, -osia) a subsequence of {1,2,...r}s then the cylindric extension of Qp to Uy x°"+% Un a fuzzy relation Qpp in Us x--- x Uy defined by Hap (1s Un) = Hop (isso (aan) [As a special case, if Qs is fuzzy set in U, then the cylindric extension of Q to Ux V isa furay relation Qye in U x V defined by ase (#0) = Has(2) (48) ‘The definition (4.17) is also valid for crisp relations; check Fig. 4.1 for an example. Example 4.5. Consider the projections Qs and Qs in Example 4.4 (4.18) and (4.14). According to (418), thei cylindric extensions to U x V are Qi = 09/(SF, Boston) + 0.9/(SF, HK) + 1/(1HK, Boston) +1/(HK, HK) +0.95/(Tokyo, Boston) +085/(Tokyo, HK) (439) Qa = 1/(SF, Boston) + 1/(HK, Boston) + 1/(Tokyo, Boston) 40.9/(SP,HK) +0.9/(HK,HK) +09/(Tokyo,HK) (4.20) Similarly, the cylindric extensions of AE, and AB» in (4.15) and (4.16) to U x V i= ey) =UxV L Abies flew (42n) and Alas = [les =UxV (4.22) o From Examples 4.4 and 4.5 we see that when we take the projection ofa fuzzy relation and then cylindsically extend it, we obtain a fuzzy relation that is larger than the original one. To characterize this property formally, we frst introduce the concept of Cartesian product of fuzzy sets. Let At, Ay be fuzzy gets i Oty om Uy respectively. The Cartesian product of Ayy nn, denoted by Ay x +> x dys is @ fuzzy relation in U, x ++ x Un whose membership function is defined as Base ly) = Ha (ts) #20 Hae (Un) (423) ‘where + represents any t-norm operator. Lemma 4.1. If Q is @ fuzzy relation in Uy x +++ x Uy and Qty. Qn are its: projections on Ui,.-Un, fespectively then (see Fig. 4.2 for illustration) QCA Kx Qn (424)Compositions of Fuszy Relations 53 1 Q2=QiENQae Qe _ U Figure 4.2. Relation between the Cartesian product and Intersection of eylinle sets. ‘where we use min for the t-norm in the definition (4.23) of Q1 x--- x Qu. Proof: Substituting (4-11) into (4.17), we have ape (tts. Un) = est (4.25) Hence, Qc (4.26) for all i = 1,2,...m, where Qup is the cylindric extension of Q; to Uy x +++ x Ur. ‘Therefore, if we use min for intersection, we have QC QB MQuz =O Xx On 27 4.2 Compositions of Fuzzy Relations Let P(U,V) and Q(V, 1) be two crisp binary relations that share @ common set V. The composition of P and Q, denoted by P 0 Q, is defined as a relation in U x W such that (2,2) € P oQ if and only if there exists at least one y € V such ‘hat (x,y) € P and (y,2) € Q- Using the membership function representation of relations (see (4.5), we have an equivalent definition for composition that is given in the following lemma.54 Fuzzy Re Extension Pincile_ Ch. 4 Lemma 4.2, PQ is the composition of P(U,V) and Q(V,W) if and only if mpcalts#) = max tiuel2,¥) Holds?) (4.28) for any (2,2) € U XW, where ¢ is any tnorm. Proof: We first show that if PoQ isthe composition according tothe definition, then (4.28) is true. If P 0 Q is the composition, then (z,2) € Po Q implies that there exists y € V such that jp(2,y) = 1 and wo(y,2) = 1. Hence, wroa(2,2) = 1 = maxyey tlur(2,y),na(V,2)), that is, (4.28) is true. If (2,2) ¢ PoQ, then for any y € V either jp(z,y) = 0 oF pig(ty2) = 0. Hence, jpsa(z,2) = 0 = maxyev t\p(2,y), Holy, =)). Therefore, (4.28) is true for any (2,2) € U x W. Conversely, if (4.28) is true, then (x, 2) € P © Q implies maxyey thip(2.y), Ho(¥,2)] = 1, which means that there exists at least one y € V such that up(2,y) = 4alys2) = 1 (See Axiom t1 in Section 3.3); this is the definition, For (2,2) ¢ PoQ, ‘we have from (4.28) that maxyev t{p(2,¥),/19(v,2)] = 0, which means that there sno y €V such that p(z,y) = aly, 2) = 1. Therefore, (4.28) implies that PQ is the composition according to the definition. © [Now we generalize the concept of composition to fuzzy relations. From Lemma 4.2 we see that if we use (4.28) to define composition of fuzzy relations (suppose P and Q are fuzzy relatioins), then the definition is valid for the special case where P- and Q are crisp relations. Therefore, we give the following definition. Definition 4.4. The composition of fuzzy relations P(U,V) and Q(V,W), de- noted by P 0Q, is defined as a fuzzy relation in Ux W’ whose membership function is given by (428). Because the t-norm in (4.28) can take a varity of formulas, for each t-norm ‘we obtain a particular composition. The two most commonly used compositions in the literature are the so-called maz-min composition and maz-product composition, which are defined as follows: ‘© The max-min composition of fuzzy relations P(U;V) and Q(V,W) is a fuzzy relation Pe Q in U x W defined by the membership function rvo(2, 2) = maxminlur(2,v) Hal] (429) where (2,2) € Ux W. ‘© The maz-product composition of fuzzy relations P(U,V) and Q(V,W) is a fuzzy relation P © Q in U x W defined by the membership function mrca (2) = maxlue(2,yuo(vs2)] (4.30) where (2,2) €U x W.Sec 42._ Compositions of Fuzzy Relations 55 We see that the max-min and max-product compositions use minimum and algebraic product for the t-norm in the definition (4.28), respectively. We now consider two examples for how to compute the compositions. Example 4.6, Let U and V bedefined asin Example 42 and W = (New York City, Beijing}. Let P(U,V) denote the fuzzy relation “very far" defined by (4.7). Define the fuzzy relation “very near” in V x W7, denoted by Q(V,W), by the relational matrix w NYC Beijing V Boston 0.95 04 HK 01 09 (4.31) Using the notation (2.7), we can write P and Q as P =03/(SF, Boston) +0.9/(SF, HK) + 1/(HK, Boston) +0/(HK, HK) +0.95/(Tokyo, Boston) +0.1/(Tokyo, HK) (4.32) 95/(Boston, NYC) + 0.1/(Boston, Beijing) + 0.1/(HK, NYC) +0.9/(HK, Beijing) (433) We now compute the max-min and max-product compositions of P and Q. First, wwe note that U x W contains six elements: (SF,NYC), (SF,Beijing), (HK,NYC), (AK Bejing), (Tokyo.NYO) and (Tokyo,Beijing). ‘Thus, our task is to determine the membership values of pag at these six elements. Using the max-min composition (4.29), we have upsa(SF, NYC) = max{min{yp(SF, Boston), 4o(Boston, NYC)], min|ep(SF, HK), nq(HK, NYC)}} = maz(min(0.3,0.95),min(0.9,0.1)] =03 (434) Similarly, we have psg(SF, Beijing) = max{min|p(SF, Boston), ua{ Boston, Beijing), min|up(SF, HK), uq(HK, Beijing))} naz[min(0.3,0.1),min(0.9,0.9)} a (435) ‘The final PoQ is PQ =03/(SF,NYC) +0.9/(SF, Beijing) + 0.95/(HK,NYC) 40.1/(HK, Beijing) + 0.95/(Tokyo, NYC) +0.1/(Tokyo, Beijing) (436)56 Fuzzy Relations and the Extension Principle Ch. & If we use the max-product composition (4.30), then following the same procedure as above (replacing min by product), we obtain PQ =0285/(SF,NYC) + 0.81/(SF, Beijing) + 0.95/(HK, NYC) 40.1/(HK, Beijing) + 0.9025 (Tokyo, NYC) +0.095 /(Tokyo, Beijing) (437) ° From (4.36), (4.87) and the relational matrices (4.7) and (4.81), we see that: a simpler way to compute Po Q is to use relational matrices and matrix product. Specifically, let P and @ be the relational matrices for the fuzzy relations P(U, V) and Q(V,W), respectively. Then, the relational matrix for the fuzzy composition PoQ can be computed according to the following method: ‘+ For max-min composition, write out cach element in the matrix produet PQ, Dut treat each multiplication as a min operation and each addition as a maz ‘operation. ‘¢ For max-product composition, write out each element in the matrix product PQ, but treat each addition as a mar operation. ‘We now check that (4.36) and (4.37) can be obtained by this method, Specif- cally, we have 03 09 0309 1 o Jo( Qi Ga) =( 098 01 (438) 095 01 095 for max-min composition, and 03 09 : 0285 081 vO) o(am 92) (a Cr) am 095 1 ; 0.9025. 0.095 for max-produet composition. In Example 4.6, the universal sets U,V and W contain a finite number of ele- ments. In most engineering applications, however, the U,V and W are real-valued spaces that contain an infinite mumber of elements. We now consider an example for computing the composition of fuzzy relations in continuous domains. Example 4.7: Let U = V = W = R. Consider the fuzzy relation AE (ap- proximately equal) and ML (much larger) defined by (4.9) and (4.10) in Example 4.3, We now want to determine the composition AE ML. Using the max-produet composition, we have teny)* Te eo (4.40) Hawome(t;2)See.43._ The Extension Princip 87 To compute the right hand side of (4.40), we must determine the y € R at which REC achieves its maximum value, where # and 2 are considered to be fixed valued in R. The necessary condition for such y is 2 alto eo? (4a) ‘Because its impossible to obtain a closed form solution fr (4.1), we cannot further mplify (4.40). In practice, for given values of x and = we can first determine the numerical solution of (441) and then substitute it into (4.40). Comparing this example with Example 46, we see that fuzzy compositions in continuous domains fare much more difficult to compute than those in discrete domains. 4.3. The Extension Pri ple ‘The extension principe isa asic identity tha allows the domain of «function to be extended from crisp points in U to fuzzy sets in U. More specifically, let f : U + V_ be a function from crisp set U to crisp set V. Suppose that a fuzzy set A in U is ‘given and we want to determine a fuzzy set B = f(A) in V thet is induced by f. If {isan oneto-one mapping, then we can define wot) = Half), EV (4.42) where f-'(y) is the inverse of f, that is, f[f~"(y)] = y- If f is not one-to-one, ‘then an ambiguity arises when two or more distinct points in U with different ‘membership values in A are mapped into the same point in V. For example, we may have f(t) = f(2a) = y but a1 ¢ rs and alas) # a(za), so the right hand side of (4.42) may take two different values (21 = f-"(y)) of wale = f"(y))- To resolve this ambiguity, we assign the larger one of the two membership values to wa(y). More generally, the memabership function for B is defined as ota) =, mame wale), ve (43) where J-'y) denotes the set ofall points 2 € U such that f(2) = y. The identity (443) i called tho extension principle. Example 4.8. Let U = (1,2,..10} and f(a in U defined by 2°, Let small be a fuzzy set small = 1/1 +1/2+08/3+ 06/44 0.4/5 (444) ‘Then, in consequence of (4.43), we have small? = 1/1 +1/4+08/9+ 06/16 +0.4/25 (445)58 Fuzzy Re 4.4 Summary and Further Readings In this chapter we have demonstrated the following: ‘ The concepts of fuzzy relations, projections, and cylindric extensions. * The max-min and max-product compositions of fuzzy relations. ‘¢ The extension principle and its applications. ‘The basic ideas of fuzzy relations, projections, cylindric extensions, compositions of fuzzy relations, and the extension principle were proposed in Zadeh [19710] and Zadeh [1975]. These original papers of Zadeh were very clearly written and are still the best sources to learn these fundamental concepts, 4.5 Exercises Exercise 4.1. Given an rary relation, how many different projections of the relation ean be taken? Exercise 4.2. Consider the fuzzy relation Q defined in Uy x +++ x Uy where 1 = {a,byc},U2 = {ot}, Ue = {x,y} and Ue = {i,j}: Q=O4/(b,t,ysi) + 0.6/(a, 9,251) + 0.9/(0, 5,958) + 1/0,85469) 40.6/(a,t,u,J) +0.2/(c, 8,044) (a) Compute the projections of Q on Uy x Us x Ur, Ui x Us and Uy. (b) Compute the cylindric extensions of the projections in (a) to Ui xUpxUsxU. Bxercise 4.3. Consider the three binary fuzzy relations defined by the relational matrices oz 06 06 0 1 0 07 0 ),@=( 0 06 01),Q=( 0 1 0 1 0 a1 0 o7 0 4 (4/40) Compute the max-min and max-produet compositions Qr © Q2,Qx © Qs and Qh ° N\e ©, Exercise 4.4. Consider fay sot A = 0.5/—1-+0.8/0-+1/14-0.4/2 and function J(z) =27. Determine the fuzzy set f(A) using the extension principle Exercise 4.5. Compute the