PID-compensated ANN Inverse Control with Application in Superheated Steam Temperature Control of Supercritical Boiler Unit

Considering the nonlinearity, large inertia and large time delay of superheated steam temperature system of the supercritical boiler unit, the nonlinear dynamic neural network inverse models of superheated steam temperature water-spray desuperheating system were established, and the historical operation data of the unit were used to train and verify the models. Based on the trained models, neural network inverse controllers with PID compensation links were constructed, and a real-time control program was programmed in MATLAB. With the full-scope simulation system of the 600MW supercritical unit, the superheated steam temperature control simulation tests were carried out under setpoint value disturbance and continuous variable load disturbances. The results show that the neural network inverse control scheme with PID compensation can effectively reduce the control deviation of superheated steam temperature and greatly shorten the process stabilization time.