MIMO系统的MPSO-PID型神经网络解耦控制研究

针对多输入多输出系统(MIMO系统)多变量、非线性、强耦合的特点,提出采用改进粒子群优化算法(MPSO)对PID型神经网络的权值进行优化的方法,实现对MIMO系统的解耦控制.其中,与基本PSO算法相比,MPSO算法后期仍能保持种群的多样性和较大的搜索空间；PID型神经网络是一种3层前向神经网络,网络各层神经元个数、连接方式、连接权值的初值都是按PID控制规律确定的.通过仿真分析,该方法有很好的控制品质:跟踪快、鲁棒性强、解耦效果好,为实际应用中强耦合系统控制方法的改进提供了理论依据.

Research on Decoupling Control by MPSO-PID Neural Network in MIMO System

Arming at multivariable,strong coupling,non-linear characteristic of multiple inputs multiple outputs system(MIMO systems),the design puts forward to realize decoupling control of MIMO system using modified particle swarm optimization(MPSO) to optimize weights of PID neural network.Compared with basic PSO algorithm,MPSO algorithm can still keep diversity of groups and greater search space at later stage.PID neural network is a feedfoward network including three layers,number of neurons,connection mode,initial values of connection weights of each layer are determined by the PID control laws.Through the simulation analysis,the method has better control quality:fast track,strong robustness,and decoupling effect is good.