REPORT

( GROUP 2 )
SUBJECT :ADVANCED ENGINEERING MATHEMATICS

TOPIC : Harmony Search for Structural Optimization

GUIDED BY : Dr. PRIYANKA HARJULE

BATCH : E6

MEMBERS :

SUHANI 2023UEE2014

YASHWANT 2023UEE2019

SUSHMIT 2023UEE2017

ADESH 2023UEE2018
HARMONY SEARCH ALGORITHM

 INTRODUCTION

The Harmony Search Algorithm (HSA) is a heuristic optimization algorithm

inspired by the process of musical improvisation. It was introduced by Geem,
Kim, and Loganathan in 2001 as a metaheuristic approach to solving
optimization problems. The fundamental idea behind HSA is to mimic the
process of musicians improvising melodies in a way that achieves the best
harmony. In the context of optimization, this translates into finding the optimal
or near-optimal solutions to complex problems.

In HSA, a solution to an optimization problem is represented as a "harmony" (a

vector in the decision space), and the algorithm iteratively improves the
harmony by searching for better solutions, guided by memory (the best
harmonies encountered so far).

 BASIC CONCEPTS
The process of HSA is inspired by the following real-world musical concepts:
Harmony Memory (HM): A memory that stores the best harmonies or solutions.
Harmony Memory Size (HMS): The number of harmonies stored in the
Harmony Memory.
Pitch Adjustment (PA): A process of fine-tuning a solution after it has been
selected.
Harmony Memory Consideration (HMCR): The probability that a new harmony
is selected from the memory.
Pitch Adjustment Rate (PAR): The probability that a selected harmony will
undergo pitch adjustment.
HSA operates on the principle of combining elements from the current best
solutions and exploring new areas of the search space. These operations allow
HSA to balance exploration (searching new areas) and exploitation (refining
known good solutions).
  STEPS INVOLVED
The algorithm proceeds through the following steps:
1. Initialize the Harmony Memory (HM):
Initialize a set of random solutions (harmonies) and store them in the
harmony memory.
2. Improvisation of New Harmony:
o For each component (decision variable) in a new solution:
o Memory consideration (HMCR): A random number is generated. If
it is less than HMCR, the component is chosen from the memory.
o Pitch adjustment (PAR): If the new solution is selected, a pitch
adjustment is applied with a certain probability (PAR).
o If no pitch adjustment occurs, a random value from the entire
search s pace is chosen.

3. Evaluation:
The fitness (or quality) of the newly generated harmony is evaluated
using an objective function.
4. Update Harmony Memory:
If the new harmony is better than the worst harmony in the memory, it
replaces the worst one.
5. Termination:
Repeat the improvisation and updating process until a stopping criterion
is met .
  HARMONY SEARCH PARAMETERS
Harmony Memory Size (HMS): The number of harmonies stored in the
memory. A larger memory allows for better exploration but increases
computational complexity.

Harmony Memory Consideration Rate (HMCR): A probability that determines

whether a component of the new harmony is taken from the harmony memory.
A higher HMCR encourages exploitation of existing solutions.

Pitch Adjustment Rate (PAR): A probability that determines whether the

selected harmony undergoes pitch adjustment. A higher PAR enhances the
refinement of solutions.

Maximum Iterations or Stopping Criterion: The maximum number of iterations

or a threshold for solution quality to terminate the search process.

 EXAMPLE
Let’s consider a simple example of using Harmony Search to solve a single-
variable optimization problem: finding the minimum of the quadratic function:

𝑓(𝑥)=𝑥2−10𝑥+25
This is a simple convex function with a global minimum at x=5 where 𝑓(5)=0

Step-by-Step Execution:
Initialize Harmony Memory (HM):

Assume we have a harmony memory size of 5, and we initialize it with random

solutions within the range [−10,10]. For example, the initial HM might look like
this:
HM={−5.3,1.2,8.6,−1.9,4.3}
Improvisation of New Harmony:

Memory Consideration (HMCR): Assume HMCR = 0.9. For each component

(decision variable), we choose a value either from the memory or randomly
from the search space.

For the first component, a random number is generated. If it is less than 0.9, we
pick from the memory. Let's say the first component of the new harmony is
selected from the memory, and it is 4.3.
Pitch Adjustment (PAR): Assume PAR = 0.3. If a value is chosen from memory,
we perform pitch adjustment with a 30% chance. If we decide to apply pitch
adjustment, the value is perturbed slightly. For example, if the selected harmony
is 4.3, it may adjust to 4.5 (a slight change within a certain range).

The second and third components are chosen in the same way, possibly with
different outcomes.

Evaluation:

Evaluate the fitness of the new harmony (solution). For example, if the new
solution is x=4.5, the fitness is f(4.5)=4.52 -10(4.5)+25=0.25

Update Harmony Memory:

If this new solution is better (has lower fitness) than the worst solution in the
memory, it replaces that solution. For example, if the worst solution in HM is
8.6, and the new solution is 4.5, we replace 8.6 with 4.5.

Termination:
The process continues until the stopping criterion is met, such as a fixed number
of iterations or the objective function reaching a satisfactory value.
  ADVANTAGES OF HARMONY SEARCH
Simplicity: The algorithm is simple to implement and does not require the
gradient of the objective function, making it suitable for non-differentiable and
complex functions.
Global Search Capability: It can effectively search large solution spaces and is
less likely to get trapped in local optima compared to traditional optimization
algorithms like gradient descent.
Flexibility: HSA can be applied to a wide range of optimization problems,
including continuous, discrete, combinatorial, and multi-objective optimization.

 APPLICATIONS OF HARMONY SEARCH

Engineering Design Problems: HSA has been successfully used to optimize
designs in mechanical, civil, and electrical engineering fields.
Machine Learning: HSA is used in hyperparameter optimization, feature
selection, and model training tasks.
Control Systems: It is applied in tuning parameters for controllers, such as PID
controllers.
Scheduling and Routing Problems: HSA has been applied to optimize tasks such
as scheduling, traveling salesman problems, and supply chain management.

 CONCLUSION
The Harmony Search Algorithm is a versatile and effective optimization tool
that can solve complex real-world problems without requiring gradient
information or problem-specific knowledge. Its intuitive approach, inspired by
music improvisation, allows it to balance between exploration and exploitation,
making it suitable for both small and large-scale optimization problems. While
it is generally successful across a wide variety of applications, careful tuning of
the algorithm's parameters, such as Harmony Memory Size (HMS), Harmony
Memory Consideration Rate (HMCR), and Pitch Adjustment Rate (PAR), is
necessary for optimal performance.
BIBLIOGRAPHY :
 Engineering Optimization-By SS RAO
 https://www.researchgate.net/publication/
259216728_Harmony_Search_Algorithm
 www.openai.com
 https://en.wikipedia.org/?title=Harmony_search&redirect=no
 https://arxiv.org/pdf/1003.1599