Novel Fault Diagnosis Method for Rolling Bearing Based on Voiceprint Recognition With FasterNet‐CAM

ABSTRACT Contact measuring tools are not suitable in some specific working environments, such as high temperature or chemical metallurgical equipment, when non‐contact sensors should be considered. In this study, a rolling bearing fault diagnosis method based on voiceprint recognition is proposed. The original signal is converted into a Mel‐spectrum that can characterize the voiceprint characteristics which based on the features of human hearing, the idea of partial convolution is used for further feature extraction, and then input into the enhanced FasterNet network for classification. The Group‐CAM is integrated with the FasterNet network to confirm the significant portions of the voiceprint associated with the decision, thereby conforming the validity of the model's judgment throughout the recognition process. The experimental results show that the proposed method has an accuracy of 99.4%, a reasoning time of 4.48 s, and a throughput of 223.3 fps after iteration, which is optimal in the compared experiment, indicating that the model meets the lightweight requirement and can identify the acoustic signals of faulty bearings effectively. The method also intuitively highlights the key parts of the acoustic signals, which ensures that the decision‐making process of the model is transparent and trustworthy and enhances the interpretability and reliability of the diagnostic process.