An adaptive acoustic signal reconstruction and fault diagnosis method for rolling bearings based on SSDAE–MobileViT

Abstract The Stack denoising autoencoder (SDAE) is suitable for acoustic signal denoising because of its ability to learn high-level features automatically, but the reconstruction effect is unstable with high-intensity noise. The reason is that the noise, which is emitted by neighboring equipment, easily disguises the acoustic signals of the target equipment. This reduces the smoothness of the signal and has an impact on the accuracy of the fault diagnosis. Accordingly, this paper presents a supervised SDAE (SSDAE)–mobile vision transformer (MobileViT) model, aiming to identify the fault location and fault degree accurately and efficiently in the presence of substantial background noise interference. First, an SSDAE is established to reduce the high-intensity noise present in the fault acoustic signals; the Huber loss between the reconstructed signal and the theoretical signal is employed to guide the fine-tuning of the model. Subsequently, the mel-frequency cepstral coefficient was used to extract the acoustic features of the reconstructed signal, and it was converted into a mel-frequency spectrogram. Finally, the MobileViT model is utilized for fault classification. Ultimately, an acoustic fault diagnosis model of rolling bearings under high-intensity noise is obtained. According to comparative experiments, the noise reduction method proposed in this paper achieved the highest level of signal-to-noise ratio increment, waveform similarity coefficient, and mean square deviation in real signals when compared with the three traditional noise reduction methods. Furthermore, the average fault diagnosis accuracy of the fault diagnosis model was found to be 99.2%, which was determined to be optimal in comparison with other fault diagnosis models.