A deep reinforcement learning‐based maintenance optimization for vacuum packaging machines considering product quality degradation
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
Abstract Vacuum loss in packaged meats can lead to product defects resulting in significant economic losses and negative public health issues. Therefore, it is crucial to study the degradation of components that are critical for the provision of vacuum and package sealing to enhance system availability and process safety. Accordingly, this article proposes a condition‐based maintenance policy that integrates quality information considering meat cuts that lack proper vacuum as defective items. A deep reinforcement learning algorithm is used to learn a set of adequate maintenance actions to be performed at each maintenance inspection while maximizing the system availability and/or minimizing the total maintenance cost including the cost of producing defectives items. A numerical case study and benchmarking were performed, demonstrating that the proposed model surpasses the corrective maintenance policy. It leads to a 2.2% increase in system reliability, a 91% reduction in maintenance costs, a 93% reduction in defects identified in production, and a 90% reduction in defects identified on supermarket shelves. Such results demonstrate that the model can (i) prescribe maintenance actions at each inspection according to critical degradation states; (ii) exploit maintenance opportunities that lead to economic savings; and (iii) reduce product reprocessing and propagation of defects to shelves. Practical applications A new machine learning‐based maintenance model promises to revolutionize the vacuum packaging industry by enhancing system reliability, reducing costs, and improving product quality. The model enables managers to make dynamic decisions based on the system state, avoiding inefficient maintenance planning and ensuring maximum productivity. By predicting quality performance through vacuum condition, the model allows for timely decision‐making and reduces the need for costly laboratory analysis. The free‐model's estimation of structural and economic relations of components enables managers to retrain and adapt maintenance policies for optimal system performance. With improved system reliability and reduced production of non‐conforming items, the industry can reduce reprocessing costs, contamination risks, and protect brand image by ensuring better control over meat hygiene. This new approach to maintenance optimization has significant implications for process safety, efficiency of operations, and profits of the vacuum packaging industry, making it a potential game‐changer in the field.
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 | 0.001 | 0.001 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| 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