Transients Analysis of a Nuclear Power Plant Component for Fault Diagnosis

Abstract

We analyze signal data collected during 148 shut-down transients of a nuclear power plant (NPP) turbine for fault diagnosis. The aim is to identify groups of transients with similar characteristics, under the conjecture that faults and malfunctioning of the same type lead to a similar behavior of the measured signal transients. The fault diagnosis method is based on the combined use of an original fuzzy-based slope and similarity analysis with spectral clustering.