摘要： 为辨识非线性系统Hammerstein模型，将非线性系统的辨识问题转化为参数空间上的优化问题，提出一种基于正态云模型的改进粒子群算法(NCPSO)。该算法采用动态变异概率，对全局最优粒子和粒子自身最优位置进行正态云变异，以产生新的粒子引导种群的飞行，有效避免早熟收敛。采用一种广义学习策略，提升粒子向最优解飞行的概率，将NCPSO算法用于对Hammerstein模型的辨识，相比其他算法，该算法辨识精度较高。

关键词: 粒子群优化算法, 正态云模型, 系统辨识, 动态变异, Hammerstein模型

Abstract: In order to identify the nonlinear system Hammerstein model, the problem of nonlinear system identification is changed into an optimization problem in parameter space. An improved Particle Swarm Optimization based on Normal Cloud(NCPSO) is proposed, in which a dynamic mutation is adopted to make the normal cloud mutation for the best performing particle(Gbest) in the swarm and the best previous position of each particle(Pbest). This mutation will generate the new Pbest and Gbest to lead the whole swarm flight, which effectively avoids the premature convergence. Then a comprehensive learning strategy is introduced to increase the probability of flying to the optimal solution. NCPSO is used to identify Hammerstein model. Experimental results show that the proposed NCPSO has better results than other algorithms.

Key words: Particle Swarm Optimization(PSO) algorithm, normal cloud model, system identification, dynamic mutation, Hammerstein model

中图分类号: 

• TP301.6