MHDO-MAGE based Performance Optimization Strategy for Steam Generator Level Control System of Nuclear Power Plant

The steam generator level control system is one of the critical control systems in a nuclear power plant, and its performance is vital for the safe and stable operation of the plant. The control parameters are key factors that affect the operational performance of the steam generator level control system. Therefore, during the commissioning of the nuclear power plant, it is usually necessary to adjust the parameters through the trial-and-error method. This tuning process mainly relies on the experience of the commissioning personnel and may potentially compromise plant safety due to human errors. Considering that the parameter tuning process of control systems can be regarded as a special type of batch processes, this paper proposes to apply an efficient data-driven optimization method, multi-source hybrid data-driven optimization method based on model accuracy gradient estimation (MHDO-MAGE), to the problem of parameter tuning for the steam generator control system. This method effectively combines the advantages of traditional MBO (Model-Based Optimization) and MFO (Model-Free Optimization) methods, while efficiently utilizing prior data information related to the optimization problem. This method was implemented on a widely adopted steam generator control system model specifically designed for steam generator simulations. The simulation results clearly indicate that the MHDO-MAGE method significantly enhances the efficiency of parameter tuning for the steam generator level control system, ultimately leading to a reduction in the costs associated with nuclear power plant commissioning.