非线性多变量零阶接近有界系统的多模型自适应控制

Multiple Model Adaptive Control for a Class of Nonlinear Multi-variable Systems with Zero-order Proximity Boundedness

针对一类多变量非线性离散时间系统,提出一种新的基于神经网络的多模型自适应控制方法.为了将非线性系统的高阶非线性项的限制条件放宽到零阶接近有界,该方法引入了一种新的非线性模型.该模型在传统线性回归模型基础上增加了非线性补偿项,使模型的估计误差有界.一个神经网络模型与非线性模型同时被用来对系统进行辨识.基于性能指标的切换机构选择性能较好的模型对应的控制器对系统进行控制.理论分析证明了零阶接近有界多模型自适应控制系统的有界输入和有界输出稳定性.仿真实验说明了提出的多模型自适应控制方法的有效性.

A novel multiple model adaptive control method using neural networks is proposed for a class of MIMO nonlinear discrete-time systems. In order to relax the restriction of the higher order nonlinear term of the nonlinear system to zeroorder proximity boundedness, this method introduces a new nonlinear model. The model adds a nonlinear compensation term to the conventional linear autoregressive model such that the estimation error is bounded. A neural network model is used to identify the system with nonlinear model simultaneously. A performance-based switching mechanism determines the controller which has the better performance to control the system. Theoretic analysis proves the bounded-input-boundedoutput stability of the zero-order proximity boundedness multiple model adaptive control system. Simulation results are presented to show the effectiveness of the proposed method.