Generalized PID neural network, which is based on classical PID control,realized through neural network which has self study function for multivariatedecoupling control. The structure,forward algorithm and back coupling algorithm weregiven Through simulating double inputs the double output system, generalized PIDneural network compared with network setting PID control algorithm, simple model of300MW unit plant coordination control system has carried on the simulation.It provedthat Generalized PID neural network has preferably self study and self adaptingdecoupling control ability through simulation. The system which inosculates thedecoupling and controller, is easy to implement and applicable for multivariatedecoupling control. It makes the decoupled system have better dynamic behavior andstatic characteristic. Especially, it makes parameters astringe fast when determininginitial value of network according to PID control law. |