A Dynamic Fuzzy Neural System Design via Hybridization of EM and PSO Algorithms

摘要

In this paper, we propose a modified hybridization of electromagnetism-like mechanism (EM) and particle swarm optimization (PSO) algorithms, called mEM PSO, for designing the proposed functional-link based Petri recurrent fuzzy neural system (FLPRFNS). The mEMPSO implements an instant update particle velocity strategy such that each particle updates its information instantaneously. For reducing the computational complexity, the randomly local search is replaced by PSO algorithm. In addition, the proposed FLPRFNS has the following characteristics, the consequent part is a functional-link based orthogonal basis functions and a Petri layer is adopted to eliminate the redundant fuzzy rules computation. In order to improve the ability of function approximation and have better convergence results, this study uses the functional expansion sine and cosine basis functions. Simulation on nonlinear control and nonlinear channel equalization are discussed to show the effectiveness and performance of our approach.