The combined effect comprehensive learning particle swarm optimizer

Abstract

This paper introduces a novel and efficient optimization method, the Combined Effect Comprehensive Learning Particle Swarm Optimizer (CECLPSO) to handle the problems of premature and slow convergence with inferior solution prevailing in PSO and its variants. These weaknesses are resolved by introducing the combined effect of two consecutive global best particles contribution on the learning strategies of particles with the integration of Comprehensive Learning. This is in contrast to the original Comprehensive Learning PSO (CLPSO) technique, in which, the particles learning strategy is based on the knowledge of only one global best gbest. The performance of the CECLPSO is compared with basic PSO (BPSO) and CLPSO algorithms, on search efficiency, with the set of benchmark functions of dimension 50. The simulation result clearly indicates that the proposed CECLPSO algorithm prevents premature convergence and obtains better solution over basic PSO and CLCPSO in optimizing higher dimensional multimodal functions.