HK-FDSA based Performance Optimization Strategy for Steam Generator Level Control System of Nuclear Power Plant

A steam generator (SG) is an essential piece of equipment between a nuclear power plant's primary and secondary loop. Hence, the steam generator level control system (SGLCS) is of great significance to nuclear power plants safe and economical operation. The SGLCS is generally constructed based on PID controllers with fixed parameters. However, traditional parameters tuning methods are experience-based, cumbersome, and time-consuming. This study proposes a prior model data-driven finite difference stochastic approximation hybrid knowledge-informed model-free optimization method (HK-FDSA) to tune the controller parameters. Firstly, with the help of a rough model for the relationship between control system performance and control parameters, the HK-FDSA could estimate the rough gradient at every current iteration point. Thus, it could obtain a gradient without online experimental calculation based on the finite differencing algorithm, which reduces the gradient estimation cost during the optimization. Secondly, to reduce the gradient estimation error of the model, the iterative point data would be brought into the prior historical data, and the model would be updated dynamically. Finally, an intelligent termination control mechanism has been integrated into the method to accomplish optimization progress control. The simulation results demonstrated the effectiveness and efficiency of the HK-FDSA based method.