Application of fuzzy logic for the monitoring and diagnosis of loose parts in the NSSS of nuclear power plants

Abstract

The minimization of false alarms from the loose part monitoring system in nuclear power plant has been a troublesome issue for the system operators and designers because of the difficulty to distinguish the loose-part impact signals from various types of other plant operational noise. The diagnosis of the impact signals, after a loose-part alarm event occurred, has been also another difficulty for the system operators, because specialized technologies and various experiences are required to estimate the impacted loose-part precisely. Therefore, the necessity to develop a loose part monitoring system equipped with the intelligent alarm discrimination and automated impact diagnosis capability has been continued. In consideration of the fuzzy nature of impact signals detected from the complex mechanical structures, and mixed with various types of noise in a nuclear power plant under operation, the loose parts monitoring system with a new signal processing technique using fuzzy logic is proposed. The proposed system is called Fuzzy-LPMS in this paper. In the Fuzzy-LPMS design, comprehensive relations among the impact signal features are taken into account in the fuzzy rule bases for the intelligent alarm discrimination and impact diagnosis. The proposed Fuzzy-LPMS consists of two major parts; an alarm discrimination part based on the fuzzy inference technique for valid signal determination, and an impact diagnosis part based on the hybrid techniques which combine the impact source localization and fuzzy inference techniques for the diagnosis of the loose-part event characteristics. Through the performance test with a mock-up facility, the proposed approach to the loose part monitoring system design has been revealed to be effective not only in suppressing false alarm generation but also in characterizing the loose-part event, i.e., a possible impacted-area (PIA) and a degree of impact magnitude (DIM). In the cases of processing fuzzy and uncertain input signals, th...