基于神经网络的通用非线性神经自适应控制研究

Universal Nonlinear Adaptive Neural Control Based on Neural Network

该文提出了一个适用于非线性不确定系统的通用的自适应神经网络控制方案,其中只有一个神经网络同时作为控制器和识别器,控制器采用广义的反向传播学习算法(GBP)作为控制器学习算法,实现了学习误差与控制误差的统一,控制器的学习过程就是控制规律控制更新算法的更新过程,不需要额外的控制算法。该控制方案算法简单,控制速度快,实时性好,是一种高效的非线性神经网络控制方案。

In this paper, one for uncertain nonlinear systems adaptive neural network control program common, of which only a neural network controller and at the same time as the identifier, the controller uses the generalized backpropagation learning algorithm(GBP) as controllers learning algorithm to achieve a unified learning error and control error learning process controller is to control law to control update algorithm update process, no additional control algorithms. The control scheme is simple algorithm, control speed, real-time, is a highly non-linear neural network control scheme.