Neural network based online self-learning adaptive PID control

An improved backpropagation neural network (BPNN) based online self-learning adaptive PID control strategy is presented. A feedback PID controller tuned by the critical proportion method works at the beginning of the control process. With the learning of a BPNN controller, it gradually compensates the deficiency of the feedback together perfectly. The design of the controller is independent on the empirical knowledge of the system, and the parameters are tuned based on the testing information and error feedback learning algorithm. The results of the simulation show that the structure of this method is simple and its performance is substantially outperform the normal PID controllers