Stacking deep learning models to analyse vibration and current signatures in rotating machinery
Bibliographic record
Abstract
Purpose We explore the enhancement of deep learning (DL) algorithms from time series data analysis for rotating machinery monitoring. Design/methodology/approach Our proposed method combines the strengths of three distinct neural network architectures: a Multilayer Perceptron (MLP), a Long Short-Term Memory (LSTM) and a Convolutional Neural Network (CNN). The skills of each architecture allow us to tune them to process data from time, frequency and time-frequency domains, respectively. The multi-domain stacking approach has proven effective, confirming its versatility and potential for applicability in several diagnostic scenarios on industry applications. Findings It is challenging to work with high-quality and extensive datasets, remarkably like electrical currents. Electrical current can be interpreted in time, frequency and time-frequency domains. Multi-domain stacking DL models can be tuned to be more suitable for electrical current analysis. Multi-domain stacking approach has applicability in several diagnostic scenarios in industry applications. Originality/value We introduce a novel approach to fault detection by exploiting machine learning algorithms and signal processing techniques across time, frequency and time-frequency domains. Our method integrates data from these domains using three specialized neural network architectures: MLP for structured data from the Fast Fourier Transform (FFT), CNN for matrix-like data from spectrograms and LSTM networks for sequential data with dependencies over time. We stack these specialized architectures to form a neural network framework that enhances data utility, improving classification accuracy rate and noise robustness. It is applicable not only to electrical signal analysis but also to vibration data, offering broad potential for preventive maintenance and operational safety across various industry applications.
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 | 0.001 | 0.002 |
| 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 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
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".