Long-term operation monitoring strategy for nuclear power plants based on continuous learning. (1st December 2022)

• Record Type: Journal Article
• Authors: Yu, Yue; Peng, Min-jun; Wang, Hang; Liu, Yong-kuo; Ma, Zhan-guo; Cheng, Shou-yu
• Is Part Of: Annals of nuclear energy. Volume 178(2022)
• Publication Date: 2022-12-01
• Subjects: Principal component analysis -- Continuous learning -- Long-term monitoring -- Training set optimization; Nuclear energy -- Periodicals; Nuclear engineering -- Periodicals; 621.4805
• DOI: 10.1016/j.anucene.2022.109323
• Languages: English
• ISSNs: 0306-4549

Abstract

Highlights: Dynamic threshold design method of principal component analysis. A strategy for principal component analysis model to continuous learning and forgetting new data. Continuous monitoring of the long-term operation of nuclear power plants. Abstract: Safety is the most important feature of nuclear power plant operation. With the continuous development and improvement of the energy market, the monitoring model and monitoring strategy inevitably need to be improved and innovated when encountering new problems, such as the long-term operation monitoring of nuclear power plants. Another key challenge is how to effectively learn new knowledge from continuously collected measurements and how to integrate new information into current monitoring models. Therefore, this paper proposes a continuous learning and forgetting strategy for long-term operation monitoring of nuclear power plants. Then, PCA model is used to combine with this strategy to achieve long-term monitoring of nuclear power plants the simulator is used to insert the fault test. Finally, the test shows that the strategy can complete the long-term operation monitoring of nuclear power plant by learning new knowledge, and can quickly detect several typical faults.