A dynamic scheduling method with Conv-Dueling and generalized representation based on reinforcement learning
Bibliographic record
Abstract
In modern industrial manufacturing, there are uncertain dynamic disturbances between processing machines and jobs which will disrupt the original production plan. This research focuses on dynamic multi-objective flexible scheduling problems such as the multi-constraint relationship among machines, jobs, and uncertain disturbance events. The possible disturbance events include job insertion, machine breakdown, and processing time change. The paper proposes a conv-dueling network model, a multidimensional state representation of the job processing information, and multiple scheduling objectives for minimizing makespan and delay time, while maximizing the completion punctuality rate. We design a multidimensional state space that includes job and machine processing information, an efficient and complete intelligent agent scheduling action space, and a compound scheduling reward function that combines the main task and the branch task. The unsupervised training of the network model utilizes the dueling-double-deep Q-network (D3QN) algorithm. Finally, based on the multi-constraint and multi-disturbance production environment information, the multidimensional state representation matrix of the job is used as input and the optimal scheduling rules are output after the feature extraction of the conv-dueling network model and decision making. This study carries out simulation experiments on 50 test cases. The results show the proposed conv-dueling network model can quickly converge for DQN, DDQN, and D3QN algorithms, and has good stability and universality. The experimental results indicate that the scheduling algorithm proposed in this paper outperforms DQN, DDQN, and single scheduling algorithms in all three scheduling objectives. It also demonstrates high robustness and excellent comprehensive scheduling performance.
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 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 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".