19th Asia-Pacific Vibration Conference (APVC2021)

In order to solve the difficulties in extracting early weak fault features and the low diagnosis efficiency of wind turbine rolling bearings, thus a fault diagnosis method of wind turbine bearing based on Ensemble Empirical Mode Decomposition and Improved Deep convolutional Neural Network (EEMD-IDCNN) is proposed in this paper. The EEMD-IDCNN method can realize an end-to-end processing of the original vibration signal and improve the adaptability of the algorithm. Firstly, the periodic extension method of signal is used to solve the end effect of Ensemble Empirical Mode Decomposition (EEMD). Secondly, the Intrinsic Mode Function (IMF) components generated by EEMD are obtained, and the Continuous Wavelet Transform (CWT) is used to get the time-frequency characteristic diagram. Then, the time-frequency characteristic diagram is convoluted to obtain the feature matrix, and the batch normalization layer is added between the convolution layer and the pooling layer to reduce the uncertainty of the data features and improve the generalization ability of fault diagnosis. Finally, through the experimental analysis of bearing data collected by actual engineering, it is proved that this method is more accurate than other methods and has a wider diagnostic range.