The integration of Model Predictive Control and deep Reinforcement Learning for efficient thermal control in thermoforming processes
Bibliographic record
Abstract
This paper presents an approach to integrate Model Predictive Control (MPC) and deep Reinforcement Learning (RL) to improve the efficiency in radiation thermal control systems, specifically in the heating phase of the thermoforming process where a considerable number of radiation heating elements are used as actuators. Because of the large action and state spaces in such systems, the exploration process during the agent training takes a long time. The strategy in this paper employs an MPC to guide and expedite the training process of deep RL agents. While MPC performs optimally with well-defined models and can handle constraints, it requires that model parameters stay constant over time and its online computational burden is notable, especially in systems with extensive action and state spaces. The Proposed approach leverages the predictive capabilities of MPC to provide an external action input that can guide the deep RL agent’s exploration and learning process. Hence, at the end of the training process, the trained agent will perform close to optimally while its online computational burden is very low compared to MPC. In our designed heating system, this hybrid method dramatically accelerates learning, achieving a remarkable 100-fold increase in average episode rewards during training compared to traditional deep RL techniques. Furthermore, the trained agent is not only robust to environmental disturbances, but its online computing burden is 14 times lower than that of MPC. This approach stands as a promising solution for efficient and effective thermal control in industrial 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) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| 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) | 0.001 | 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".