DE Rolling Optimal Control of Interference Prediction Based on GRU Model

Aiming at the output disturbance of nonlinear system caused by input interference, through traditional PID control, further reaction can only be carried out after the output disturbance, which leads to the inability to achieve more accurate and rapid regulation. However, due to the characteristics of the predictive model, model predictive control (MPC) can directly detect the interference after the interference is generated, and then more suitable input can be selected in the rolling optimization part to suppress the interference. In order to build a more accurate prediction model, the gated recurrent unit (GRU) is selected to build a prediction model for the nonlinear system, and the differential evolution algorithm (DE) is adopted in the rolling optimization part to improve the global optimization ability. It is tested in a strongly nonlinear system such as gas turbine, and compared with the traditional PID control by adding input interference. The simulation results show that the model predictive control combined with GRU and DE has better command response and interference prediction suppression effect.