Comparison Of Particle Swarm And Differential Evolution Optimization Algorithms Considering Various Benchmark Functions

Authors : Hasan Özcan

Pages : 899-905

Doi:10.2339/politeknik.369076

View : 50 | Download : 22

Publication Date : 2017-12-20

Article Type : Research Paper

Abstract :This study aims to compare Particle Swarm Optimization insert ignore into journalissuearticles values(PSO); and Differential Evolution insert ignore into journalissuearticles values(DE); methods for various input parameters. Both optimization methods show high performance in optimization of any physical system including simple and complex constraints and objectives. Average and standard values of both methods were evaluated by utilizing 8 benchmark functions and a graphical representation and comparison of corresponding methods was presented for 50x50 and 100x100 population sizes and dimensionalities. It is concluded that DE and PSO show the best fitness value for Sum of Different Powers benchmark function for both number of populations. Approach to the optimum is found to be faster through the PSO method. Both methods are flexible to be used for simple and complex engineering problems with high performances with ease of programming.
Keywords : Particle swarm, differential evolution, optimization, benchmark function

ORIGINAL ARTICLE URL