Fault Diagnosis of Rolling Bearing Based on CBAM-ICNN

Rolling bearing is one of the key components of mechanical equipment, and it is of great practical significance to ensure its normal operation. Note that the bearing fault diagnosis method based on convolutional neural network cannot adaptively extract key features. A fault diagnosis method is proposed based on short-time fourier transform and convolutional neural network with convolutional block attention module (STFT-CBAM-ICNN) in this paper. Firstly, the one-dimensional vibration signals are transformed into two-dimensional time-frequency images by using short-time fourier transform. Secondly, for the purpose of increasing the network’s attention to important features and improve the ability of feature extraction, the CBAM attention mechanism is embedded into an improved convolutional neural network. In addition, the transformed image data is fed into the CBAM-ICNN module for feature extraction and fault classification. Finally, the effectiveness of the proposed STFT-CBAM-ICNN model in the fault diagnosis of rolling bearing is verified by several sets of comparative experiments.