Self-tuning PID control by neural-networks

Abstract

It has been proved that a multilayered neural network (NN) can approximate any continuous function within arbitrarily small error. Thus, NNs enable us to represent any nonlinear system. Furthermore, adaptive control theory or parameter tuning of PID controller is mainly limited in linear systems. In this paper, using a nonlinear mapping capability of NNs, we derive a tuning method of PID controller based on a backpropagation method of multilayered NNs. Simulated and experimental results show that the proposed method can identify the appropriate parameters of PID controller when it is implemented to both linear and nonlinear plants.