Interpretable degradation tensor modeling through multi-scale and multi-level time-frequency feature fusion for machine health monitoring
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
A health indicator (HI) is crucial for comprehensively characterizing degradation processes influenced by multiple factors. However, HIs derived from physical features typically capture only partial degradation characteristics, making them insensitive to early faults and unable to show monotonic trends. Consequently, incorporating all relevant information is vital for effective machine health monitoring. Tensor features in data-driven methods offer the advantage of aggregating diverse information while preserving explicit structural relationships. Nevertheless, existing deep learning approaches for tensor feature fusion in degradation modeling often require large volumes of high-quality training data and lack interpretability. This study proposes an interpretable degradation tensor modeling methodology that enables multi-scale and multi-level fusion of time-frequency tensor features. This approach generates a set of HIs with diverse characteristics, enhancing early fault detection and monotonic performance evaluation. Initially, time-frequency spectrograms derived from different scales of vibration signals are represented as tensor features. A degradation tensor model is introduced to optimize tensor weights and achieve feature-level fusion for HI construction. An interesting finding is that time-frequency spectrograms with larger scales yield HIs with greater sensitivity to early machine faults, while smaller-scale spectrograms produce HIs with more pronounced monotonic trends. The methodology further integrates process control techniques with the proposed HIs for decision-level fusion, facilitating accurate early fault detection and monotonic deterioration assessment. Validation through two endurance tests demonstrates the superior performance of the proposed methodology compared to famous and classical approaches. Additionally, the optimized tensor weights can identify informative frequency bands associated with machine faults, enhancing fault diagnosis interpretability.
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.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.002 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 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 it