非线性挠性结构的神经网络变结构控制

针对非线性挠性结构的模态截断 ,模型不精确与不确定性 ,以及非线性特性 ,采用了神经网络与变结构相结合的控制方法 ,针对线性化部分设计变结构控制器 ,把模型不确定性以及非线性部分作为干扰 ,用神经网络辨识切换函数的实际值与理想值之差 ,该差值反映了模型与实际系统的差别。该方法对摄动具有很强的鲁棒性 ,大大减小了系统在切换面上的抖动。针对挠性结构的模态不可测性 ,以及传感器的失误概率随其数量的增大而迅速增大的特性 ,采用了神经网络观测器观测系统状态。

Neural Network-Assisted Variable Structure Control for Nonlinear Flexible Structure

In this paper a neural network\|assisted variable structure method is presented for nonlinear flexible structure to deal with mode truncation,imprecision of model and uncertainty as well as nonlinearity.Variable structure controller is designed for linearized part,at the same time,the imprecision of model and nonlinear part are regarded as perturbance,which is identified by neural network.Neural networks identify the difference between actual value of switching function and ideal value,which implies the difference between model and actual system.This method is good at robustness and it decreases the chattering on switching surface.And,a neural network observer is adopted for immeasurable mode of flexible structure since the misplay probability of sensors increases quickly as its quantity.