Combining Feature Learning and Transfer Learning in Balancing Anomaly Detection for Gas Turbine Engine Vibration Analysis
Bibliographic record
Abstract
Abstract Rotor imbalance is a vital measure that indicates the health state of a gas turbine (GT). Abnormal balancing patterns will lead to excessive vibration and gradually compromise structural integrity. This paper presents the construction of anomaly detection (AD) models that recognize abnormal balancing patterns for two aeroderivative GTs, AGT-A and AGT-B, from Siemens Energy. Such a diagnostic tool can predict at an early stage whether a high vibration would occur during the vibration test and avoid engine reject for re-balance. Machine learning (ML) algorithms have been extensively utilized to conduct GT design space exploration and condition monitoring. However, ML has not been implemented to improve the efficiency of GT manufacturing processes, mainly due to data scarcity. The authors propose a combined feature learning and transfer learning technique to leverage the data resources of GT manufacturing processes. The physical and operational similarities between GTs belonging to the same series imply the transferability of features between models. The normal balancing patterns of the data-rich AGT-A were first learned by a sparse autoencoder to detect balancing anomalies. Then, the learned features were used to initialize the balancing AD model for the data-poor AGT-B. The test accuracy of the AGT-B AD model was increased from 75% to 92% with transfer learning. The presented methodology can facilitate and enable various data-driven analysis tasks for the manufacturing processes of original equipment manufacturers.
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 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 | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.001 |
| 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 itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, 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".