DOI: 10.1615/ICHMT.1992.ExpSystComputSimEnergEngin
ISBN Print: 1-56700-031-2
A Real-Time Expert System for Anomaly Diagnosis Based on Reactor Noise
ABSTRAKT
This paper describes the outline of a real-time expert system under development for anomaly diagnosis of a PWR plant based on reactor noise analysis. The noise signal processing includes parametric time-domain and the related frequency-domain methods to supply the expert system with input information. The expert system has three basic components: a noise analysis subsystem, an operating controller and a knowledge-base. The knowledge-base consists of five main parts: monitoring and diagnostic rules, plant structure, plant operation, process data attributes and cause-consequence trees. Operation of the noise diagnostic expert system can be divided into three operational modes: learning mode, monitoring mode and diagnosis mode. The functions of the expert system are being checked using the reactor noise data from an actual reactor obtained during an abnormal occurrence.