Bearing fault diagnosis based on MCNN-LSTM and cross entropy loss function

Rolling bearings are widely used in all kinds of large machinery and equipment, which play an important role in supporting the mechanical rotating body, reducing friction resistance and ensuring rotation accuracy. Due to the universality and criticality of rolling bearings, real-time state analysis of rolling bearings is of great significance to ensure the normal operation of mechanical equipment. At present, intelligent fault diagnosis method based on vibration signal analysis is widely used in this field. MCNN-LSTM model is used for bearing fault diagnosis. However, in the classification problem, the use of mean square error loss function leads to gradient vanishing, resulting in a low learning rate in the initial training of the model. Therefore, improving its loss function and using cross entropy loss function for training can effectively avoid the problem of gradient vanishing and improve the learning rate of the model. CWRU public data set is used to train and test the model. The experimental results show that the improved cross entropy loss function can not only improve the learning rate of the model, but also achieve higher prediction accuracy and F1 score.