Fault Diagnosis of Motor Bearing Based on the Bayesian Network

On the basis of the analysis of vibration characteristics of motor bearing faults and the influence from testing noise on site, calculation of the symptom parameters representing the vibration signals according to the measured vibration signals is proposed, and the sensitivity analysis is carried out to these parameters which can refine effective symptom parameters. As there are limitations for motor bearing fault intelligent diagnosis methods based on genetic algorithm and neural network, while Bayesian network has a good learning, inference and astringency. Therefore, the effective combination of the symptom parameters and Bayesian network is made and a new intelligent diagnosis method is posed.