Application of Genetic Algorithm to Digital FIR Filter Design

• Main Author: 王鴻獻
• Other Authors: 貝蘇章
• Format: Others
• Language: en_US
• Published: 2001
• Online Access: http://ndltd.ncl.edu.tw/handle/05959904029128994724

碩士 === 國立臺灣大學 === 電信工程學研究所 === 89 === Genetic algorithm(G.A.) has been widely applied to different areas recently due to its simplicity and efficiency. Being an optimization algorithm, it was applied to digital signal processing in this dissertation. This dissertation is consisted of seven chapters. In chapter 1, we simply introduce genetic algorithm. A numerical maximization problem is taken as an example to demonstrate how genetic algorithm actually works. In chapter 2, we use genetic algorithm to design higher-order differentiators. We consider two cases for the coefficients of the FIR differentiators. The first one is the infinite precision case, and the second is the finite precision case. For the second case, we set the coefficients in the form of sums or differences of powers-of-two(SDPOT). In chapter 3, we apply genetic algorithm to the design of FIR bilevel Laplacian-of-Guassian(BLoG) filter. Compared with gradient-descent algorithm, better results can be obtained. In chapter 4, we apply genetic algorithm to the design of FIR Quadrature-mirror-filter(QMF) bank. With G.A., the complex nonlinear optimization problem could be simply solved. In chapter 5, we introduce the application of G.A. to the synthesis of antenna arrays. Antenna arrays synthesis is somewhat similar to the design of FIR filters. In chapter 6, we apply genetic algorithm to polygonal approximation problems. Finally, we conclude this dissertation and suggest our future work in chapter 7.