基于高斯基神经网络的自学习控制及应用

Self-Learning Control Using Gaussian Potential Function Neural Networks and Its Application

本文提出一类基于高斯基神经网络的自学习控制器，该控制器由两个GPFN网络组成，一个完成PID学习控制，另一个完成未知被控对象模型的建模．为加快网络的学习过程，文中提出了递归最小二乘法（RLS）用于神经网络的学习，并分析研究了自学习控制系统的收敛性和稳定性．仿真和实验结果表明，这类智能控制是成功的．

This paper presents a new self-learning controller based on Gaussian potential function neural networks. The controller consists of a control network and model network. In order to speed up learning rate, a general least squares approach for neural network learning is proposed. Convergence and stability of this self-learning system is proved. Simulation results have shown that the new controller can be successfully applied in nonliner control systems.