Particle swarm optimization (PSO) is one of the most recently developed evolutionary algorithms (EA). For PSO, the setting of the inertia weight is a key factor in determination of the performance of searching for the optimal solutions to the applications. The previously published algorithms cannot adapt their mechanisms of the parameter settings according to the landscape of some particular objective function corresponding to an optimization problem. To address this problem, we proposed a novel kind of robust PSO (RPSO) within a co-evolving framework by using the fuzzy logical controllers (FLCs) to control the inertia weights and encoding the fuzzy rules of the FLCs with cultural genes (CG) for the parameter settings. Compared RPSO with several kinds of adaptive PSOs, the simulation results on a suite of test functions show that the use of this framework improves the performance of the PSO to some extent, especially the robustness against the dimensional variation of the test functions.
Qiang Luo(#)(*),Dongyun Yi,A co-evolving framework for robust particle swarm optimization,Applied Mathematics and Computation,2008,199(2):611-622. (ESI Top 10%)
Download Matlab package for RPSO