ageart (22) in Example 4.7 for (2,2 ,0),(,1)- (0,0), (0,1),Chapter 5 Linguistic Variables and Fuzzy IF-THEN Rules 5.1 From Numerical Variables to Linguistic Variables In our daily life, words are often used to describe variables. For example, when wwe say “today is hot,” or equivalently, “today’s temperature is high,” we use the word “high” to describe the variable “today’s temperature.” That is, the variable “today’s temperature” takes the word “high” as its value. Clearly, the variable “to- day's temperature” also can take numbers like 25%, 19%e, ete, as its values. When variable takes numbers as its values, we have a well-established mathematical framework to formulate it. But when a variable takes words as its values, we do not have a formal framework to formulate it in classical mathematical theory. In order to provide such a formal framework, the concept of linguistic variables was Introduced. Roughly speaking, if a variable can take words in natural languages as its values, it is called a linguistic variable. Now the question is how to formulate the words in mathematical terms? Here we use fuzzy sets to characterize the words. ‘Thus, we have the following definition Definition 5.1. Ifa variable can take words in natural languages as its valwes, It is called a nguistic variable, where the words are characterized by fuzzy sets defined in the universe of discourse in which the variable is defined, Example 5.1. The speed of a cars a variable x that takes values in the interval [0, Vise]: Where Vinar is the maximum speed of the car. We now define three fuzzy sets “slow,” “medium,” and “fast” in (0, Vnaz] a8 shown in Fig, 5.1. If we view 2 as a linguistic variable, then it ean take “slow,” “medium” and “fast” as its values ‘That is, we can say “z is slow,” “x is medium,” and “ is fast." Of course, x also ‘ean take numbers in the interval (0, Vee] a8 its values, for example, 2 = 50mph, 35mph, ete. 0 Definition 5.1 gives a simple and intuitive definition for linguistic variables. In the fuzzy theory literature, 2 more formal definition of linguistic variables was usu- 5060. Linguistic Variables ad Fuzzy IF-THEN Rules _Ch. 5 slow medium — fast pop of ea met) Oss oso) ome Figure 5.1. ‘The speed of car at a lnguistie variable that can take forzy sets Slow” "meium” and at” a ite vale, ally employed (Zadch [1973] and [1975}). This definition is given below. Definition 5.2. A linguistic variable is characterized by (X,,U, M), where: ‘¢ X is the name of the linguistic variable; in Example 5.1, X is the speed of the ‘© T’is the set of linguistic values that X can take; in Example 5.1, 7 = {slow, ‘medium, fast} ‘© U is the actual physical domain in which the linguistic variable X takes its ‘quantitative (crisp) values; in Example 5.1, U = [0, Vinasl- © M is a semantic rule that relates cach linguistic value in T with fuzzy set in U; in Example 5.1, M relates “slow,” “medium,” and “fast” with the membership functions shown in Fig. 5.1, ‘Comparing Definitions 5.1 with 5.2, we see that they are essentially equivalent. Definition 5.1 is more intuitive, whereas Definition 5.2 looks more formal. From these definitions we see that linguistic variables are extensions of numerical variables Im the sense that they are allowed to take fuzzy sets as their values; see Fig. 5.2. ‘Why is the concept of linguistic variable important? Because linguistic variables fare the most fundamental elements in human knowledge representation, When we ‘use sensors to measure a variable, they give us numbers, when we ask human experts, to evaluate a variable, they give us words, For example, when we use a radar gun ‘to measure the speed of a car, it gives us numbers like 39mph, 42mph, etc; when‘numerical variable Ainguistie valabte we ask a Inman to tell us about the speed of the ear, he/she often tells us in words like “it’s slow,” “it’s fast,” etc. Henco, by introducing the concept of linguistic variables, we are able to formulate vague descriptions in natural languages in precise ‘mathematical terms. This is the frst step to incorporate human knowledge into ‘engineering systems in a systematic and efficient manner. 5.2 Linguistic Hedges With the concept of linguistic variables, we are able to take words as values of (linguistic) variables. In our daily life, we often use more than one word to describe iable. For example, if we view the speed of a car as a linguistic variable, then its values might be “not slow,” “very slow,” “slightly fast,” “more or less medium,” etc. In general, the value of a linguistic variable is a composite term 2 = 2129+: 2, that is a concatenation of atomic terms 11,.02,...%,. These atomic terms may be classified into three grou + Primary terms, which are labels of fuzzy sets; in Example 5.1, they are “slow “medium,” and “fast.” ‘+ Complement “not” and connections “and” and “or 1+ Hedges, such as “very,” “slightly,” “more or less," ete. ‘The terms “not,” “and,” and “or” were studied in Chapters 2 and 3, Our task now is to characterize hedges.62 Fuzy Although in its everyday use the hedge very does not have a well-defined mean- ing, in essence it acts as an intensifier In this spirit, we have the following definition for the two most commonly used hedges: very and more or less, Definition 5.8. Let A be a fuzzy set in U, then very A is defined as a fuzzy set in U with the membership function cry aCe) = Wale)? 61) and more or less Ais 8 fuzzy set in U with the membership function Hmore or tees Alt) = [pale)]? (5.2) Example 5.2. Let U = {1,2,..5} and the furzy set small be defined as small = 1/1 + 08/2-+ 0.6/8 + 0.4/4 +0.2/5 63) ‘Then, according to (5.1) and (5:2), we have /1+0.64/2+0.36/3+0.16/4+0.08/5 (54) very (very small) /1 + 0.4006/2 + 0.1296/3 + 0.0256/4 +0.0016/5 (5) more or less small = 1/1 + 0.8944/2 + 0.7746/3 + 0.6825/4 40.4472/5 (56) 5.3. Fuzzy IF-THEN Rules Tn Chapter 1 we mentioned that in fuzzy systems and control, human knowledge is represented in terms of fuzzy IF-THEN rules. A fuzzy IF-THEN rule's conditional statement expressed as IF. < fuzzy proposition >, THEN < fuzzy proposition > (5.7) ‘Therefore, in order to understand fuzzy IE-THEN rules, we first must know what are fuzzy propositions. 5.3.1 Fuzzy Propositions ‘There are two types of fuzzy propositions: atomic fuzzy propositions, and compound fuzzy propositions. An atomic fuzsy proposition is single statement zie (8)‘ee. 5.3, Fuzzy I-THEN Rules 63, ‘where x is a linguistic variable, and A is a linguistic value of z (that is, A is ‘a fuzzy set defined in the physical domain of 2). A compound fuzzy proposition is a composition of atomic fuzzy propositions using the connectives “and,” “or,” and “not” which represent fuzzy intersection, fuzzy union, and fuzzy complement, respectively. For example, if represents the speed of the car in Example 5.1, then the following are fuzzy propositions (the first three are atomic fuzzy propositions and the last three are compound fuzzy propositions): zis 69) zisM (6.10) zisF ju) zis Sor zis not M (12) ‘ta not S ond x ie not F (23) (@ is S and x is not F) or 2 is M (a4) where $, M and F denote the fuzzy sets “slow,” “medium,” and “fast,” respectively, Note that in a compound fuzzy proposition, the atomic fuzzy propositions are independent, that is, the 2's in the same proposition of (5.12)-(5.14) can be different variables. Actually, the linguistic variables in a compound fuzzy proposition are in general not the same. For example, let x be the speed of a car and y = # be the acceleration of the car, then if we define fuzzy set large(L) for the acceleration, the following is a compound fuzzy proposition zis FandyisL Therefore, compound fuzzy propositions should be understood as fuzzy relations, How to determine the merabership functions of these fuzzy relations? ‘+ For connective “and” use fuzzy intersections. Specifically, let x and y be Tinguistic variables inthe physical domains U and V, and A and B be fuzzy sets in U and V, respectively, then the compound fuzzy proposition zis Aandy isB (6.15) Js interpreted as the fuzzy relation ANB in Ux V with membership function Hana(,y) = tlea(z), nay) (5.16) where (0,2] x [0,1] > (0,1 is any t-norm. ‘+ For connective “or” use fuzzy unions. Specifically, the compound fuzzy propo- sition risAoryisB (a7) "Note shat in Chapiors 2 and 8, A and B are fury set defined in the same universal st and AUB and AO B ace furey ses in U; here, AUB and ANB are fusny relations in = V, Where U may of ay’ not eal64 Linguistic Viable and Furry IF-THEN Rules Ch 5 Is interpreted as the fuzzy relation AUB in Ux V with membership function LHa(2), walu)) (5.18) where s : [0,1] x [0.1] - [0,1] is any s-norm, Havaley) ‘© For connective ‘not” use fuzzy complements. That is, replace not A by A, which is defined according to the complement operators in Chapter 3, Example 5.3. The fuzzy proposition (5.14), that is, FP = (xis $ and zis not F) or zis M (5.19) is a fuzzy relation in the product space [0, Vor]? with the membership function vv(oigza.20) = afthis(an) cu (oa)] waees)) (620) where s,t and ¢ are s-norm, t-norm and fuzzy complement operators, respectively, the fuzzy sets $ ~ small, M = medium, and F = fast are defined in Fig. 51, and oases ‘We are now ready to interpret the fuzzy IF-THEN rules in the form of (5.7). 5.3.2 Interpretations of Fuzzy IF-THEN Rules Because the fuzzy propositions are interpreted as fuzzy relations, the key question remaining is how to interpret the IF-THEN operation. In classical propositional caloulus, the expression IF p THEN q is written as p> q with the implication “+ regarded as a connective defined by Table 5.1, where p and q are propositional ‘variables whose values are either truth (T) or false (F). From Table 5.1 we see that. if both p and q are true or false, then p+ q is true; if pis true and q is false, then pqs false; and, if p is false and q is true, then p+ q is true. Hence, pq is, ‘equivalent to pva (621) and @aradve (6.22) in the sense that they share the same truth table (Table 5.1) as p>, where V land A represent (classical) logic operations “not,” “or,” and “and,” respectively Because fuzzy IF-THEN rules can be viewed as replacing the p and q with fuzzy propositions, we can interpret the fuzzy IF-THEN rules by replacing the V and ‘A operators in (5.21) and (5.22) with fuzzy complement, fuzzy union, and fuzzy intersection, respectively, Since there are a wide variety of fuzzy complement, fuzzy union, and fuzzy intersection operators, a number of different interpretations of furay IE-THEN rules were proposed in the literature. We list some of them below.