Bayesian model-based fault diagnosis for the rotor

Purpose - The purpose of this paper is to deal with the fault of the rotor system of aeroengine that has too much uncertainty and design a structural diagnosis framework for the rotor. Design/methodology/approach - Bayesian network (BN) is especially suited for capturing and reasoning with uncertainty. This paper adopts the techniques of BN to implement the probability computation of fault occurrence using system information. The rotor system is analyzed in detail and the familiar faults and their corresponding fault symptoms are extracted, then the rotor's BN model based on above information is established. Meanwhile, a framework of the fault diagnosis system based on the network model is developed. Using this model, the conditional probabilities of the faults happened are computed when the observation of the rotor is presented. Findings - The diagnosis methods developed are used to diagnose the actual four kinds of faults of the rotor. The BN model can identify the faults occurred by those probabilities computed. Originality/value - The diagnosis system using BN described in this paper is satisfying and can handle the faults of the rotor.