Quality classification model with machine learning for porosity prediction in laser welding aluminum alloys
Bibliographic record
Abstract
The growing implementation of aluminum alloys in industry has focused interest on studying transformation processes such as laser welding. This process generates different kinds of signals that can be monitored and used to evaluate it and make a quality analysis of the final product. Internal defects that are difficult to detect, such as porosity, are one of the most critical irregularities in laser welding. This kind of defect may result in a critical failure of the manufactured goods, affecting the final user. In this research, a porosity prediction method using a high-speed camera monitoring system and machine learning (ML) algorithms is proposed and studied to find the most performant methodology to resolve the prediction problem. The methodology includes feature extraction by high-speed X-ray analysis, feature engineering and selection, imbalance treatment, and the evaluation of the ML algorithms by metrics such as accuracy, AUC (area under the curve), and F1. As a result, it was found that the best ML algorithm for porosity prediction in the proposed setup is Random Forest with a 0.83 AUC and 75% accuracy, 0.75 in the F1 score for no porosity, and 0.76 in the F1 score for porosity. The results of the proposed model and methodology indicate that they could be implemented in industrial applications for enhancing the final product quality for welded plates, reducing process waste and product quality analysis time, and increasing the operational performance of the process.
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 | 0.000 |
| Research integrity | 0.001 | 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".