Comparisons study of APSO OLPSO and CLPSO on CEC2005 and CEC2014 test suits

Abstract

Particle swarm optimization (PSO) is originally designed to solve continuous optimization problems. Recently, lots of improved PSO variants with different features have been proposed, such as Adaptive particle swarm optimization (APSO), Orthogonal Learning particle swarm optimization (OLPSO) and Comprehensive Learning particle swarm optimization (CLPSO). In order to find out whether these PSOs have any particular difficulties or preference and whether one of them would outperform the others on a majority of the tested problems, we analyze the performance of different PSOs on various tested problems. In this paper, we evaluate the performance of APSO, OLPSO, and CLPSO on more complex benchmark functions. The comparison is performed on a large amount of real-parameter optimization problems, including the CEC 2005 and the CEC 2014 benchmark functions. Finally, we find out that the OLPSO achieves higher solution quality than the other two PSOs on most problems based on the simulation results on benchmark functions.