Date of Award

9-6-2018

Publication Type

Master Thesis

Degree Name

M.A.Sc.

Department

Electrical and Computer Engineering

Supervisor

Kwan, Hon

Rights

info:eu-repo/semantics/openAccess

Abstract

Harmony Search (HS) is an emerging metaheuristic algorithm inspired by the improvisation process of jazz musicians. In the HS algorithm, each musician (= decision variable) plays (= generates) a note (= a value) for finding the best harmony (= global optimum) all together. This algorithm has been employed to cope with numerous tasks in the past decade. In this thesis, HS algorithm has been applied to design digital filters of orders 24 and 48 as well as the parameters of neural network problems. Both multiobjective and single objective optimization techniques were applied to design FIR digital filters. 2-dimensional digital filters can be used for image processing and neural networks can be used for medical image diagnosis. Digital filter design using Harmony Search Algorithm can achieve results close to Parks McClellan Algorithm which shows that the algorithm is capable of solving complex engineering problems. Harmony Search is able to optimize the parameter values of feedforward network problems and fuzzy inference neural networks. The performance of a designed neural network was tested by introducing various noise levels at the testing inputs and the output of the neural networks with noise was compared to that without noise. It was observed that, even if noise is being introduced to the testing input there was not much difference in the output. Design results were obtained within a reasonable amount of time using Harmony Search Algorithm.

Share

COinS