Anomaly Detection and Analysis in Nuclear Power Plants
Abhishek Chaudhary,Junseo Han,Seongah Kim,Aram Kim,Sunoh Choi
DOI: https://doi.org/10.3390/electronics13224428
IF: 2.9
2024-11-13
Electronics
Abstract:Industries are increasingly adopting digital systems to improve control and accessibility by providing real-time monitoring and early alerts for potential issues. While digital transformation fuels exponential growth, it exposes these industries to cyberattacks. For critical sectors such as nuclear power plants, a cyberattack not only risks damaging the facility but also endangers human lives. In today's digital world, enormous amounts of data are generated, and the analysis of these data can help ensure effectiveness, including security. In this study, we analyzed the data using a deep learning model for early detection of abnormal behavior. We first examined the Asherah Nuclear Power Plant simulator by initiating three different cyberattacks, each targeting a different system, thereby collecting and analyzing data from the simulator. Second, a Bi-LSTM model was used to detect anomalies in the simulator, which detected it before the plant's protection system was activated in response to a threat. Finally, we applied explainable AI (XAI) to acquire insight into how distinctive features contribute to the detection of anomalies. XAI provides valuable explanations of model behavior by revealing how specific features influence anomaly detection during attacks. This research proposes an effective anomaly detection technique and interpretability to better understand counter-cyber threats in critical industries, such as nuclear plants.
engineering, electrical & electronic,computer science, information systems,physics, applied