Belt conveyor roller bearing fault diagnosis based on CGM-MobileNet

Aiming at the problem of multiple working conditions and high noise of belt conveyor in the actual working environment, which leads to the low accuracy of roller bearing fault diagnosis, a CGM-MobileNet belt conveyor roller bearing fault diagnosis method based on CGM-MobileNet is proposed. Firstly, the original one-dimensional vibration signals are obtained from two-dimensional time-frequency image by Continuous Wavelet Transform(CWT), Gramian Angular Field(GAF) and Markov Transition Field(MTF) respectively, and spliced into three-channel two-dimensional time-frequency image; secondly, a lightweight convolutional neural network MobileNet is used as the fault classification network to ensure the diagnostic efficiency; and finally, the CBAM attention mechanism is introduced to improve the stability and accuracy of the network. Using the proposed method for several experiments, the average accuracy rate reaches 99.6%, and the proposed method can meet the high efficiency and accuracy of roller bearing fault diagnosis.