Neural Network PID Control System for the FTGS

This paper proposes the neural network PID control system for the nonlinear Francis hydroturbine generator set FTGS including the Francis turbine neural network model FTNNM. The neural network model FTNNM reflect nonlinear characteristic of the Francis turbine truly. In the neural network PID control system, the regular PID controller receives the PID parameters from the neural network NNPID in different mode of operation and gives the optimal control signal in time. The Levenberg-Marquardt algorithm is used to train the FTNNM and the NNPID . The convergence speed of the offline training is fast and the accuracy of the model is high. Due to the simple and fast continuous variable parameter control algorithm, It satisfy the high real-time need of the hyroturbine governing system.