Research on the effect of wind turbine bearing fault diagnosis method based on multi-feature calculation and Bayesian optimized machine learning method

Wind turbine bearings are one of the most important components of wind turbine generating equipment. Failure problems in wind turbine bearings can affect the operation of the entire plant. The data from sensors can be processed accurately and quickly through machine learning methods to diagnose the bearing failure. In this paper, six sets of experimental data are derived using a combination of feature extraction, principal component analysis, and Bayesian optimization of decision trees. Results are shown that the Bayesian optimized decision tree has higher diagnostic accuracy compared to the traditional decision tree. The principal component analysis method has some optimization effect on the original data, but the accuracy of the data after applying to feature extraction will be reduced. The Bayesian optimized decision tree based on feature extraction has the best results, with an accuracy of 99.8%. The findings of this paper have some reference value in the field of wind turbine bearing fault diagnosis in the future.