Data-Analysis Methods in Detecting, Visualizing and Predicting Nuclear Power Plant Component Ageing Phenomena

Estimating remaining useful lifetime of nuclear power plants has become an important research issue. Our approach in Long-Term Degradation Management project is to try to solve component ageing problems with data-analysis methods. We use e.g. Principal Component Analysis and K-means clustering method to detect data structures or transients from three different data sources for nuclear power plants. We have developed a method for detecting trends in a single cluster. The outcome of the methodologies are shown in three case examples. We were able to visualize anomalies in data, but we did not find evidence for corrosion with this data source.