A simple evolutionary algorithm for IIR filter design: differential evolution algorithm

Abstract

ÖzetçeHerhangi bir sayısal işaret işleme (DSP) algoritması veya işlemcisi, sayısal süzgeç olarak tanımlanabilir. Sonsuz darbe cevaplı (IIR) süzgeçlerin temel avantajı, aynı dereceden sonlu darbe cevaplı (FIR) süzgeçlere göre çok daha iyi performanslara sahip olmalarıdır. Bununla birlikte, bu süzgeçlerin hata yüzeyleri çok modlu bir yapıya sahip olabilmektedir. Diferansiyel gelişim (DE) algoritması üç temel avantaja sahip yeni bir sezgisel yaklaşımdır. Bu avantajlar : çok modlu bir araştırma uzayındaki evrensel minimayı bulabilme kabiliyetleri, yerel yakınsama hızlarının çok yüksek olması ve az sayıda kontrol parametresi kullanmasıdır. Bu çalışmada, DE algoritması IIR süzgeç tasarımına uygulanmış ve genetik algoritma ile performansları mukayese edilmiştir. AbstractAny digital signal processing algorithm or processor can be reasonably described as a digital filter. The main advantage of an infinite impulse response filter is that it can provide a much better performance than the finite impulse response filter having the same number of coefficients. However, they might have a multi-modal error surface. Differential Evolution (DE) algorithm is a new heuristic approach mainly having three advantages; finding the true global minimum of a multi modal search space, fast convergence, and using a few control parameters. In this work, DE algorithm has been applied to the design of digital IIR filters and compared its performance to that of genetic algorithm.

A Heuristic Method: Differential Evolution for Harmonic Reduction in Multilevel Inverter System

“…DE algorithm is a population based algorithm like genetic algorithms using similar operators; crossover, mutation and selection [9]. In this work, we have compared the performance of DE algorithm to that of some other well known versions of genetic algorithms.…”

Performance Study of Differential Evolution with Various Mutation Strategies Applied to Water Distribution System Optimization

“…The dither mutation and dither crossover factor introduced by Karaboga and Okdem (2004) was randomized according to:…”

“…Liu and Lampinen (2005) claimed that DE, with a population size within the range of [3D, 10D] (D is the dimension of the optimization problem), F within the range of [0.5, 1.0], and CR within the range of [0.8, 1.0], generally showed good performance when applied to numerical optimization problems. Karaboga and Okdem (2004) introduced a DE with a dither mutation weighting factor (F dither ) and a dither crossover rate (CR dither ) such that the values of mutation weighting factor and crossover rate were randomized in a given range for each individual rather than specified particular values. Within this work, Karaboga and Okdem (2004) addressed that the dither DE outperformed the standard DE with a given fixed F and CR based on testing a number of numerical optimization problems.…”

“…Karaboga and Okdem (2004) introduced a DE with a dither mutation weighting factor (F dither ) and a dither crossover rate (CR dither ) such that the values of mutation weighting factor and crossover rate were randomized in a given range for each individual rather than specified particular values. Within this work, Karaboga and Okdem (2004) addressed that the dither DE outperformed the standard DE with a given fixed F and CR based on testing a number of numerical optimization problems. A total of five mutation strategies have been frequently used for DE application to diverse optimization fields (Price et al 2005).…”

Differential evolution using a localized Cauchy mutation operator

“…DE shares a common terminology of selection, crossover and mutation operators with GA however it is the application of these operators that make DE different from GA; while, in GA crossover plays a significant role, it is the mutation operator which affects the working of DE [12].…”