‘See 5.3, Foray I-THEN Rates 65 ‘Dable 6.1, Truth table for p> @ Pa [psa TTP Tt tT Fl F Pot| t perl tr In the following, we rewrite (5.7) as IF_< PP, > THEN < FP, > and replace the p and q in (5.21) and (5.22) by FF; and FPs, respectively, where FP, and FP, ave fuzzy propositions, We assume that FP, is a fuzzy relation defined in U =U x-++x Up, Pe isa fuzzy relation defined in V = Vj x +++ Vy, and 2 and y are linguistic variables (vectors) in U and V, respectively. * Dienes-Rescher Implication: If we roplace the logic operators ~ and V in (6.21) by the basic fuzzy complement (3.1) and the basic fuzzy union (3.2), respectively, then we obtain the so-called Dienes-Rescher implication. Specifi- cally, the fuzzy IF-THEN rule IF < FP, > THEN < FP; >is interpreted as a fuzzy relation Qp in U x V with the membership function Hao (ay) = maz[l ~ per, (2), Hr rs(y)] (5.23) ‘+ Lukasiewicz Implication: If we use the Yager snorm (8.10) with w = 1 for the V and basic fuzzy complement (81) for the~in (5.21), we obtain the Lutasiewice implication, Specifically, the fuzzy IF-THEN rule IF < FP, > THEN < FP, > is interpreted as a fuzzy relation Qy, in U x V with the membership function May (Bu) = min{L, 1 ~ pre, (2) + weey(u)] (6.24) Zadeh Implication: Here the fuzzy IF-THEN rule IF < FP; > THEN < FP, > is interpreted as a fuzzy relation Qz in U x V with the membership funeti Has (zu) = maztmin(upr,(2),Hrrs(y))-4 ~ wer (2)] (5.25) Clearly, (5.25) is obtained from (5.22) by using basic fuzzy complement (3.1), basic fuzzy union (3.2), and basic fuzzy intersection (8.8) for -V and A, re. spectively. ‘* Gédel Implication: The Gédel implication is a well-known implication for- mula in classical logic. By generalizing it to fuzzy propositions, we obtainFuzzy IF-THEN Rules _Ch.5 the following: the fuzzy IP-THEN rule JF < FP, > THEN < FP, >is interpreted as a fuzzy relation Qo in U x V with the membership function Lrt—~—— (6.26) 1 is interesting to explore the relationship among these implications. The fol- lowing lemma shows that the Zadeh implication is smaller than the Dienes-Rescher implication, which is smaller than the Lukasiewicz implication. Lemma 5.1. For all (7,9) € U x V, the following is true Ha 2(0) $ Hap(2s¥) < Hae (ev) (27) Proof: Since 0 < 1—yens(2) < 1 and 0 < pres(y) <1, we have mas[l — Hep, (2), Hep,(u)) <1 — wer (2) + wer, y) and maz(l — per, (z),yer,(y)] <1. Hence, gp (ty) = mall ~ en (2), wers(y)) < min(L, 1 seen, (@) + raya) = Hax(z,y). Since min|upp,(z), ure, (y)] < urr,(y), we have maz[min(ure,(z), se) tune (o)] $ maziuen Iwan (2) which say (Bs9) < Has ev) Conceptually, we can replace the GV and A in (5.21) and (6.22) by any fuzzy complement, s-norm and t-norm, respectively, to obtain @ particular interpretation. So a question arises: Based on what criteria do we choose the combination of fuzzy complements, s-norms, and t-norms? This is an important question and we will discuss it in Chapters 7-10. Another question is: Are (5.21) and (5.22) still “equivalent” to p ~» g when p and q are fuzzy propositions and what does this equivalent” mean? We now try to answer this question. When p and q are crisp propositions (that is, p and g are either true or false), pq is a global implication the sense that Table 5.1 covers all the possible cases. However, when p and q are fuzzy propositions, p + q may only be a local implication in the sense that p> q hhas large truth value only when both p and q have large truth values. For example, when we say “IF speod is high, THEN resistance is high,” we are concerned only ‘with a local situation in the sense that this rue tells us nothing about the situations when “speed is slow,” “speed is medium,” ete. Therefore, the fuzzy IF-THEN rule IF THEN (5.28) should be interpreted as IP THEN ELSE < NOTHING > (5.29) where NOTHING means that this rule does not exist. In logic terms, it becomes PI G=PAg (6.30) Using min or algebraic product for the ‘ in (5.90), we obtain the Mamdani impli- cations,‘See. 5:3, Fuzzy IF-THEN Rules er ‘© Mamdani Implications: The fuzzy IF-THEN rule (5.28) is interpreted as a fuzzy relation Qaray or Quip in U x V with the membership function Haase (29) = min|urr, (2). Hers(V)] 31) Hauer (20) = Hee, (2) UV) (6.32) ‘Mamdani implications are the most widely used implications in fuzzy systems and fuzzy control. They are supported by the argument that fuzzy IF-THEN rules are local. However, one may not agree with this argument. For example, one ‘may argue that when we say “IF speed is high, THEN resistance is high,” we implicitly indicate that “IF speed is slow, THEN resistance is low.” In this sense, fury TF-THEN rules are nonlocel. This kind of debate indicates that when we represent human knowledge in terms of fuzzy IF-THEN rules, different people have different interpretations. Consequently, different implications are needed to cope with the diversity of interpretations. For example, if the human experts think ¢hat ther rules are local, then the Mamdani implications should be used; otherwise, the global implications (5.23)-(5.26) should be considered. ‘We now consider some examples for the computation of Qn, Qr,Qz,Qurss and Our. Example 5.4. Let x1 be the speed of a car, zy be the acceleration, and y be ‘the force applied to the accelerator. Consider the following fuzzy TF-THEN rule: TP 2, i9 slow and 23 is small, THEN y is large 6.33) where “slow” is the fuzzy set defined in Fig. 5.1, that is, 1 if <3 Haoelts)=4 See if 25 < x <5 6a 0 if m>55 “small” is a fuzzy set in the domain of acceleration with the membership function ={ 98 if 02 Let the domains of 24,22 and y be U; = (0,100],U2 = 0,30], and V = [0,3], respectively. If we use algebraic product for the t-norm in (5.16), then the fuzay proposition Hearge(¥) -{ y FP, = 2 is slow and 22 is small (537)68 Linguistic Variables and Fuzzy IF-THEN Rules _Ch. 5 bs small a5) co x inure 5.3. stration for how t0 compute Batowl1)semat 2) i Example 5 is a fuzay relation in Ui x Uz with the membership function rr (21%) = dsowl21)Hamat(22) 0 if 255 ory >10 -{ oe af x) < 35 and 22 < 10 (6.38) s-esfho-e) 7 35 < ry <55 and zy < 10 Fig. 5.8 illustrates how to compute pep, (11,22). If we use the Dienes-Rescher implication (5.28), then the fuzzy IF-THEN rule (6.83) is interpreted as a fuzzy relation Qp(xi,z2,y) in Ur x Us x V with the ‘membership function Hao (12, ¥) = maz[l ~ pr, (2,72), Marge) (5.39) From (5.38) we have 1 if 1 2550r2>10 1 pers es) = nf if mS5andm<10 (6.40) 1 eygosad if 35cm < 55 and ay <10 ‘To help us combining 1 ~ purr (21,32) of (5-40) with puarge(¥) of (5.36) using the ‘maz operator, we illustrate in Fig. 5.4 the division of the domains of 1—rp, (21,22)Sec. 6.3. Furay I-THEN Roles 69 ley yel (xx) xpSS yas Sexje55 for x>10 and x3<10 and x3€10 Figure 5.4, Divion of the domain of 1 ar (e122) fd vnacgal) and thie combinations for Exam 9 and jHorge(¥) and their combinations, From Fig. 54, we obtain bao 21.2259) = 1 ifm > 55 0r 2g > 10 or y>2 24/10 if 21 $35 and 2) < 10 and y <1 1 epsom) §F35 alpwalé) TeTP TY] ET) TY] FF tle} e|ao|r |r ie Flr) re | at] |r lr Fletriel| | rir other logic function. Logic formulas are defined recursively as follows: ‘© The truth values 0 and 1 are logic formulas. ‘» If pis a proposition, then p and p are logie formulas. 1s If p and g are logic formulas, then pV q and p Aq are also logic formulas. ‘+ The only logic formulas are those defined by (a)-(c). When the proposition represented by a logic formula is always true regardless of the truth values of the basic propositions participating in the formula, itis called a tautology: when itis always false, itis called a contradiction, Example 6.1. The following loge formulas are tautologies: (>a) (Va) (6.1) 406 (@Adve) (62) ‘To prove (6.1) and (6.2), we use the truth table method, that is, we list all the ‘possible values of (6.1) and (6.2) and see whether they are all true, Table 6.2 shows the results, which indicates that (6.1) and (6.2) are tantologies. © Proof p+ 4) + (Ba) and (P+) +4 (ena) VB TDA [STIR GVO LOI) ADD. T T aaah] masa aaa aaa erry Various forms of tautologies can be used for making deductive inferences. They are referred to as inference rules. ‘The three most commonly used inference rules‘See. 6: From Casical Logi to Fuzzy Logie 7 ‘* Modus Ponens: This inference rule states that given two propositions p and p> q (called the premises), the truth of the proposition q (called the conclusion) should be inferred, Symbolically i is represented as PARA) >a (6.3) A more intuitive representation of modus ponens is Premisel: ris A Premise2: IF xis ATHEN yis B Conclusion: — yis B ‘+ Modus Tollens: This inference rule states that given two propositions g and Pp ~ 4g, the truth of the proposition p should be inferred. Symbolicaly, it Decomes GA@>9) > (4) A more intuitive representation of modus tollens is, Premise1: — yis not B Premise2: IF is ATHEN yis B Conclusion: x is not A © Hypothetical Syllogism: This inference rule states that given two propos- tions p -» q and q — r, the truth of the proposition p— r should be inferred. Symbotically, we have (PIDAG>N) +>") (65) A more intuitive representation of it is Premise: IF 2i¢@ ATHEN yis B Premise2: IP yis BTHEN zisC Conclusion: IF 2 is ATHEN zis C 6.1.2 Basic Principles in Fuzzy Logic In fuzzy logic, the propositions are fuzzy propositions that, as defined in Chapter 5, are represented by fuzzy sets. The ultimate goal of fuzzy logie is to provide foundations for approximate reasoning with imprecise propositions using fuzzy set theory as the principal tool. To achieve this goal, the so-called generalized modus ‘ponens, generalized modus tollens, and generalized hypothetical syllogism were pro- posed. They are the fundamental principles in fuzzy logic.16 Fumy Logi and Approximate Reasoning _ Ch. 6 ‘+ Generalized Modus Ponens: This inference rule states that given two fuzzy propositions 2 is A’ and IF x is A THEN y is B, we should infer a new fuzzy proposition y is B such that the eloser the A to A, the closer the B” to B, where A,A',B and B’ are fuzzy sets; that is, Premise: zis Al Premise 2: IF xis ATHEN yis B Conclusion: — yis BY ‘Table 6.3 shows the intuitive criteria relating Premise 1 and the Conclusion in generalized modus ponens. We note that if a causal relation between “2 AP and “y is B” is not strong in Premise 2, the satisfaction of criterion p3 and criterion p5 is allowed. Criterion p7 is interpreted as: “IP z is A THEN vis B, ELSE y is not B.” Although this relation is not valid in classical logic, ‘we ofien make such an interpretation in everyday reasoning, Intuitive enter relating Premise 1 and the Conclusion forgiven in generalled modus pontns Fie 7 Premise 1) | ys BY (Conclusion) ‘aitarion pi ZA ya criterion p2| 2 is very A yis very B criterion p3.| zis very A vis B criterion p | 2 is more or less A | y is more or less B criterion p5 | 2 is more or less A visB criterion p6 | is not A ys unknown criterion p? | __ is not 4 wis not B ‘+ Generalized Modus Tollens: This inference rule states that given two fuzzy propositions y is B' and IF 2 is A THEN y is B, we should infer a new fuzzy proposition 2 is 4’ such that the more difference between B' and B, the ‘more difference between A’ and A, where A’, A, B’ and B are fuzzy sets; that is, Premise: yis BY Premise2: IF xis ATHEN yis B Conclusion: zis A! ‘Table 6.4 shows some intuitive criteria relating Premise 1 and the Conclusion in generalized modus tollens. Similar to the criteria in Table 6.8, some criteria in Table 6.4 are not true in classical logic, but we use them approximately in cour daily life.Sec. 6.1. From Classical Loge to Fuzzy Logic 7 ‘Table 6.4 Intuitive criteria relating Premise 1 and the Conclusion for given Premise 2 in generalized modus tllens. yD (Prenie 1) [ais A” (Conclusion) —] ‘terion tT yi not B zis not a caiterion 2 | yis not very B isnot very A criterion 8 | y is not more ot less B | 2 is not mare ot less A criterion 4 yisB ‘ris unknown criterion 65 yis B ig ‘+ Generalized Hypothetical Syllogism: This inference rule states that given ‘wo fuzzy propositions IF 2 is A THEN y is B and IF y is B! THEN 2 is C, ‘we could infer a new fuzzy proposition IF x is A THEN z is C’ such that the closer the B to B, the closer the C" to C, where A,B, B',C and C! are fuzzy seta; that is, Premise: IF ris ATHEN yia B Premise2: IF yis B! THEN is C Conclusion: IF 2 is ATHEN zis C! ‘Table 6.5 shows some intuitive criteria relating y i# BY with z is C’ in the «generalized hypothetical syllogism. Criteria 62 is obtained from the following intuition: To match the B in Premise 1 with the B’ = very B in Premise 2, we may change Premise 1 to IF x is very A THEN y is very B, so we have IP xis very A THEN = is C. By applying the hedge more or less to cancel the very, we have IF x is A THEN 2 is more or less C, which is criterion s2 Other criteria can be justified in a similar manner. ‘Table 6.5. Intuitive crlterla relating y fe Bn Premise 2 and = ie C" inthe ‘Conclusion in generalized hypothesieal sgt, UBB" Premise 9) [7 8 C (Conclusion) | ‘aiterion aT yee TiC ctiterion 2 | yisvery B |= is more or ess C criterion 3 | yis very B zc | ion 94 | yis more or ess B | 2 is very criterion 55 | y is more or loos B rc criterion 6 |" yis not B 218 unknown evitaion s7 | yis not B z isnot C38 Furzy Logic and Approximate Reasoning Ch. 8 We call the criteria in Tables 6.3-6.5 intuitive criteria because they are not nec- cessarily true for a particular choice of fuzzy sets; this is what approzimate reasoning means, Although these criteria are not absolutely correct, they do make some sense. ‘They should be viewed as guidelines (or soft constraints) in designing specific infer- We have now shown the basic ideas of three fundamental principles in fuzzy logic: generalized modus ponens, generalized modus tollens, and generalized hypothetical syllogism, The next question is how to determine the membership functions of the fuzzy propositions in the conclusions given those in the premises. The compositional rule of inference was proposed to answer this question, 6.2 The Compositional Rule of Inference ‘The compositional rule of inference is a generalization of the following, procedure (referring to Fig. 6.1): suppose we have a curve y = f(x) from x €U toy € V and are given z =a, then from ¢= a and y= f(z) we ean infer y = b = f(a). Let us generalize the above procedure by assuming that ais an interval and f(r) is an interval-valued function as shown in Fig. 6.2. To find the interval b which is inferred from a and f(z), we frst construct a cylindrical set ag, with base a and find its intersection 7 with the interval-valued curve. Then we project J on V’ yielding the interval b yev foo xeU Figure 6:1 Inferring y= b ftom 2 =a snd y= J(2) Going one step further in our chain of generalization, assume the AT is a fuzzy set in U and Q is a fuzzy relation in U x V. Again, forming a cylindrical extension,Sec.6.2_ The Compositional Rule of Inference 79 Figure 6.2. Infering interval b from interval @ and interval-valued fanetion f(2) Figure 6.3. lafersing fuxay set D” from furry set A’ and fussy relation Q. Ap of A and intersecting it with the fuzzy relation @ (soe Fig. 6.3), we obtain a fuzzy set AiG which is analog of the intersection Tin Fig. 6.2. Then, projecting 4g Q on the y-axis, we obtain the fuzzy set Bt More specifically, given jz4-(2) and woz, y), we have pala) (66) bay (20)80 Reasoning Ch. 6 (see (4.18)) and, consequently, agra(9) = thing (eu) nalesu)] = tua (e)nalew)] (62) Finally, from (4.12) we obtain B’, the projection of 4’ Q on V, as or(v) = sup ta (2), n9(29)) (68) (68) is called the compositional rule of inference. In the literature, the syibol “+” is often used for the t-norm operator, so (6.8) is also written as ory) = supln (2) + aleu)] 6.9) ‘The compositional rule of inference is also called the sup-star composition In Chapter 5, we learned that a fuzzy IF-THEN rule, for example, IF 2 is A THEN y is B, is interpreted as a fuzzy relation in the Cartesian product of the domains of 2 and y. Different implication principles give different fuzzy relations; see (5.23)-(5.26), (5.31), and (5.32). Therefore, the Premise 2s in the generalized modus ponens and generalized modus tollens can be viewed as the fuzzy relation Q in (6.9). For generalized hypothetical syllogism, we see that itis simply the compo- ion of two fuzzy relations, so we ean use the composition (4.28) to determine the conelusion. In summary, we obtain the detailed formulas for computing the con- clusions in generalized modus ponens, generalized modus tollens, and generalized hypothetical syllogism, as follows: '* Generalized Modus Ponens: Given fuzzy set A’ (which represents the premise « is A’) and fuzay relation A+ B in U x V (which represents the premise JF 2 is A THEN y is B), a fuzzy set BY in V (which represents the conclusion y is B!) is inferred as pi(y) = sup tivar(2) Hae (e9)) (6.10) * Generalized Modus Tollens: Given fuzzy set B’ (which represents the premise y is B") and fuzzy relation A> B in U x V (which represents the premise IF x is A THEN y is B), a fuzzy set A’ in U (which represents the conclusion 2 is A") is inferred as H4r(2) = sap thae(y)sH4-+6( 9) (6.11) * Generalized Hypothetical Syllogism: Given fuzzy relation A -+ B in U x V (which represents the premise IF z is A THEN y is B) and fuzzy relation B! -+ Cin V x W’ (which represents the premise IF y is B! THEN =is C), a fuzzy relation A> C’ in U x W (which represents the conclusion JP is A THEN 2 is C’) is inferred as wp thaa-rn(esy),er-r0(V,2)] (6.12) teascres2) = a Using different t-norms in (6.10)-(6.12) and different implication rules (6.23)- (6.26), (6.31) and (5.52), we obtain a diversity of results. These results show the properties of the implication rules. We now study some of these properties 6.3. Properties of the Implication Rules In this section, we apply specific implication rules and t-norms to (6.10)-(6.12) and see what the pr (y),s14»(2) and p.4-40*(2,2) look like for some typical cases of A’ and B'. We consider the generalized modus ponens, generalized modus tollens, and generalized hypothetical syllogism in sequal 6.3.1 Generalized Modus Ponens Example 6.2. Suppose we use min for the tnorm and Marans product impli cation (5.52) for the 414-2 (2, y) in the generalized modus ponens (6.10). Consider four cases of A’: (a) A’ = A, (b) A’ = very A, (c) A! = more or less A, and (a) A = A. Our task is to determine the corresponding B’. We assume that sup,crra(2)] = 1 (the fuzzy set A is normal). If A! = A, we have (4) sup oninia(e) ae) o(a) = suplea(=hin(u)] = nely) (6.13) HA’ very A, we have te (9) = sp (min (2) aleno(a (19 Since sup,cy(ja(z)] = 1 and x ean take any values in U, for any y € V there exists 2 €U such that j4(e) > wa(y)- Thus (6.14) can be simplified to voy) = soplaloun (a) = woly) (6.15) wa’ jore or less A, then from w'{"(¢) > a(x) > wa(z)un(z), we have Hoey) = sup (minlu}{"(2), Hale uo())) =swiv) (6.16)82 Funny Logi and Approximate Reasoning _Ch. 6 Finally, if A! = A, we obtain dary) = suptmin(t — Hala), Halz)en(9))) (6.17) Since for fixed y € V, wa(2)up(y) is an increasing function with f.4(2) while 1 a(x) isa decreasing function with 44(2), the sup,eq min in (6.17) is achieved when 1 = ea(2) = HaCe)sn(a) that is, when (2) = segs. Hence, no(v) T+ as) From (6.13), (6.15), (6.16), (6.18), and Table 6.3 we see that the particular gener- alized modus ponens considered in this example satisfies critera pl, p3 and p5, but does not satisy criteria p2, p4, pO, and p7. O Example 6.3. In this example, we still use min for the tnorm but use Zadeh implication (5:25) for the 4-+9(2;y) in the generalized modus ponens (6.10 Again, we consider the four typical cases of A" in Example 6.2 and assume that supzev|ala)) = 1 (a) For A’ = A, we have dae (y) = (6.18) or(y) = sup min{u4(z),maz{min(ya(z),49(y)), 1 — a(z)]} (6.19) Since supzct:#a(2) = 1, the supzey min in (6.19) is achieved at the particular zo €U when algo) = mazimin(s14(20), 42 (u)),1 ~ Hale0)] (6.20) Tf pa(2o) < aly), then (6.20) becomes: Halo) = maa(20)s1 ~ Ha(20)] (621) which is true when jt4(¢o) > 0.5; thus from (6.19) and (6.20) we have p(y) = 1420). Since supzepljta(2)] = 1 it must be true that y.a(ro) = 1, but this leads to un(d) > algo) = 1, which is impossible. Thus, we cannot have ja(o) < aly). Now consider the only possible case s.a(70) 2 s4n(y). Tn this case, (6.20) becomes a(zo) = maz[yn(y),1 — Haleo)] (6.22) If yp(y) < 1~na(to), then jra(z0) = 1~p2a(z0), which is true when jra(z0) = 0.5. If oly) = 1 palao), then from (6.22) we have jra(z0) = nly) > 0.5. Hence, a(zo) = maz[0.5, 4p(y)] and we obtain or (y) = palo) = maz[0.5,45(0)] (6.23) (b) For A! = very A, we have ary) = sup min{u4(2),maz{min(na(2),noly))s1 — wale)]} (6.24)‘See. 6.3. Propenias ofthe Implication Rules 83 Similar to the A' = A case, the sup,cy min is achieved at 2» € U- when H44(0) = mazimin(H.a(20), Hew), 1 — Ha(zo)] (6.25) alco) < pny) then HA(z0) = maz|ja(=o),1— paleo) (6.26) which is true only when j14(79) = 1, but this leads to the contradiction 4(y) > 1. Thus a(t0) 2 wo) the only posble cas, I wa(za) 2 waly), then (625) HA(zo) = maz{n(y),1 — #a(za)] (627) If waly) < 1 — wa(zo), then p4(z0) A-1, Hence, if way) <1— alta) — aloo), which is true when j24(¢9) = 94, we have p(y) = wh(0o) = 34. It p(y) > 1 ~ naleo), we have a(u) = 14(20) = way) > 32. In summary, we 3- v5 obtain Hp (y) = H4(0) = mace: shely)) (6.28) (©) A’ = more o leas A, we have ‘4 (v) = sup min{al(2),masininina(@) mata) ~ wale)]) (629) ‘where the supggy minis achieved at x9 €U when W300) = marlin (wa(zo),na(y)),4 — Ha(to)] (630) Similar to the A! = very A case, we can show that j14(z9) < p(y) is impossible. For #a(to) 2 Holy), we have Hi{?(@0) = maz{un(y),1 - walzo)] (631) If wp(y) <1 paleo), then p'{®(29) = 1 azo), which is true when ja(zo) = 238. Thus, if unly) <1 alo) = 41, we have p(y) = wi{"(z0) = Se. It p(y) > 1 — malo), we have ysr(y) = 1{*(20) = wn(y) > 4=*. To summarize, we obtain v5 2 es(y) = wi{*(20) = mast atu) (6.32) (4) Finally, when 4’ = 4, we have tuon(u) = sup min{1 — a(x), mazlmin(ya(z),ae(W))s1—Hale)]} (6:88)84 Fuzzy Loge and Approximate Reasoning Ch. 6 By inspecting (6.33) we see that if we choose zp € U such that 14(20) = 0, then 1 pa(zo) = 1 and maz{min(y.a(z), aa(y)),1— Ha(2)] = 1, thus the supzey min is achieved at x = zo. Hence, in this case we have voy) =1 (6.34) From (6.23), (6.28), (6.32), and (6.34), we see that forall the criteria in ‘Table 6.3, only criterion p6 is satisfied. (This approximate reasoning is truely approsimatel) a 6.3.2 Generalized Modus Tollens Example 6.4. Similar to Example 6.2, we use min for the tnorm and Mamdani’s product implication (6.32) for the j24-.n(t,v) im the generalized modus tollens (6.11). Consider four cases of BY: (a) B' = B, (b) B' = not very B, (c) B' = not more or less B, and (d) B' = B. We assume that stpycylis(y)] = 1. If BY = B, we have from (6.11) that (2) = sup inl — poly) nae] (6.5) The supycy min is achieved at yo € V when 1 — jsn(vo) = saCe)in(v0), which implies pve) = raza hence, ale) =1~ tole) = eae (30) IB! = not very B, then ba(e) = sup mint — p(y) Hale)ue()) (6.37) where the supyey min is achieved at yp € V when 1—15(yo) = wa(z)an(yo), which gives po(gn) = VADHA Hence, aCe VAT EA he) (e35) Hae(2) = Hale) (00) = If BY = not more or lesa B, we have a2) = sup minfh — 1S) pala )nn(V) (639) Again, the supyey min is achieved at yo € V when 1 ~ pi{"(yo) = a(z)s0(v0)s wich gs en) = S282 YEEE, ene, + 2walz) — eq) +1 (6.40) (2) = Hale)ua(w) = TatSee. 