基于深度强化学习的用于柔性作业车间的动态调度方法及装置
By constructing a Markov decision process for flexible work shops, designing a continuous and infinite weight action space and reward function, and using the DDPG algorithm to train a policy network to generate the optimal weight combination rules, the problem of limited rule space in flexible work shop scheduling is solved, and scheduling performance is improved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- NANJING UNIV OF AERONAUTICS & ASTRONAUTICS
- Filing Date
- 2024-04-03
- Publication Date
- 2026-07-17
AI Technical Summary
In existing technologies, the dynamic scheduling method for flexible workshops has a limited rule space to choose at each scheduling decision point, which limits the improvement of scheduling performance and fails to meet the needs of rapid response.
We employ a deep reinforcement learning-based approach to construct a Markov decision process, design a continuous infinite weight action space and reward function, and use the DDPG algorithm to train a policy network to generate the optimal weight combination rule, thereby improving scheduling performance.
By selecting a more effective rule than a single scheduling rule at each scheduling decision point, the scheduling performance of the flexible workshop is significantly improved, and the scheduling effect of the production system is optimized.
Smart Images
Figure CN118153896B_ABST