Comparison of neural networks based on accuracy and robustness in identifying impact location for structural health monitoring applications
Bibliographic record
Abstract
Structural health monitoring systems must provide accuracy and robustness in predicting the structure’s health using the minimum intervention to ensure commercial viability. Characterization of impact is useful in assessing its severity, deciding if detailed damage analysis is necessary, and re-evaluating the present health of the structure under monitoring with better confidence. In this characterization process, the impact location is significant since some positions within a structure are more sensitive to damage. The inherent noise and uncertainties present in the sensor response pose a substantial hurdle to estimating the external impact correctly. This paper quantitatively compares three of the widely used neural networks, namely, Artificial Neural Network (ANN), Convolutional Neural Network (CNN), and Long Short-Term Memory network (LSTM), to estimate impact location from the lead zirconate titanate (PZT) sensor response. For this purpose, a square aluminum plate of 500 × 500 mm was equipped with four PZT sensors; each placed 100 mm away in both the plate directions from a corner and impact loads were given on a grid covering the whole plate. The PZT responses were used to train the three neural networks under study here, and their estimations were compared based on the Mean Absolute Error (MAE). In addition, increasing Gaussian noise was added to the PZT responses, and the robustness of the three neural networks was monitored. It was found that the ANN gives better accuracy with a Mean Absolute Error of 22 mm compared to Convolutional Neural Network (MAE = 31 mm) and Long Short-Term Memory (MAE = 25 mm). However, CNN is more robust when encountering noise with a 2% reduction in accuracy, while LSTM and ANN lost 7% and 11% accuracy, respectively.
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 | 0.003 | 0.013 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.002 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 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".