基于带近邻因子的粒子群算法的非线性系统辨识

Nonlinear System Identification Based on Particle Swarm Optimization with Near Neighborhood Factor

针对非线性系统辨识的问题,提出了一种改进的粒子群算法。该算法引入近邻因子,增加了当前粒子的社会学习功能,可有效克服基本粒子群算法易陷于局部最优解的常见弊病。算法对未知非线性系统具有充分的逼近能力,对噪声不敏感,实现了对一类非线性系统的有效辨识。

A modified particle swarm optimization algorithm for nonlinear system identification is presented. By using a near neighborhood factor, each particle is attracted towards the best previous positions visited by its neighbors. The proposed algorithm emphasizes the social learning of particles, so it can effectively overcome the shortcoming of getting into local optimum by the classical algorithm. The nonlinear System identification and the related experiment analysis based on the modified particle swarm optimization algorithm presented show the good performance.