基于神经网络的一类非线性关联大系统鲁棒滑模分散控制

针对一类具有未知界扰动和子系统部分已知的非线性大系统,结合神经网络逼近方法、滑模控制研究了一种新的分散鲁棒自适应控制方法。所设计的分散控制器分为两部分,一是等效控制器,二是滑模控制器。滑模控制器用来减小系统的跟踪误差,起鲁棒控制作用。文中用神经网络逼近非线性未知函数,将网络权值误差引入到网络权值的自适应律中用以改善系统的动态性能。仿真算例证明了所设计的鲁棒分散控制器是有效的。

ROBUST SLIDING MODE DECENTRALIZED CONTROL FOR A CLASS OF NONLINEAR INTERCONNECTED LARGE-SCALE SYSTEM WITH NEURAL NETWORKS

A new decentralized robust control method is discussed for a class of nonlinear interconnected large-scale system with unknown bounded disturbance and unknown nonlinear function term. A decentralized control law is proposed which combines the approximation method of neural network with sliding mode control. The decentralized controller consists of an equivalent controller and an adaptive sliding mode controller. The sliding mode controller is a robust controller used to reduce the track error of the control system. The neural networks are used to approximate the unknown nonlinear functions, meanwhile the approximation errors of the neural networks are applied to the weight value updated law to improve performance of the system. Finally, an example demonstrates the availability of the decentralized control method.