Stages of implementation of the process control system at the NPP based on the predictive analytics model

Abstract

The article discusses the stages of development and implementation of a process control system at nuclear power plants (NPPs) based on Model Predictive Control (MPC) predictive analytics models. The authors propose a comprehensive approach that integrates Industrial Internet of Things (IIoT) technologies, digital twins, predictive diagnostics, and intelligent operator support within the framework of the digital transformation of the nuclear industry. Particular attention is paid to the sequential execution of stages: equipping equipment with smart sensors, developing digital twins, implementing predictive analytics based on machine learning methods (specifically, MSET), and integrating MPC models to optimize real-time control. Case studies of applying this approach to specific power block systems (essential service water supply, reactor coolant pump oil supply, and power output) are considered. The final stage presents the development of an Operator Information Support System (OISS) that provides forecast visualization, decision explanation, and cognitive load reduction. In conclusion, it is emphasized that the implementation of MPC models contributes to increasing the safety, reliability, and economic efficiency of NPPs. The application of end-to-end predictive models based on machine learning methods allows the nuclear industry to systematically transition from reactive maintenance strategies to proactive management. Practical operational references of similar information support systems at the Novovoronezh NPP-2 power units clearly confirm the viability of this approach, demonstrating a reduction in the number of false trips of technological protections by more than twenty percent. Thus, linking predictive analytics with the MPC loop establishes a reliable mathematical foundation for verifying the design lifetime and justifying the safety of NPP elements in the long term.