26 June 2023 in Mechanical Systems and Signal Processing
23 December 2022 in Reliability Engineering & System Safety
26 December 2022 in Mechanical Systems and Signal Processing
25 July 2024 in Reliability Engineering & System Safety
4 May 2024 in Mechanical Systems and Signal Processing
24 February 2025 in IEEE Transactions on Instrumentation and Measurement
With the rapid development of sensing and computing technology, transfer learning has become increasingly favored for mechanical fault diagnosis due to its ability to handle distribution differences across different domains. However, the interpretability of backbone models used in transfer learning, such as convolutional neural network, recurrent neural network, graph neural network, and transformer, is limited, hindering their acceptance and adoption by industrial users. To address this problem, we propose an interpretable wavelet constrained transformer for diagnostic tasks, designed to extract local features and aggregate global information. Specifically, our model applies the dual-tree complex wavelet constraint to the transformer structure, ensuring approximate shift invariance. This improves diagnostic accuracy while reducing the number of parameters. Additionally, we explore the Einstein summation for matrix multiplication in frequency band blending after wavelet transform to reduces computational complexity and accelerates convergence speed. To enhance the model’s transferability across different domains, we incorporate uncertainty-constrained loss on the model output using temperature scaling and uncertainty reweighting. This effectively reduces class confusion and improve accuracy in the target domain. Considering the necessity of non-contact measurement in mechanical systems for real-world applications, we utilize acoustics signals to verify the effectiveness of our transferable and interpretable model. The experimental results show that, compared with other commonly used models, our model significantly improves cross domain diagnostic accuracy without affecting interpretability.