线性时变系统的分散自校正控制

Decentralized Self-Tuning Control of Linear Time-Varying System

本文对由存在有界扰动和非线性关联的子系统组成的线性时变离散大系统提出了一种多项式逼近的分散自校正控制算法．也可用神经网络去实现自校正控制，采用神经网络作为预报和控制器，网络学习算法用Widrow－Hoff规则．证明了这种分散自校正控制系统的稳定性．仿真结果表明系统是稳定的，可以很好地跟踪期望输出．

In this paper,a decentralized self-tuning control algorithm using polynomial approximation is presented for linear time-varying discrete large-scale systems consiting of subsystems with bounded disturbances and nonlinear interconnections. We also propose using neural networks to realize the decentralized selftuning control. The neural networks are employed to predict the plant model and use as a controller. The Widrow-Hoff rule is used as learning algorithm for the networks. The simulation results demonstrate that the system is stable and can track the desired output quite well.