Objectives To address the performance degradation in fault diagnosis of rotating machinery caused by noise interference in practical applications, a novel fault diagnosis approach based on Mel-frequency cepstral coefficients (MFCC) and a parallel dual-channel convolutional neural network (PDCNN) is proposed. This method aims to improve the quality of fault feature extraction from vibration signals and enhance fault diagnosis capabilities under noisy conditions.

Methods The MFCC is used to extract features from vibration signals contaminated by noise. Meanwhile, a novel parallel dual-channel convolutional neural network structure is designed to explore both global features and deeper, finer details of the data, thereby enhancing the diagnostic performance of the method in strong noise environments.

Results Experimental evaluation results under different noise conditions show that the proposed method achieves a fault diagnosis accuracy of over 98% in environments with strong noise. Its robustness to noise and diagnostic performance significantly surpass traditional methods.

Conclusions The findings of this study can provide valuable references for gearbox fault diagnosis in environments with strong noise.