Research on the Fault Diagnosis of Wind Turbine Gearbox Based on Bayesian Networks

The speed-up gearbox is one of key components in the large-sized wind turbine. During the operation, some faults often cause long maintenance downtime and higher cost. In this investigation, the gearbox faults were diagnosed by Bayesian Networks method. Based on the analysis of fault factors, the different signal features of fault diagnosis were confirmed. According to Bayesian Networks theory, the fault model of speed-up gearbox was established. The probability of sub-node was obtained by the conditional probability relationship of different nodes. Using the conditional independence of each node, and simplifying the probability distribution, the fault probability was counted out. Finally, the availability of Bayesian Networks method is proved by a calculation case on the test-platform. The study shows that the method can improve the fault diagnosis and operation level of the large-sized wind turbine when be used to judge the fault position in the gearbox.