6.3. Properties ofthe Implication Rules 85, Finally, when B" = B, we have (2) = sup minlne(y) ae)na(o)] sup aCe )un(y) vey = ale) (oa) From (6.36), (6.38), (6.40) and (6.41) we see that among the seven intuitive criteria in Table 6.4 only criterion t5 is satisfied. © 6.3.3 Generalized Hypothetical Syllogism Example 6.5. Similar to Examples 62 and 6.4, we use min for the t-norm and Maméani product implication for the wa (2s4) and jy—-o(¥,2) in the general- ized hypothetical syllogism (6.12). We assume suey jin()] = 1 and consider four typical cass of B': (a) B' = B, (b) BY — very Bs (0) B' = more or less B, and (@) BB. If B= B, we have from (6.12) that asver(t2) = sup minlya(z)na(y),Hey)Hc(=)] sup wa (y))minfal2) wo(2)] vin|wa(z), wo(2)] (6.42) If BY = very B, we have waror(t,2) = sup miniual2)ea(0), Haw eol2)] (6.43) If wale) > wo(2), then it is always true that pra(2)pin(y) > sh(a)we(=), thus, a-ror(22) = S0Pyex HB )HC(2) = Ho). Wiza(2) < (2), then the supgey min Js achieved at yo €'V, when 14(2)nn\o) = wn)vic(=), which gives w9{do) = FALE; hence, in this case wascx(2,2) = Ha(2)se(yo) = SAB. In summary, we obtain cls) if pole) < male) ws if nole) 2 ale) If BY = more or less B, then using the same method as forthe B= very B ease, wo have pacser(ay2) = { (6.44) pale) if pale) ole) Cae) Finally, when B’ = B, we have Haver (@2) = sup miniua(2)un(y) (0 ~ ae(o))wo(2] (6.16)86. Fuzy Logi and Approximate Reasoning _Ch. 6 where the supyey minis achieved at yo € V when j4(z)48(40) = (1-15(b0))Hc(=)s that is, when ja(yo) = z—#@ELay. Hence, bale)uo(2) Hatz) + Hol®) ay Bavor(2,2) 6.4 Summary and Further Readings In this chapter we have demonstrated the following: + Using truth tables to prove the equivalence of propositions. ‘Basic inference rules (Modus Ponens, Modus Tollen, and Hypothetical Sylo- sism) and their generalizations to fuzzy propositions (Generalized Modus Po- zens, Generalized Modus ‘Tollens, and Generalized Hypothetical Syllogist), + The idea and applications of the compositional rule of inference. ‘© Determining the resulting membership functions from different implication rules and typical cases of premises. ‘A comprehensive treatment of many-valued logic was prepared by Rescher [1960] ‘The generalizations of classical logic principles to fuzzy logic were proposed in Zadeh [1973], Zadeh (1975) and other papers of Zadeh in the 1970s. ‘The compositional rule of inference also can be found in these papers of Zadeh. 6.5 Exercises Exercise 6.1. Use the truth table method to prove that the following are ‘autologies: (a) modus ponens (6.3), (b) modus tollens (6.4), and (c) hypothetical syllogism (6.5) Exercise 6.2. Let U = {21, 22,23} and V = {y1,ua}, and assume that a fuzzy IF-THEN rule “IF 2 is A, THEN y is B” is given, where A = 5/2, + 1/22 + 6/2 and B = 1/y-+A/ya. ‘Then, given a fact “xis A'" where A’ = 6/2,+.9/2a+.7/22, use the generalized modus ponens (6.10) to derive a conclusion in the form “y is BI’ where the fuzzy relation A+ B is interpreted using: (9) Dienes-Rescher implication (5.28), (0) Lulasiewice implication (5.24), (6) Zadeh implication (5.25), and (@) Mamdani Product implication (6.32).Sec. 65. Exercises 87 Exercise 6.3. Repeat Exercise 6.2 with A= .6/z:+1/za+ 9/2», B= 6/ui+ Mp, and A! = 5/21 +.9/20+ Vas. Exercise 6.4. Let U,V, A, and B be the same as in Exercise 6.2. Now given a fact “y is BY where B= 9/y,+.7/ya, use the generalized modus tollens (6.11) to derive a conclusion “2 is 4." where the fuzzy relation A -> Bis interpreted using (2) Lukasiewicz implication (5.24), and (b) Mamdani Product implication (5.32) Exercise 6.5. Use min for the t-norm and Lukasiewics implication (5.24) for the j4+(2,y) in the generalized modus ponens (6.10), and determine the membership function yp(y) in terms of yn(y) for: (a) A’ = A, (b) A! = very A, (©) A= more or less A, and (@) A’ Exercise 6.6. Use min for the t-norm and Dienes-Rescher implication (5.23) for the 14+8(2,y) in the generalized modus ponens (6.10), and determine the ‘membership function p(y) in terms of un(y) for: (a) A’ = A, (b) A! = very A, (©) A’ = more or less A, and (d) A! = A. Exercise 6.7. With min as the tnorm and Mamdani minimum implication (631) for the 14-2(2,y) in the generalized modus tollens (6.11), determine the ‘membership function j(z) in terms of y.4(2) for: (a) B' = B, (b) B' = not very By (6) B' = not more or less B, and (a) BY= B. Exercise 6.8. Consider a fuzzy logic based on the standard operation (min, max, 1 ~a). For any two arbitrary propositions, A and B, in the logic, assume that we requite that the equality an v(AnB) (6.48) holds. Imposing such requirement means that pairs of truth values of A and B become restricted to a subset of [0,1]?. Show exactly how they are restricted.88. Fuzzy Logic and Approximate Reasoning cnPart I Fuzzy Systems and Their Properties We learned in Chapter I that a fuzzy system consists of four components: fuzzy rule base, fuzzy inference engine, fuzzfier and defuzzifier, as shown in Fig. 1.5. In this part (Chapters 7-11), we will study each of the four components in detail. We will see how the fuzzy mathematical and logic principles we learned in Part I are used in the fuzzy systems, We will derive the compact mathematical formulas for different types of fuzzy systems and study their approximation properties. In Chapter 7, we will analyze the structure of fuzzy rule base and propose a number of specific fuzzy inference engines. In Chapter 8, a number of fuzzifiers and defuzzifiers will be proposed and analyzed in detail. In Chapter 9, the fuzzy Inference engines, fuzzfiers, and defuzzifiers proposed in Chapters 7 and 8 will be combined to obtain some specific fuzzy systems that will be proven to have the universal approximation property. In Chapters 10 and 11, the approximation accuracy of fuzzy systems will be studied in detail and we will show how to design fuzzy systems to achieve any specified accuracy. 89Chapter 7 Fuzzy Rule Base and Fuzzy Inference Engine Consider the fuzzy system shown in Fig. 1.5, where X Up x0 x Un CRO and V CR. We consider only the multi-input-single-output case, because a multi ‘output system can always be decomposed into a collection of single-output systems. For example, if we are asked to design a 4-input-3-output fuzzy system, we can frst design three 4-input-I-output fuzzy systems separately and then put them together as in Fig. 71. -input-1-ourpuy tuzzy system — Linput-1-outpu . x3 fuzzy system x a tuzzy system Figure 7-1. A mult-inpur-multi-output foray system fat be decomposed into a collection of multsinputsoge futpet fay systems. In this chapter, we will study the details inside the fuzzy rule base and fuzzy 90Sec. 7, Faray Rule Base o1 inference engine; fuzzifiers and defuzzifiers will be studied in the next chapter. 7.1 Fuzzy Rule Base TAAL Structure of Fuzzy Rule Base A fuazy rule base consists of a set of fuzzy IF-THEN rules. Its the heart of the fuzzy tystem in the sense that all other components are used to implement these rules in ‘reasonable and efficient’ manner. Specifically, the fuzzy rule base comprises the following fuzzy IF-THEN rules: Ru TF 2 30 A and... and aq is AL, THEN yis BE (7.1) where A! and BY are fuzzy sets in Us C Rand VC R, respectively, and x = (21,22, --t9)” € U and y € V are the input and output (Linguistic) variables of the fuzay system, respectively. Let. M be the number of rules in the fuzzy rule base; that is, = 1,2,.. Min (7.1). We call the rules in the form of (7.1) canonical fuzzy IP-THEN rules because they include many other types of fuzzy rules and fuzzy propositions as special cases, as shown in the following lemma. Lemma 7.1. The canonical fuzzy IF-THEN rules in the form of (7.1) include the following as special cases: (a) “Partial rules" IF 2, is Al and ... and tm i9 Aly, THEN y is BI (72) where m in Rul? is defined according to various implications (5.23)-(5.26), (5.31), and (5.82). If we accept the first view of a set of rules, then the M rules in the form of (7-1) are interpreted as a single fuzzy relation Qy in U x V defined by Qu=Urv® cay) This combination is called the Mamdani combination. If we use the syimbol + to represent the e-norm, then (7.14) ean be rewritten as Hay (1) = tmaor 94> Ho 9) (ra5) For the second view of a set of rules, the M fuzzy IF-THEN mules of (7-1) are interpreted as a fuzzy relation Qo in U x V, which is defined as Qc= [au 716) or equivalently, He CU) = Haun O61) #°-4 Hn) Gan where + denotes t-norm. This combination is called the Gédel combination. Let A" be an arbitrary fuzzy set in U and be the input to the fuzzy inference engine. Then, by viewing Qy or Qg as a single fuzzy IF-THEN rule and using the generalized modus ponens (6.10), we obtain the output of the fuzzy inference engine as seth if wo use the Mamdani combination, or as up tha) ae 0) (719) Het) Ds Beane (X40) (7.18) ey) if we use the Gédel combination, In summary, the computational procedure of the composition based inference is ssiven as follows:Fuzzy Rule Base and Fuzy Inference Engine Ch. 7 «Step 1: For the M fuzzy IF-THEN rules inthe form of (7.1), determine the membership fmetions p14, ¢...4, (By 2) for 1 = 2,2) uy AE according to (7.13). # Step 2: View Al x---x AL as the FP, and B! as the FPy in the impli- cations (6.28)(6.26), (6.81) and (5.82), and determine pigun #1y--y m0) = ate oxAivB1(ise-ymy) fOr 1 = 1,2, -05.M according to any one of these implications 1 Step 3: Determine poy (4) oF Ago (%,¥) acvording to (7.16) or (7.17). ‘Step 4: For given input 4’, the fuzzy inference engine gives output BY ac- cording to (7.18) or (7.19) 7.2.2 Individual-Rule Based Inference In individual-rule based inference, each rule in the fuzzy rule base determines an ‘output fuzzy set and the output of the whole fuzzy inference engine is the combina- tion of the M individual fuzzy sets. The combination can be taken either by union cor by intersection, ‘The computational procedure of the individual-rule based inference is summa- tized as follows: ‘© Steps 1 and 2: Same as the Steps 1 and 2 for the composition based inference, ‘* Step 3: For given input fuzzy set A’ in U, compute the output fuzzy set B} in V foreach individual rule Ru! according tothe generalized modus ponent (6.10), that is, og) = sup th (pawn (9 (720 ‘© Step 4: The output of the fuzzy inforence engine is the combination of the M fusay sets {Bi, ... By} either by union, that poy) = neyo) oe) (721) ‘or by intersection, that is, BW) = Hasla) + HE (9) (722) where ++ and + denote ssnorm and t-norm operators, respectively.Sec 1.2 Fuzzy Inference Engine or 7.2.3 The Details of Some Inference Engines From the previous two subsections we see that there are a variety of choices in the fuzzy inference engine. Specifically, we have the following alternatives: (i) ‘composition based inference or individual-rule based inference, and within the com- position based inference, Mamdant inference or Gadel inference, (il) Dienes-Rescher implication (6.23), Lukasiewicz implication (5.24), Zadeh implication (5,25), Godel implication (5.26), or Mamdant implications (5.31)-(5.82), and (ii) different oper- ations for the t-norms and s-norms in the various formulas, So a natural question is: how do we select from these alternatives? In general, the following threo eriteria should be considered: ‘* Intuitive appeat: The choice should make sense from an intuitive point of view. For example, if a set of rules are given by a human expert who believes that these rules are independent of each other, then they should be combined by ‘* Computational efficiency: ‘The choice should result in a formula relating BY with 4’, which is simple to compute. ‘+ Special properties: Some choice may result in an inference engine that has special properties. If these properties are desirable, then we should make this choice. ‘We now show the detailed formulas of a number of fuzzy inference engines that are commonly used in furzy systems and fuzzy control. + Product Inference Engine: In product inference engine, wo use: (i) individual- rule based inference with union combination (7:21), (8) Mamdani’s product implication (5.32), and (i) algebraic product fo all the t-norm operators and ‘maz for all the storm