Application of Bayesian Theory in Fault Diagnosis of Turbo-generators

Building on the analysis of the features of the sealing oil system faults in turbo-generators this paper mainly discusses how to employ Bayesian theory to perform fault diagnosis by providing mathematical formulae concerning the solution to the fault diagnosis and determining the Bayesian network inference methodology based on the prior information of the samples. It is demonstrated that the application of Bayesian theory, combined with the leaky noisy-OR model which helps to reduce the amount of data required, is conducive to improving the diagnosis speed and efficiency. This paper testifies the validity of this approach and realizes a forecast of the faults at early stages and a rapid diagnosis of their possible causes as well