Fault Detection and Diagnosis in Nuclear Power Systems

Fault detection and diagnosis (FDD) in nuclear power systems encompasses the identification, localisation and characterisation of deviations from normal operation within reactors and associated equipment. The primary aim is to maintain safety margins, enhance reliability and reduce unscheduled outages by rapidly distinguishing between sensor errors, component malfunctions and process anomalies. Approaches span model-based techniques, which rely on first-principles representations of reactor physics and thermohydraulics, and data-driven strategies that leverage statistical analysis or machine learning to detect patterns indicative of incipient faults. Emerging hybrid frameworks combine knowledge-driven rules with adaptive algorithms to manage complex interactions among multiple subsystems. Prognostics and Health Management (PHM) further extends FDD by forecasting remaining useful life of critical components, enabling predictive maintenance schedules. Advances in signal processing—such as empirical mode decomposition—and deep learning architectures have improved sensitivity to subtle perturbations in neutron flux, coolant flow and vibration signals. Real-time implementation of FDD systems supports automated control adjustments and operator decision-support, reinforcing global safety standards and facilitating extended operation of existing reactors as well as advanced designs.

Research from Nature Portfolio

No recent Nature Portfolio content available.

Fault Detection and Diagnosis in Nuclear Power Systems publication trend

The graph below shows the total number of articles in fault detection and diagnosis in nuclear power systems across all publications each year (not limited to Nature Index journals).

Technical terms

Fault detection and diagnosis (FDD): The process of identifying and characterising deviations from expected system behaviour.

Model-based methods: Techniques that use mathematical or physical models of system dynamics to detect inconsistencies indicative of faults.

Data-driven methods: Approaches that employ statistical or machine learning algorithms to uncover anomalies directly from operational data.

Prognostics and Health Management (PHM): A framework combining monitoring, diagnostics and remaining-useful-life estimation to support predictive maintenance.

Pressurised Water Reactor (PWR): A nuclear reactor type in which water under high pressure acts as both coolant and neutron moderator.

Ensemble Empirical Mode Decomposition (EEMD): A signal processing technique that decomposes complex time series into intrinsic mode functions for feature extraction.

Neural network: A computational model inspired by the brain’s architecture, used for pattern recognition and classification.

Support Vector Machine (SVM): A supervised learning algorithm that finds an optimal boundary to separate data classes.

K-Nearest Neighbour (KNN): A classification method that assigns labels based on the closest training examples in feature space.

References

1. Fault Diagnosis Techniques for Nuclear Power Plants: A Review from the Artificial Intelligence Perspective. Energies (2023).
2. Fault Detection and Isolation of a Pressurized Water Reactor Based on Neural Network and K-Nearest Neighbor. IEEE Access (2022).
3. Health State Identification Method of Nuclear Power Main Circulating Pump Based on EEMD and OQGA-SVM. Electronics (2023).