On a method of single neural PID feedback compensation control

Abstract

A control structure based on PID controller, single neural PID controller and single neural PID identifier is proposed. The PID controller is used to maintain the stability in the early stage of the study process of the neural network as well as when the system is under disturbance. The single neural PID identifier performs online learning based on the control error. Then it transfers the parameter results to the single neural PID controller, successfully avoiding offline learning. Afterwards, the single neural PID controller performs further study based on the control parameters and the output of the PID controller, producing a feedback compensation control quantity in order to compensate the model error of the single neural PID identifiers. The simulation results shows that compared with traditional PID control method, the single neural PID feedback compensation control method obtains significant improvement in various control features and has relatively excellent robustness and static features.