operators. Specifically, ftom (7.20), (7.21), (5.82), and (7-13), we obtain the product inference engine as sory) = as up (pact) [] ag dae) (7.23) ‘That is, given fuzzy set A‘ in U, the product inference engine gives the fuzzy ‘set B’ in V according to (7.23). + Minimum Inference Engine: In minimum inference engine, we use: (i) individual-rule based inference with union combination (7.21, (fi) Mamidani's ‘minimum implication (5.31), and (ii) min for all the t-norm operators and smaz fo all the s-norm operators. Specifically, from (7.20), (7-21), (8:31), and (7.13) we have (0) = miloup mina) say(@r)rontag a) ootw))) (2.24)98 Fray Rule Base and Fury inference Engine _Ch.7 ‘That is, given fuzzy set A’ in U, the minimum inference engine gives the fuzzy set BY in V according to (7.24). ‘The product inference engine and the minimum inference engine are the most commonly used fuzzy inference engines in fuzzy systems and fuzzy control. The ‘main advantage of them is their computational simplicity; this is especially true for the product inference engine (7.23). Also, they are intuitively appealing for many practical problems, especially for fuzzy control. We now show some properties of the product and minimum inference engines. Lemma 7.2, The product inference engine is unchanged if we replace “individual- rule based inference with union combination (7.21)" by “composition based infer- ence with Mamdani combination (7.15).” Proof: From (7.15) and (7.18) we have e(y) = gupta) xno )] (725) ‘Using (5.32) and (7.13), we can rewrite (7.25) as voto) = gop ase (3) Taye] (7.26) Because the max, and supxey are interchangeable, (7.26) is equivalent to (7.28). a ‘Lemma 7.3. If the fuzzy set A’ is a fuzzy singleton, that is, if Lifx=x* ‘where x" is some point in U, then the product inference engine is simplified to go(y) = msl] ] og (22) 0 WI] (7.28) and the minimum inference engine i simplified to ory) = Ph amin (29) ay (28) 4ne(W))] (7.29) Proof: Substituting (7.27) into (7.28) and (7.24), we see that the suPxcu is achieved at x = x", Hence, (7.23) reduces (7.28) and (7.24) reduces (7.29). ‘Lemma 7.2 shows that although the individual-rule based and composition based inferences are conceptually different, they produce the same fuzzy inference engine in certain important cases. Lemma 7.3 indicates that the computation within theSec. 2 Fuzzy Inference Engine 90 furzy inference engine can be greatly simplified ifthe input isa furry singleton (the 1most difficult computation in (7:28) and (7.24) is the super, which disappears in (7.28) and (7.29). {A disadvantage ofthe product and minimum inference engines is that if at some X EU the jis (2;)'s are very small, then the jg»(y) obtained from (7.28) or (7.24) ‘will be very sinall. ‘This may cause problems in implementation, The following three fuzzy inference engines overcome this disadvantage. ‘* Lukasiewicz Inference Engine: In Lukasiewice inference engine, we use: {@) individual-rule based inference with intersection combination (7.2), Lukasiewice implication (5.24), and (ii) min for all the t-norm operators. Specifically, from (7.22), (7.20), (5.24) and (7.13) we obtain sayy) = ppp mina) Has 0) {up inna (9), min 1 ~ gg) + WDD} ip sup minfa(x).1 mindy (a) + Horo (780) ‘That is, for given fuzzy set A’ in U, the Lukasiewicz inference engine gives the fuzzy set B' in V according to (7.30). ‘+ Zadeh Inference Engine: In Zadch inference engine, we use: (i) individual- rule based inference with intersection combination (7.22), (ji) Zadeh implica- tion (5.25), and (il) min forall the tmorm operators. Specifically, from (7.22), (7.20), (6.25), and (7.13) we obtain bay) rn M{ gp mina) ma (min (24) Hay (En) tas) nag) (731) '* Dienes-Rescher Inference Engine: In Dienes-Rescher inference engine, ‘we use the same operations as in the Zadeh inference engine, except that ‘we replace the Zadeh implication (5.25) with the Dienes-Rescher implication (6.23). Specifically, we obtain from (7.22), (7-20), (5.23), and (7.13) that too) = i gup miniua(s),maz(l— mining (ea) maca))) (082) Similar to Lemma 7.3, we have the following results for the Lukasiewicz, Zadeh and Dienes-Rescher inference engines,100 Fuzzy Rule Base and Funy inference Engine Ch 7 Lemma 7.4: If A’ isa fuzzy singleton as defined by (7.27), then the Lukasiewicz, ‘Zadeh and Dienes-Rescher inference engines are simplified to ‘bor (u) = mnt ~ nlin(ugg(27)) + Hosta] (7.33) {max|min p14; (2), 4tay (Ba) Har(O))> 1 mine 21) (7.34) Bl (maz{h — nin(ug (24), oi ]} (7.38) respectively Proof: Using the same arguments as in the proof of Lemma 7.3, we ean prove this lemma, 0 We now have proposed five fuzzy inference engines. Next, we compare them through two examples. Example 7.2: Suppose that a fuzzy rule base consists of only one rule TP 2; in Ay and -- ond ty is Any THEN yis BB (738) ed 1 l He? \ Hy 7 = igure 7.5. Output membership functions using the prod tet and minimum inference engines fr the 4g) © 0-5See. 7.2. Fuzzy Inference Engine 103 Hy. Hy Figure 7.6. Output membership functions using the Lasiewicr, Zadeh and Dience-Reecher inference engines for the ag (zp) <0. case Figure 7.7. Output membership function us the prod: ‘ut Inference engine forthe came of to rules,104 Furzy Rule Base and Fury inference Engine _Ch. 7 7.3. Summary and Further Readings In this chapter we have demonstrated the following: «© The structure of the canonical fuzzy IF-THEN rules and the criteria for eval- ating a set of rules ‘* The computational procedures for the composition based and individual-rule based inferences. ‘© The detailed formulas of the five specific fuzzy inference engines: product, ‘minimurn, Lukasiewicz, Zadeh and Dienes-Rescher inference engines, and their ‘computations for particular examples. Lee [1990] provided a very good survey on fuzay rule bases and fuzzy inference engines. ‘This paper gives intuitive analyses for various issues associated with fuzzy rule bases and furzy inference engines. A mathematical analysis of fuzzy infer- ence engines, similar to the approach in this chapter, were prepared by Driankov, Hellendoorn and Reinfrank (1993) 7.4 Exercises Exercise 7-1. If the third and sixth rules ia (7.12) (Example 7.1) are missing, at what points do the IF part propositions of all the remaining rules have zero membership values? Exercise 7.2. Give an example of fuzzy sels B",..., B® such thatthe set of the six rules (7.12) is continuous ~ Exercise 7.3. Suppose that a fuzzy rule base consists of only one rule (7.36) with po(v) = exn(-v") (75) Let the input AY to the fuzzy inference engine be the fuzzy singleton (7.27). Plot the output membership functions jp) using: (a) product, (b) minimum, (€) Lukasiewicz, (d) Zadeh, and (e) Dienes-Rescher inference engines Exercise 7-4. Consider Example 7.5 and plot the p(y) using: (a) Lukas inference engine, and (b) Zadeh inference engine. Bxercise 7.5, Use the Gadel implication to propose a so-called Gadel inference engineChapter 8 Fuzzifiers and Defuzzifiers We learned from Chapter 7 that the fuzzy inference engine combines the rules in the fuzzy rule base into a mapping from fuzzy set A in U to fuzzy set BY in V Because in most applications the input and output of the fuzzy system are real- valued numbers, we must construct interfaces betwoen the fuzzy inference engine and the environment. The interfaces are the fuzzifier and defuzzitier in Fig. 1.5. 8.1 Fuzzifiers ‘The fuztifir is defined as a mapping from a real-valued point x" € UC R" to a fuzzy set A’ in U. What are the criteria in designing the fuzzifier? First, the fuzzifier should consider the fact that the input is at the crisp point x", that is, the fuzzy set A’ should have large membership value at x". Second, if the input to the fuzzy system is corrupted by noise, then itis desireable that the fuzzifier should help. to suppress the noise. Third, the fuzzfier should help to simplify the computations involved in the fuzzy inference engine. From (7.23), (7.24) and (7-30)-(7.32) we see that the most complicated computation in the fuzzy inference engine isthe therefore our objective is to simplify the computations involving supxey We now propose three fuzzifiers ‘+ Singleton fuzzifier: The singleton fussifier maps a real-valued point x" € 0 into a fuzzy singleton A in U, which has membership value 1 at x" and 0 at all other points in Uthat wax) (tuzex oH O otherwise © Gaussian fuzzifier: The Gaussian fussifier maps x" € U into fuzey set A in U, which has the followings Gaussian membership function:106 Furze and Defuzzifirs Ch. 8 where a; ate positive parameters and the t-norm + is usually-chosen.as alge- braie product or min. # ‘Triangular fuzzifler: The triangular fuzsifier maps x" € U into fuzzy set A in U, which has the following triangular membership function rw ={ (1 EEG wooo = 5S) if Jes —2T] 1; therefore, the height of the 0h fuzay et in (7.90) is say = sup ming a(2) 1 ~ platings (20) + Hao) = sup nae(2) =1 (9.13)122 Fuzzy Systems 38 Nonlinear Mappings Ch. 9 ‘where we use the fact that for ll the three fuzifiers (8.1)-(8.3) we have supzgy Ha-(2) 1, Similarly, the height ofthe 'th fuzzy set in (7.32) is also equal to one, Hence, th the center average defuzifer (8.18) we obtain (9.12). 0 ‘The fuzzy systems in the form of (9.12) do not make a lot of sense because it sives a constant output no matter what the input is. Therefore, the combinations of fuzzy inference engine, fuzzfer, and defuzzfer in Lemma 9.3 do not result in useful fuzzy systems 9.1.2 Fuzzy Systems with Maximum Defuzzifier ‘Lemma 9.4. Suppose the fuzzy set BI in (7.1) is normal with center g!, then the fuzzy systems with fuzzy rule base (7.1), product inference engine (7.28), singleton fuzziier (8.1), and maximum defuzzifer (8.23) are of the following form: fe) =a" (os) where fe € (1,2, Mis such that TI ap (ed > [Tae (9.15) for all 1=1,2,..,M- Proof: From (7.28) (Lemma 7.3) and noticing that 2" in this case, we have sup aae(y) = sup afbe(TT oa (eda0 (w)) (0.16) Since supyey and max are interchangeable and BY is normal, we have sup ne-(y) = Hisoup TT ny (edo (oa7) where [+ is defined according to (9-15). Since wpi(y!*) <1 when I # Ie and bpie(gl*) = 1, we have noi") (0.8)see 1,_The Formulae of Some Clases of Fury Systems 123 Hence, the supyey in (9.16) is achioved at g!*. Using the maximum defuzifier (6.23) we obtain (9.14). © ‘From Lemma 9.4 we see thatthe fuzzy systems in this case are simple functions, that is, they are piece-wise constant function, and these constants are the centers of the membership functions in the THEN parts of the rules. From (9.15) we soe that as long as the product of membership values of the IF-part fuzzy sets of the rule is greater than or equal to those of the other rules, the ontput of the fuzzy system remains unchanged. Therefore, these kinds of fuzzy systems are robust to smnall disturbances in the input and in the membership functions j4(2i). However, these fuzzy systems are not continuous, that is, when I+ changes from one number to the other, f(2) changes in a discrete fashion. If che fuzzy systoms are used in decision making or other open-loop applications, this kind of abrupt change may be tolerated, but it is usually unacceptable in closed-loop contro ‘The next lemma shows that we can obtain a similar result if we use the minimam inference engine. Lemma 9.5. If we change the product inference engine in Lemma 9.4 to the ‘minimum inference engine (7-24) and keep the others unchanged, then the fuzzy systems are of the same form as (9.14) with t+ determined by (oa (@0] = mila 420) (0.19) where ! = 1,2, Proof: From (7.29) (Lemma 7.3) and using the facts that supyey and max} are interchangeable and that BY are normal, we have u sup p(y) = nelsup min( yy (21), 44, (2a) ae ()] vb ve = ldiin 4g (2) = mln ayo) om Also from (7.20) we have that jn (7) = ming; (iq)-(2))s thus the supyey in (9.20) is achieved at g. Hence, the maximum defuzrifir (8.23) gives (9.14). 