Research on switch cabinet fault diagnosis algorithm based on voiceprint feature fusion

As an important equipment in industrial production, the health of switchgear directly affects the success or failure of production. However, the traditional fault diagnosis methods are not easy to obtain accurate fault features, and the measurement of feature distribution difference between different working conditions is not fully domain adaptive, so it is difficult to achieve better recognition accuracy. In addition, certain background noise is generated during the operation of the switch cabinet, which produces certain interference and affects the accuracy of fault identification. In order to overcome the above limitations, an air compressor fault diagnosis method based on feature fusion is proposed. Firstly, the Mel cepstrum coefficient feature and wavelet transform feature of the switchgear are extracted respectively. Then, at the decision-making level, the confidence score and prediction bounding box are fused late. According to the evaluation index, the best network model is obtained to complete the classification. The experimental results show that the proposed feature fusion method achieves better recognition effect.