Abstract:Brushless DC motor is one of the important power devices for large equipment, and the operating status of the motor is highly consistent with the operating status of the equipment. However, current motor fault diagnosis methods are difficult to make accurate state judgments on motors in environments with multiple motors or electromagnetic interference. In order to achieve state diagnosis of brushless DC motors in complex environments, a fusion of convolutional neural network algorithm and long short-term memory network algorithm was studied. Researching the bidirectional propagation of long short-term memory network algorithms to capture the impact characteristics of complex environments on motors, thereby improving the diagnostic accuracy of the model. The experimental results show that the average convergence time of the proposed model on the mechanical and electrical equipment fault diagnosis dataset is 8.91 minutes, and the average convergence time on the motor fault dataset is 12.66 minutes, both of which are lower than the control models in the same group. Secondly, the F1 value of the proposed model is 84.17%, which is 0.87% and 5.08% higher than the control model, respectively. In addition, in the comparison of voltage detection before and after motor faults, the proposed model provides more detailed detection results when motor faults occur. According to the experimental results, it can be concluded that the proposed motor diagnosis model has excellent performance and meets the accuracy requirements of the motor diagnosis industry.
