Anomaly Detection and Similarity Search in Neutron Monitor Data for Predictive Maintenance of Nuclear Power Plants

Anomaly detection and similarity search in time series data is an area of wide research in the field of data mining. In this paper we introduce a nearest neighbor based technique for performing anomaly detection over time series data. It is based on the observation that any anomalous behavior is surrounded by a large variation in slope of the graph obtained by plotting the time sequence. Time series comprising of the count of delayed neutrons have been analyzed for the purpose of predictive maintenance in nuclear power plants. We aim to identify anomalies in the neutron counts possibly due to leaks in the nuclear reactor channel.