An interpretable transfer learning method for bearing diagnosis across different systems, faults, and signal types
Bibliographic record
Abstract
Bearing fault diagnosis is critical for the maintenance of mechanical systems. This paper proposes a transfer learning approach across different systems, faults, and signal types with limited labeled data. The core idea of this study is to integrate feature reshaping based on continuous wavelet transform and model fine-tuning, enhancing the model’s adaptability across different tasks. Feature reshaping based on spectral analysis improves the transferability of data within the model, while model fine-tuning aims to enhance diagnostic accuracy and accommodate the requirements of the target domain. To validate the feasibility and generalizability of the proposed method, two case studies were conducted. The results of case study 1 demonstrate that the method can achieve effective transfer learning across different machines, fault types, and label quantities, yielding high accuracy. Case study 2 explores transfer learning between different signal types, showing that acoustic signals can be successfully transferred to a vibration-based model. In addition, this paper uses Shapley Additive Explanation (SHAP) to interpret the transfer learning model. The SHAP analysis reveals that the model effectively captures the key time–frequency features associated with bearing faults. Feature reshaping enhances the signal-to-noise ratio, enabling the model to focus more on fault-related features rather than noise. SHAP analysis clearly highlights the feature differences between various fault types and identifies the critical factors underlying the model’s decision-making process. These findings validate the importance of feature reshaping and fine-tuning in improving the classification performance of the model.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.006 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.002 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".