Bearings, as core components of mechanical equipment, play a critical role in ensuring equipment safety and reliability. Early fault detection holds significant importance. Addressing the challenges of insufficient robustness in bearing fault diagnosis under industrial high-noise conditions and the difficulty of extracting fault features from a single modality, this study proposes a three-channel multimodal fault diagnosis method that integrates a Convolutional Auto-Encoder (CAE) with a dual attention mechanism (M-CNNBiAM). This approach provides an effective technical solution for the precise diagnosis of bearing faults in high-noise environments. To suppress substantial noise interference, a CAE denoising module was designed to filter out intense noise, providing high-quality input for subsequent diagnostic networks. To address the limitations of single-modal feature extraction and restricted generalization capabilities, a three-channel time–frequency signal joint diagnosis model combining the Continuous Wavelet Transform (CWT) with an attention mechanism was proposed. This approach enables deep mining and efficient fusion of multi-domain features, thereby enhancing fault diagnosis accuracy and generalization capabilities. Experimental results demonstrate that the designed CAE module maintains excellent noise reduction performance even under −10 dB strong noise conditions. When combined with the proposed diagnostic model, it achieves an average diagnostic accuracy of 98% across both the CWRU and self-test datasets, demonstrating outstanding diagnostic precision. Furthermore, under −4 dB noise conditions, it achieves a 94% diagnostic accuracy even without relying on the CAE denoising module. With a single training cycle taking only 6.8 s, it balances training efficiency and diagnostic performance, making it well-suited for real-time, reliable bearing fault diagnosis in industrial environments with high noise levels.

As a core component of mechanical equipment, the operational status of bearings directly determines equipment safety, making early fault diagnosis critically important. However, bearing vibration signals are susceptible to substantial noise interference and exhibit both nonlinear and non-stationary characteristics, rendering traditional single-mode diagnostic methods ineffective at extracting fault features. Therefore, this paper proposes a three-channel multimodal fault diagnosis network (M-CNNBiAM) integrated with a convolutional autoencoder (CAE). Based on a convolutional neural network (CNN) architecture, this network employs CAE for signal denoising, utilizes continuous wavelet transform (CWT) to construct time-frequency features, and incorporates dual enhancement modules: convolutional attention (CBAM) and window attention (S-W-MSA).On one hand, it extracts complementary features from the raw vibration signal and the wavelet transform frequency domain signal, fusing them at the channel dimension. On the other hand, it embeds Shifted Window Attention (SW-MSA) and Window Self-Attention (W-MSA) between convolutional layers to capture global-local features. Combined with CBAM to enhance fault location attention, it mitigates the vanishing gradient problem through residual connections, enabling the extraction of frequency domain features. To address the characteristics of one-dimensional time-series signals, a bidirectional gated recurrent unit (BiGRU) is introduced to collaborate with CNN for extracting temporal features. Experiments demonstrate that on the West China University public dataset and self-test dataset, M-CNNBiAM achieves an average diagnostic accuracy of 95.84% under -10dB high-noise conditions, outperforming comparative methods and validating its superior performance in complex noise environments.