Fuzzy Logic
Prepared By
Artificial Intelligence
Biprodip Pal
CSE 609
Lecturer
Dept. of CSE
RUET
Fuzzy Logic
One way to represent inexact data and knowledge, closer
to humanlike thinking, is to use fuzzy rules instead of
exact rules when representing knowledge.
Fuzzy rules represent in a straightforward way knowledge
that is subjective, ambiguous, vague, or contradictory
Applications
Control of automatic washing machines
Automatic camera focusing
Control of transmission systems in new models of cars
Automatic landing systems for aircraft
Automatic air-conditioning systems
Cooperative robots
…
Main components of FLS
Fuzzy input and output variables, defined by their fuzzy
values
A set of fuzzy rules
Fuzzy inference mechanism
Input : exact / crisp (e.g., 7) /fuzzy values (e.g.," moderate").
Output : fuzzy / exact (crisp).
Fuzzy Sets
In fuzzy sets an object can belong to a set partially. The
degree of membership is defined through membership
function:
μA(u): U [0,1]
where U is called the universe, and A is a fuzzy subset of U.
Fuzzy Sets
Table: Degree of membership of ‘tall men’
Fuzzy Sets
Fig.: (a) Crisp set (b) Fuzzy Set for Tall men
Crisp and Fuzzy Subset Representation
A Fuzzy Logic System.
Algorithm : Fuzzy logic algorithm
1. Define the linguistic variables and terms (initialization)
2. Construct the membership functions (initialization)
3. Construct the rule base (initialization)
4. Convert crisp input data to fuzzy values
using the membership functions (fuzzification)
5. Evaluate the rules in the rule base (inference)
6. Combine the results of each rule (inference)
7. Convert the output data to non-fuzzy values
(defuzzification)
A Simple FLS to Control an Air Conditioner.
Linguistic Variables
Linguistic variables are the input or output variables of
the system whose values are words or sentences from a
natural language, instead of numerical values.
temperature (t) : represents the temperature of a room.
Membership Functions
A membership function is used to quantify a linguistic
term.
Used in the fuzzification (steps map the non-fuzzy input
values to fuzzy linguistic terms)
Also used in defuzzification (vice versa)
Membership Functions (Cont.)
Membership functions (Triangular)
Membership functions (Triangular)
Membership functions (Trapezoidal)
Membership functions (Trapezoidal)
Membership functions (Gaussian)
Membership functions (Bell)
Membership functions (Sigmoid)
Fuzzy Rules
Simple IF-THEN rule
Constructed to control the output variable
Fuzzy Rules (Cont…)
Fuzzy Set Operations
MAX Max {μA(x), μB(x)}
MIN Min{μA(x), μB(x)}
NOT (complement) 1-μA(x)
Inference Procedure
Mamdani-style inference
Sugeno-style inference
…
Example Problem
Estimate the level of risk involved in a software
engineering project based on two inputs: project
funding and project staffing.
Inference Example
X: If project funding is adequate than I shall take the risk of
doing it.
Y: Project funding is inadequate, so risk is high, someone
else will do it.
X:Than if project staffing is small than I can proceed too.
Y: As manager says project funding will be around 35% and
project staffing will be around 60% so what to do?
Inference Example
Input
project_funding = 35%
and
project_staffing = 60%.
Inference Example
Linguistic Variables
project_funding project_staffing risk
inadequate small low
marginal Large normal
adequate high
Inference Example
Company Uses values like 35%, 50% etc.
We are comfortable with linguistics.
project_funding project_staffing risk
Inadequate (x≤40) Small(…) Low
Marginal(20 ≤x≤ 80) Large(…) normal
Adequate(x ≥ 80) high
Inference Example
Fuzzy Membership Value
Project_funding Project_staffing
Inference Example
Calculation
Let for inadequate region(trapezoidal)
c=30
d=40
So, µ(35) = (x-d)/(c-d) =(35-40)/(30-40)=5/10=.5
inadequate
…
Inference Example
Conversion from crisp input to fuzzy value according to
membership
µ(35)inadequate =.5
project_funding = 35%
µ(35)marginal =.25
µ(35)adequate =0
Inference Example
Conversion from crisp input to fuzzy value according to
membership
µ(60)small =.1
project_staffing = 60%
µ(60)large =.7
Inference Example
Rules
If project_funding is adequate or project_staffing is small
then risk is low.
If project_funding is marginal and project_staffing is large
then risk is normal.
If project_funding is inadequate then risk is high.
Inference Example
Rule Evaluation:
Rule 1 - If project_funding is adequate or project_staffing is small
then risk is low
µrisk=nlow =max( µ(35)adequate , µ(60)small )
=max(0 , .1)
=.1
Rule 2 - If project_funding is marginal and project_staffing is large
then risk is normal
µrisk=normal =max(
.25 µ(35)marginal , µ(60)large )
=max(.25 , .7)
=.7
Rule 3 - If project_funding is inadequate then risk is high
µrisk=high =.5
Inference Example
Final result : Union on all of the scaled functions
Inference Example
Inference Example
Defuzzification
Centroid method:
COG=
Inference Example
Defuzzification
COG =
(0+10 +20)*.1+(30 + 40 +50 +60)*.2 +(70 +80 +90 +100)*.5
.1* 3+.2*4 +.5*4
=67.4
Inference Example
The result is that this project has 67.4% risk associated
with it given the definitions above.
***
