Adaptive control systems based on neural networks

摘要

The paper advances the using of the neural networks (NN's) in adaptive control systems. Adaptive control solves the problem of the sensitivity to variation of the plant parameters. In the case of neural adaptive control, the controller parameters are changed by a NN trained off-line. The training patterns are obtained using any design method of the controller for many different values of the plant parameters. A useful tool to train any neural adaptive controller has been developed-a propgram for Windows'95 that implement the backpropagation algorithm in a general manner.