Fault Diagnosis of Wind Turbine Bearings in New Energy Systems Based on MR-DCA Method

A wind turbine bearing fault diagnosis method based on MR-DCA is proposed to address the problem of difficult identification of weak faults in wind turbine bearings. The use of maximum correlation kurtosis deconvolution and resonance sparse decomposition methods for preprocessing input samples can effectively filter out noise in the original signal and highlight fault impact components. The two-dimensional time-frequency map of the obtained fault components and the original signal are used as training samples for the network. After two feature learning modules, attention mechanism is used to screen the input features. Weight reassignment can effectively improve the computational efficiency and recognition accuracy of the model. To verify the performance of the model, weak fault data of wind turbine bearings from Case Western Reserve University were used for fault diagnosis analysis, and ablation experiments were conducted to validate the effectiveness of each module of the diagnostic model. The results indicate that the proposed method has higher recognition accuracy and faster training and iteration speeds. The proposed model can achieve good diagnostic performance in the diagnosis of weak faults in wind turbine bearings.