0 Again, we obtain a clas of fuzzy systems that are simple functions Its diffcult to obtain closed-form formulas for furzy systems with maximum de- fuzzifer and Lukasiewicz, Zadeh, or Dienes-Rescher inference engines. The dificulty comes from the fact that the supyey- and min operators are not interchangeable in general, therefore, from (7.30)-(7-52) we see that the maximum defuzzification becomes ain optimization problem for & non-smooth function. Tn these cases, for a sven input 2, the output of the fuzay system has to be computed in a step-by-step fashion, that is, computing the outputs of furzfier, furry inference engine, and de- fazzfir in sequel. Note that the output ofthe fuzzy inference engine isa function,124 Fuzzy Systems a5 Nonlinear Mappings Ch. 9 not a single value, so the computation is very complex. We will not use this type of fuzzy systems (maximum defuzzifier with Lukasiewicz, Zadeh, or Dienes-rescher inference engine) in the rest of this book. 9.2. Fuzzy Systems As Universal Approximators In the last section we showed that certain types of fuzzy systems can be written as compact nonlinear formulas. On one hand, these compact formulas simplify the computation of the fuzzy systems; on the other hand, they give us a chance to ana- lyze the fuzzy systems in more details. We see that the fuzzy systems are particular types of nonlinear funetions, so no matter whether the fuzzy systems are used as controllers or decision makers or signal processors or any others, it is Interesting to know the capability of the fuzay systems from a function approximation point of view. For example, what types of nonlinear functions can the fuzzy systems rep- resent or approximate and to what degroe of accuracy? If the fuzzy systems can approximate only certain types of nonlinear functions to a limited degree of accu- racy, then the fuzzy systems would not be very useful in general applications. But if the fuzzy systems can approximate any nonlinear function to arbitrary accuracy, then they would be very useful in a wide variety of applications. In this section, we prove that certain classes of fuzzy systems that we studied in the last section have this universal approximation capability. Specifically, we have the following main theorem. ‘Theorem 9.1 (Universal Approximation Theorem). Suppose that the input universe of discourse U' is a compact set in R". Then, for any given real continuous function g(x) on U and arbitrary ¢ > 0, there exists a fuzzy system f(z) im the form of (9.6) such that sup If (2) ~ 9(2)1 <« (9.21) ‘That is, the fuzzy systems with product inforence engine, singleton fuzzifier, center average defuzzifier, and Gaussian membership functions are universal approxima- tors. One proof ofthis theorem is based on the following Stoue-Weierstrass Theorem, which is well known in analysis, Stone-Weierstrass Theorem (Rudin {1976]). Let Z be aset of real continuous functions on a compact set U. If (i) Z is an algebra, that is, the set Z is closed ‘under addition, multiplication, and scalar multiplication; (i) Z separates points on U, that is, for every z,y € Use # y, there exists f € Z such that f(x) # f(y); and (Gi) Z vanishes at no point of U, that is, for each x ¢ U there exists f € Z such that f(2) 4 0; then for any real continuous function g(2) on U and arbitrary ¢ > 0, there exists f € Z such that supzey [f(e) — g(z)| 0 has the property that f(x) > 0,¥ar € Hence. ¥ vanishes at no poi of Tn summary of the above and the Stone-Weierstrass Theorem, we obtain the conclusion af his theorem: © “Theorem 8.1 shows that fuzzy systems can approximate continuous factions to itary acaracy the following corlay extends the result to discrete functions, Corollary 0.1. For any squarointgrabe function gl2) on the eompact set UC RY that is, for any g € La(U) = {9 :U + Ry lg()/*de < co}, there exists funy system f(2) inthe form at (06) such that po (902) aera? < (030) Proof: Since U is compact, fy, dz = E < co. Since continuous functions on 1 form a dense subeet of Ea(U) (Rudin [1978 for aay 9 € Za(U) there exis a continuous function g on U' such that ({,,lg(2) ~ g(z)F'dz)"/? < e/2. By Theorem 9.1, there exists f € Y such that supyecs|f(#) ~g(2)| < ¢/(2B"). Hence, we have Of te) ~ ated” = (fg) ~ aera + Gf tale) — eee < (f (oup ste) ~ alee!” + e/2 (931) 0 ‘Theorem 9.1 and Corollary 9.1 provide a justification for using fuzzy systems in a variety of applications. Specifically, they show that for any kind of nonlinear operations the problem may require itis always possible to design a fuzay system that performs the required operation with any degree of accuracy. They also provide ‘theoretical explanation for the success of furzy systems in practical applications. However, Theorem 9.1 and Corollary 9.1 give only existence result; that is, they show that there erists a fuzzy system in the form of (9.6) that can approximate any function to arbitzary accuracy. They do not show how to find such a fuzzy system. For engineering applications, knowing the existence of an ideal fuzzy system is ‘not enough; we must develop approaches that ean find good fuzzy systems for theSec. 9.3. Summary and Further Readings 12 particular applications. Depending upon the information provided, we may or may not find the ideal fuzzy system. In the next few chapters, we will develop a variety of approaches to designing the fuzzy systems. 9.3 Summary and Further Readings In this chapter we have demonstrated the follow pt ing ‘© The compact formulas of some useful classes of fuzay systems ‘* How to derive compact formulas for any classos of fuzzy systems if such com- pact formulas exist. ‘¢ How to use the Stone-Weierstrass Theorem, ‘The derivations of the mathematical formulas of the fuzzy systems are new. A related reference is Wang, (1994). The Universal Approximation Theorem and its proof are taken from Wang [1902]. Other approaches to this problem can be found in Buckley (1992b] and Zeng and Singh (1994) 9.4 Exercises Exercise 9.1. Derive the compact formula forthe fuzzy systems with fuzzy rule base (7.1), Zadeh inference engine (71), singleton fuzzifer (81), and center average defurifer (8.18). Exercise 9.2. Repeat Exercise than Zadeh inference engine. Exercise 9.8. Show that the fuzzy systems in the form of 9.1) have the wni- versal approximation property in Theorem 9.1 Exercise 9.4. Can you use the Stone- Weierstrass Theorem to prove that fuzzy systems in the form of (9.7) oF (96) with of = 1 ate universal appraximators? Explain your answer Bxercise 9.8. Use the Stone Weierstrass Theorem to prove that polynomials ae universal apppeoximator Exercise 9.6, Plot the fuzzy systems fi(2) and fx) for 2 € U = [-1,2} x [41,2], where (2) is the fazy aytem with the two rules (8.27) and (8.28), product inference engine (7.28), singleton fuzziier (8.1), and maximum defuzrifier (8.23), and f(z) is the same as f,(z) except that the maximum defuzzifier is replaced by the center average defuzzter (8.18) 11 using Lukasiewicz inference engine ratherChapter 10 Approximation Properties of Fuzzy Systems | In Chapter 9 we proved that fazzy systems are universal approximators; that i, they ‘can approximate any fanction on a compact set to arbitrary accuracy. However, this result showed only the existence of the optimal fuzzy system and did not provide ‘methods to find it. In fact, finding the optimal fuzzy system is much more difficult than proving its existence. Depending upon the information provided, we may or may not find the optimal fuzzy system. : ‘To answer the question of how to find the optimal fuzzy system, we must first see what information is available for the nonlinear function g(z) : UC RY > R, which we are asked to approximate. Generally speaking, we may encounter the following three situations ‘© The analytic formula of g(2) is known, 's The analytic formula of (2) is unknown, but for any x € U we ean determine the corrspending g(2). That is, (z) isa black box—we know the input-output behavior of g(z) but do not know the details inside it * The analytic formula of g(2) is unknown and we are provided only a limited number of input-output pairs (2°, g(2')), where 2? € U cannot be arbitrarily chosen. The first case is not very interesting because if the analytic formula of g(2) is mown, we can use it for whatever purpose the fuzzy system tries to achieve. Tn the rare case where we want to replace g(z) by a fuzzy system, we can use the methods for the second case because the first case is a special case of the second ‘one. Therefore, we will not consider the first case separately: ‘The second case is more realistic. We will study it in detail in this and the following chapters. 128‘Sec. 10.1, Pratiminary Concepts 129 ‘The third case is the most general one in practice. This is especially true for fuzzy control because stability requirements for contol systems may prevent us from choosing the input values arbitrarily. We wil study this case in detail in Chapters 1215, So, in this chapter we assume that the analytic formula of g(x) is unknown but we can determine the input-output pairs (x;9(2)) for any 2 € U. Our task is to design a fuzzy system that can approximate g(z) in some optimal manner. We first introduce some concepts. Definition 10.1: Pseudo-Trapezoid Membership Function. Let [a,d] CR. The pseudo-irapezoid membership function of furzy set A is a continuous function in F. given by Ta), 2 [a,8) H, x6 [bye D(z), 2 € (esd) 0, rE R-(a,d) where a <6 <¢< da 0.130 Approximation Properties of Fuzzy Systems |__Ch. 10 Ha Ceisaba.ea) 4 ACxs-00-00.¢1.41) aCxtsbs,00,00) ch be a th by igure 10.1. Beamples of pseudo-rapeaoid membership, fonctions Definition 10.3: Consistency of Fuzzy Sets. Fuzzy sets A®,A®,.., AN in WC Rare said to be consistent on W if 14s(2) = 1 for some x € W implies that bae(a) = 0 for alli #5, Definition 10.4: High Set of Fuzzy Set. The high set of a fuaay set Ain W CR is a subset in W defined by hngh(A) = (2 € Whale) = sup wale')) (104) If Aisa normal fuzzy set with pseudo-trapezoid membership function 4(154,,¢,d), then hgh(A) = (6, Definition 10.5: Onder Between Fuzzy Sets. For two fuzzy sets A and B in W CR, wesay A> Bif hgh(A) > hgh(B) (that is, ifz € hgh(A) and 2’ € hgh(B), then x >2') We now show some properties of fuzzy sets with pseudo-trapezoid membership functions. Lemma 10.1. If 4}, 42,.... A% are consistent and normal fuzzy sets in W CR with pseudo-trapezoid membership functions 4.a(1; abi, Cid) (E = 1,2y-r9 Io ‘then there exists a rearrangement {i f2,..y8x} Of (1y2s0y.V} such that AN AR co cae (10.5) Proof: For arbitrary i, € {1y2s NJ, it must be true that [ci] [bj ¢4Sec. 102. sign of Fuzzy System 131 since otherwise the fuzzy sets A, ..., AY would not be consistent. Thus, there exists a rearrangement {fsfo, in} of (1,2, such that [bis seis] < [bigs€ia] <*> < [Bin sein] (10.6) which implies (10.5). 0 Lemma 10.1 shows that we can always assume AY < A? < +++ < AN without loss of generality. Lemma 10.2. Let the fuzzy sets A', A?,..., A” in W CR be normal, consistent ‘and complete with pseudo-trapezoid membership functions j14s(z;ai,s,cidi). TE Alc Abc 0 we can choose hy and fh small enough such that UE lochs + |B llaha < e Hlence from (10.11) we have supsey lg(2) ~ $31 = Ilo Slee

