2025 International Conference on Sensing, Measurement & Data Analytics in the era of Artificial Intelligence (ICSMD 2025)

Current intelligent bearing fault diagnosis models based on end-to-end learning generally suffer from insufficient interpretability, primarily due to the lack of guidance from physical mechanisms. To address this issue, this paper proposes a fault diagnosis method that integrates physical information. By enhancing features through post-processing and converting one-dimensional signals into two-dimensional images, the method effectively embeds physical knowledge. Experiments on publicly available bearing datasets demonstrate that the proposed method achieves a diagnostic accuracy of 97.40%, showing significant advantages over baseline models in terms of stability and accuracy. This validates the effectiveness of physics-guided information in enhancing model performance and interpretability.

fault diagnosis,rolling bearing,cross-operating condition,convolutional neural network,physics informed model