The research of PID neural network decoupling controller and its application in unit load system

Based on back-propagation (BP) arithmetic, a neural network PID decoupling control strategy is presented for multi-input and multi-output (MIMO) system, which is applied to thermal power unit load system. The idea is the parameter such as proportion, integration and differentiation of PID controller are tuned on-line through self-learning of a network based on a certain control law. And the corresponding control law is taken across classical PID control algorithm. Then decoupling control is carried out in the process. Simulation results show that the almost dynamic decoupling and completely static decoupling are obtained, the closed loop system has zero static error, and the advantages of higher speed response and stronger robustness are developed.