A Particle Swarm Optimizer with Chaotic Self-Feedback for Global Optimization of Multimodal Functions

Abstract

This paper proposes an improved particle swarm optimization utilizing iterative chaotic map with infinite collapses (ICMIC) perturbations (ICMICPSO) for global optimization of multimodal functions. The chaotic perturbation generated by the ICMIC is incorporated into the particle's velocity updating rule to make the particles have a larger potential space to fly. With the coefficient of chaotic perturbation decaying, the dynamics of ICMICPSO algorithm is a chaotic dynamics first and then a steepest descent dynamics. The proposed ICMICPSO method as hybrid optimization is tested on several widely used multimodal functions. Numerical results are compared with that of some other chaotic PSO methods available in the usual literature. The performance studies demonstrate that the effectiveness and efficiency of the proposed ICMICPSO approach are comparably to or better than that of the other CPSO variants in this paper.