基于变学习率CMAC网络的自适应逆控制研究

提出一种改进学习算法的CAMC网络结构，并应用于非线性系统控制。该算法可保证网络的学习率随着系统工作点的变化而自适应变化，加快了网络的收敛速度，提高了系统的自适应能力。文中分析了CAMC网络用于自适应逆控制过程中，网络学习率对网络收敛特性的影响，论证了自适应学习率在网络学习中的作用，并给出了学习率自适应学习的具体训练方法。最终将该方法应用于三阶机械手模型的逆运动控制，给出了基于普通CMAC的逆运动控制的控制曲线和基于改进学习算法后的CMAC的逆运动控制的控制曲线，并给出了分析和对比，论证了改进的学习算法的优越性。

The Adaptive Reversion Control of Nonlinear System Based on Improved CMAC Network

A new construction of CMAC based on an improved learning algorithm is brought forward, which is used for nonlinear system control. The algorithm changes the learning rate according to different work state of the system, and improves the adaptability of the system. In the paper, the effect on the learning character of the network when changing the learning rate of the CMAC is analyzed. The effect of the adaptive best - fitting learning rate learning method is proved, and the detailed course of this method is presented. At last, the method is used for the reverse kinematic control of the manipulator, and the control results based on normal CMAC and the improved learning method are obtained, and the comparison and analysis are presented.