Modified Comprehensive Learning Particle Swarm Optimization for Numerical and Takagi-Sugeno Fuzzy System Modeling

Abstract

Modifications for comprehensive learning particle swarm optimization (M-CLPSO) is proposed for numerical problems and modeling Takagi-Sugeno(T-S) Fuzzy System. A self-adaptive strategy is adopted to adjust the value of acceleration coefficient dynamically. In the late stage of the evolution, Gaussian disturbance is hydride with algorithm to help the stagnant particles to escape from standstill state. The effectiveness of the proposed algorithm is verified by numerical experiments and T-S fuzzy system modeling. The experiments results show that the proposed method can provide comparable improvement in solving function optimization problems and can generate good fuzzy system model with high accuracy.