You might also like

• A Course in Fuzzy Systems and Control

  

  100% (9)

  A Course in Fuzzy Systems and Control

  441 pages
• A Course in Fuzzy Systems and Control

  

  100% (4)

  A Course in Fuzzy Systems and Control

  441 pages
• Fuzzy Systems and Control Guide

  

  No ratings yet

  Fuzzy Systems and Control Guide

  220 pages
• Introduction To Fuzzy Logic and Fuzzy Control

  

  No ratings yet

  Introduction To Fuzzy Logic and Fuzzy Control

  23 pages
• Fuzzystb Manual

  

  No ratings yet

  Fuzzystb Manual

  56 pages
• Fuzzy Systems & Control Guide

  

  No ratings yet

  Fuzzy Systems & Control Guide

  441 pages
• Course: In'Fuzzy

  

  No ratings yet

  Course: In'Fuzzy

  220 pages
• Fuzzy Logic Control

  

  No ratings yet

  Fuzzy Logic Control

  126 pages
• A Course in Fuzzy Systemsand Control

  

  No ratings yet

  A Course in Fuzzy Systemsand Control

  220 pages
• Fuzzy Logic Controller

  

  No ratings yet

  Fuzzy Logic Controller

  7 pages
• Jaekel03 Eolss Fuzzy Control Systems

  

  No ratings yet

  Jaekel03 Eolss Fuzzy Control Systems

  31 pages
• PyTorch - Advanced Deep Learning

  

  No ratings yet

  PyTorch - Advanced Deep Learning

  237 pages
• Essentials of Robust Control (Textbook)

  

  100% (2)

  Essentials of Robust Control (Textbook)

  425 pages
• D. Dubois, H. Prade (Auth.), M. J. Patyra, D. M. Mlynek (Eds.) - Fuzzy Logic - Implementation and Applications-Vieweg+Teubner Verlag (1996)

  

  No ratings yet

  D. Dubois, H. Prade (Auth.), M. J. Patyra, D. M. Mlynek (Eds.) - Fuzzy Logic - Implementation and Applications-Vieweg+Teubner Verlag (1996)

  325 pages
• Fuzzy Control Precise Control in A Fuzzy World

  

  No ratings yet

  Fuzzy Control Precise Control in A Fuzzy World

  46 pages
• System Identification: Arun K. Tangirala

  

  No ratings yet

  System Identification: Arun K. Tangirala

  13 pages
• Nonlinear Internal Model Control With Automotive Applications

  

  No ratings yet

  Nonlinear Internal Model Control With Automotive Applications

  243 pages
• Documents - Pub Programming in Ansi C by e Balagurusamy 3rd Edition Solutions

  

  No ratings yet

  Documents - Pub Programming in Ansi C by e Balagurusamy 3rd Edition Solutions

  228 pages
• Advanced Traffic Management Systems

  

  100% (1)

  Advanced Traffic Management Systems

  12 pages
• Adaptive Control Theory: Pole-Placement and Indirect STR

  

  100% (1)

  Adaptive Control Theory: Pole-Placement and Indirect STR

  48 pages
• Simulink Homework

  

  No ratings yet

  Simulink Homework

  66 pages
• Real-Time Optimization by Extremum-Seeking Control - Kartik B. Ariyur

  

  No ratings yet

  Real-Time Optimization by Extremum-Seeking Control - Kartik B. Ariyur

  231 pages
• Controls Engineering in FRC

  

  No ratings yet

  Controls Engineering in FRC

  238 pages
• Control Techniques

  

  No ratings yet

  Control Techniques

  340 pages
• Am 09 601 Vehicle Dynamics

  

  100% (3)

  Am 09 601 Vehicle Dynamics

  104 pages
• Fuzzy Logic Fundamentals: 4.1 Background

  

  No ratings yet

  Fuzzy Logic Fundamentals: 4.1 Background

  17 pages
• Controller Tuning With Ev

  

  No ratings yet

  Controller Tuning With Ev

  228 pages
• Next-Generation Sequencing Data Analysis 2nd Edition

  

  No ratings yet

  Next-Generation Sequencing Data Analysis 2nd Edition

  86 pages
• Implementacao Fuzzy Controller Matlab Mcds - Exercise - 8

  

  No ratings yet

  Implementacao Fuzzy Controller Matlab Mcds - Exercise - 8

  7 pages
• Control Theory for Engineers

  

  100% (1)

  Control Theory for Engineers

  163 pages
• Automatic Control of Air Conditioner Using FUZZY LOZIC

  

  No ratings yet

  Automatic Control of Air Conditioner Using FUZZY LOZIC

  6 pages
• 16 08 05 Fuzzy Set Fuzzy Logic

  

  No ratings yet

  16 08 05 Fuzzy Set Fuzzy Logic

  21 pages
• Final Exam in System Identification For F and STS Answers and Brief Solutions

  

  No ratings yet

  Final Exam in System Identification For F and STS Answers and Brief Solutions

  4 pages
• Intelligent Control & Fuzzy Logic

  

  No ratings yet

  Intelligent Control & Fuzzy Logic

  87 pages
• Data-Driven Aerospace Engineering With ML

  

  No ratings yet

  Data-Driven Aerospace Engineering With ML

  28 pages
• The Kalman Filter: State-Space Derivation For Mass-Spring-Damper System

  

  No ratings yet

  The Kalman Filter: State-Space Derivation For Mass-Spring-Damper System

  10 pages
• Fuzzy Logic Example

  

  No ratings yet

  Fuzzy Logic Example

  4 pages
• Fuzzy Logic: UIET (Computer Science Deptt)

  

  100% (1)

  Fuzzy Logic: UIET (Computer Science Deptt)

  56 pages
• Mean-Value Modelling and Robust Control of The Airpath of A Turbocharged Diesel Engine

  

  No ratings yet

  Mean-Value Modelling and Robust Control of The Airpath of A Turbocharged Diesel Engine

  168 pages
• Linear System

  

  No ratings yet

  Linear System

  68 pages
• Simulink Onramp

  

  No ratings yet

  Simulink Onramp

  1 page
• Data Acquisition: An Introduction: Bruxton Corporation

  

  100% (1)

  Data Acquisition: An Introduction: Bruxton Corporation

  7 pages
• Fuzzy & ANN Control Systems in Python

  

  No ratings yet

  Fuzzy & ANN Control Systems in Python

  20 pages
• Fuzzy Logic

  

  No ratings yet

  Fuzzy Logic

  95 pages
• LQR and LQG

  

  No ratings yet

  LQR and LQG

  13 pages
• Mech ch3

  

  No ratings yet

  Mech ch3

  7 pages
• Fronczak, Reniers - Model-Based Systems Engineering - 4TC00 Dictaat 2014-2015 PDF

  

  No ratings yet

  Fronczak, Reniers - Model-Based Systems Engineering - 4TC00 Dictaat 2014-2015 PDF

  105 pages
• HW 1 Sol

  

  0% (1)

  HW 1 Sol

  7 pages
• Control Systems Notes

  

  100% (1)

  Control Systems Notes

  66 pages
• Fuzzy Logic Applications: Bram Heyns

  

  No ratings yet

  Fuzzy Logic Applications: Bram Heyns

  7 pages
• Speed Control of BLDC Motor Using Fuzzy Logic Controller and Comparing It With Pid Controller - Ignited Minds Journals

  

  No ratings yet

  Speed Control of BLDC Motor Using Fuzzy Logic Controller and Comparing It With Pid Controller - Ignited Minds Journals

  15 pages
• DC Motor Fuzzy-PI Controller Design

  

  No ratings yet

  DC Motor Fuzzy-PI Controller Design

  5 pages
• Lect 1 - Intro To Automatic Control PDF

  

  No ratings yet

  Lect 1 - Intro To Automatic Control PDF

  15 pages
• Euler Lagrange Equation

  

  No ratings yet

  Euler Lagrange Equation

  14 pages
• Linear System Theory: 2.1 Discrete-Time Signals

  

  No ratings yet

  Linear System Theory: 2.1 Discrete-Time Signals

  31 pages
• Neural Network Sliding-Mode Position Controller For Induction Servo Drive

  

  No ratings yet

  Neural Network Sliding-Mode Position Controller For Induction Servo Drive

  12 pages
• Stability

  

  No ratings yet

  Stability

  16 pages
• ET410 Scope of The Capstone Project

  

  No ratings yet

  ET410 Scope of The Capstone Project

  5 pages