Abnormal Process State Detection by Cluster Center Point Monitoring in BWR Nuclear Power Plant

This paper proposes a new method to detect abnormal process state. The method is based on cluster center point monitoring in time and is demonstrated in its application to data from Olkiluoto nuclear power plant. Typical statistical features are extracted, mapped to ndimensional space, and clustered online for every time step. The process signals in the constant time window are classified into two clusters by the K-means method. By monitoring features of the process signals, in addition to signal trends and alarm lists, the operator gains a tool that helps in early detection of the pre-stages of a process fault. By using cluster center point time series monitoring, faults in the process can be seen by at first glance or automatically by notification in the alarm list. This provides a definite advantage to any operating personnel and ultimately improves safety at the nuclear power plant.

• Type: Journal
• Year: 2009
• Where: DMIN
• Authors: Jaakko Talonen, Miki Sirola
• Keywords: Alarm List, Cluster Center Point, Data Mining, DMIN 2009, Nuclear Power Plant