Recognition of gearbox operation fault state based on CWT-CNN

In view of the problem that the features extracted by the traditional fault diagnosis do not have adaptive ability and are difficult to match specific faults，a method for fault detection of gearbox based on the continuous wavelet transform （CWT） and two-dimensional convolutional neural network （CNN） was proposed.This method constructed the time-frequency diagrams to raw vibration signals through CWT，then built the CNN model using the diagrams as input，and finally formed a deep distributed fault feature expression through the multiple convolutions and pooling operations.The back propagation algorithm was used to adjust the structural parameters of each layer of the network，making the model establish an accurate mapping from the signal characteristics to the fault states.In experiments under different working conditions and fault states，the fault recognition accuracy reached 99.2%，which verified the effectiveness of the proposed method.Using this method of adaptive learning the abundant information in the signal can provide a basis for the intelligent fault diagnosis.