Scholarlink

<p>The stable operation of Pressurized Water Reactors (PWRs) critically depends on the effective control of the pressurizer (PZR), a nonlinear multi-input multi-output (MIMO) subsystem responsible for regulating primary coolant pressure and water level during both steady-state and transient conditions. This study proposes a Fuzzy Logic-based Adaptive PID (APID) controller for simultaneous pressure and level control in the PZR system. In contrast to contemporary machine learning (ML) and artificial intelligence (AI)-based control methods which often entail high computational cost, limited interpretability, and certification challenges the proposed APID controller emphasizes real-time feasibility, transparency, and regulatory compliance, which are essential in nuclear safety applications. The controller retains the intuitive structure of classical PID control while integrating rule-based fuzzy logic for online gain adaptation. This hybrid architecture facilitates explainable decision-making, enhances operator trust, and supports formal verification. The controller was deployed on a low-cost STM32F407 microcontroller, meeting strict memory and timing constraints, and was validated through extensive Processor-in-the-Loop (PIL) testing. Utilizing a high-fidelity, nonlinear two-region thermodynamic model of the PZR, simulation results across diverse operating scenarios including load-following transients, coolant flow loss, and setpoint shifts demonstrate significant performance enhancements over conventional PID controllers. Key improvements include up to 90% reduction in pressure overshoot, faster settling times, minimal steady-state error, and robustness against large disturbances, all achieved without the unpredictability commonly associated with data-driven ML models. By bridging adaptive intelligent control with embedded real-time constraints, this work offers a certifiable, high-performance, and transparent solution for next-generation nuclear reactor control systems, aligning theoretical innovation with practical deployment.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Processor-in-the-Loop Simulation and Multivariable Control System Design for Pressurizer System in Nuclear Power Plants

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年7月2日

查看全文
  • Amal A. Sheta,
  • Elsayed H. Ali,
  • Refaat M. Fikry,
  • Tarek A. Mahmoud,
  • Sayed M. El-Araby,
  • Mohammed I. Mahmoud

摘要

The stable operation of Pressurized Water Reactors (PWRs) critically depends on the effective control of the pressurizer (PZR), a nonlinear multi-input multi-output (MIMO) subsystem responsible for regulating primary coolant pressure and water level during both steady-state and transient conditions. This study proposes a Fuzzy Logic-based Adaptive PID (APID) controller for simultaneous pressure and level control in the PZR system. In contrast to contemporary machine learning (ML) and artificial intelligence (AI)-based control methods which often entail high computational cost, limited interpretability, and certification challenges the proposed APID controller emphasizes real-time feasibility, transparency, and regulatory compliance, which are essential in nuclear safety applications. The controller retains the intuitive structure of classical PID control while integrating rule-based fuzzy logic for online gain adaptation. This hybrid architecture facilitates explainable decision-making, enhances operator trust, and supports formal verification. The controller was deployed on a low-cost STM32F407 microcontroller, meeting strict memory and timing constraints, and was validated through extensive Processor-in-the-Loop (PIL) testing. Utilizing a high-fidelity, nonlinear two-region thermodynamic model of the PZR, simulation results across diverse operating scenarios including load-following transients, coolant flow loss, and setpoint shifts demonstrate significant performance enhancements over conventional PID controllers. Key improvements include up to 90% reduction in pressure overshoot, faster settling times, minimal steady-state error, and robustness against large disturbances, all achieved without the unpredictability commonly associated with data-driven ML models. By bridging adaptive intelligent control with embedded real-time constraints, this work offers a certifiable, high-performance, and transparent solution for next-generation nuclear reactor control systems, aligning theoretical innovation with practical deployment.