混沌粒子群优化的模糊神经PID控制器设计.pdfVIP

2008 2 ( ) Feb. 2008 35 1 J OUR NAL OF XIDIAN UNIV ER SITY Vol. 35 No. 1 PID 赵 俊, 陈建军 (西安电子科技大学 机电工程学院, 陕西 西安 710071) : 针对常规 PID 控制的线性局限性及传统模糊控制和模糊PID 控制中积分误差规则难以获 , 系 统存在稳态误差的问题, 提出一类以模糊神经网络和PID 神经网络组成的模糊神经PID 控制器. 以整 个 神经网络的权值为优化参数, 利用基于混沌策略的粒子群全局优化算法离线优化和误差反传算法在线 调整相结合的方法获得控制器参数, 并设计了混沌优化与粒子群结合的两步方案. 仿真结果表明: 与传 统PID模糊模糊 PID 控制相比, 系统的瞬态和稳态性能有了明显提高, 且保持了 一定的鲁棒性及高跟 踪精度. 该方法有效地拓展了PID 控制的使用范围, 并为智能方法与PID 控制的结合提供了 一种新的参 考方案. : 稳态误差; PID 神经网络; 模糊神经PID 控制器; 混沌优化; 粒子群全局优化 + :T P273 . 2 :A : 1001-2400( 2008) 01-0054-06 Design of the fuzzy neural PID controller based on hybrid PSO ZH A O J un, CH E N J ian-j un ( School of Mechano-electronic Engineering, Xidian Univ . , Xian 710071, China) Abstract: Considering the problems of the linearity lim it of PID control and the steady-stat e error in f zzy , f zzy PID control for it cannot easily obt ain the control r les of t he integral error, so a f zzy ne ral PID cont roller w hich consists of a f zzy ne ral network and a PID ne ral netw ork is designed. The paramet ers of the controller are optimized by the mixed learning methods integrating the offline particle sw arm opt imization algorit hm combined wit h chaos strategies of global searching ability , w ith the online BP algorithm of local searching ability . Sim lation res lts show that the designed novel controller and the proposed optimization algorithm have obvio sly improved the performance of the t ransient state and steady state in the control processing . Compared w ith convent ional PID, f zzy control and f zzy