Research on Fault Diagnosis Method Based on FMEA/FTA and Bayesian Network

Abstract

The construction of Bayesian network (BN) model is a bottleneck problem that needs to be solved urgently in the field of fault diagnosis. Combining failure mode and effect analysis(FMEA) and fault tree analysis(FTA) can solve this problem well. In this paper, a BN model based on FMEA/FTA is proposed. The transformation methods of FMEA to BN and FTA to BN are analyzed. The structural matrix is used to realize the information transformation. The noisy-max model is used to determine the BN parameters. Taking the rub-impact fault between the high pressure rotor and the front shaft seal of a 600 MW turbogenerator unit as an example, the application of the BN model based on FMEA/FTA in fault diagnosis is realized from three aspects which are BN model construction, prior probability and conditional probability assignment, and diagnosis reasoning, respectively.