This paper presents the validity and usefulness of early warning in online monitoring system for nuclear power plants. Early warning is one of the core functions of the online monitoring system, which uses pattern recognition to predict and alert potential problems in the equipment or system. This function was developed by using the AAKR technique and has been operated since 2016. We show that the early warning system is operating properly through an analysis of the operation result of the system, and present two cases that represent the role and reliability of the system. In the first case, the system detected a failure of measuring instrument in advance. And the other case, the system went through pattern relearning to overcome environmental changes.

No abstract