Decoupling control for pressurizer pressure and water level in nuclear power plants based on a Proportional–Integral–Derivative Neural Network

Abstract

The pressurizer plays a crucial role in nuclear power plants, ensuring reactor safety by maintaining stable pressure and water levels. It is characterized by nonlinearity, time-varying dynamics, and strong coupling. Conventional single-loop PID control systems often fail to decouple pressure and water level effectively, resulting in suboptimal performance, particularly under large load variations. To address these challenges, this study presents a novel decoupling control system based on a Proportional–Integral–Derivative Neural Network (PIDNN). The PIDNN is employed to mitigate the intense coupling effects between control loops. Furthermore, the Dung Beetle Optimizer (DBO) algorithm is integrated to optimize the initial weights of the PIDNN. A three-region pressurizer model is developed in MATLAB/Simulink for validation. The simulation results demonstrate that the proposed method significantly enhances the control performance and dynamic response speed of the pressurizer’s pressure and water level under typical load rejection transients in nuclear power plants, compared to traditional PID control. This method also exhibits advantages in robustness, adaptability, and engineering applicability, offering an innovative and effective solution for the decoupling control of pressurizers in nuclear power plants.