一种面向机械设备分布式制造的超启发强化学习调度方法
By employing the Q-Learning super-inspired reinforcement learning scheduling method, combined with three-dimensional vector encoding and a "pre-insertion" strategy, the problems of maximum completion time and total energy consumption in DFJSPC are solved, achieving efficient resource utilization and energy consumption reduction in the mechanical equipment manufacturing process.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- KUNMING UNIV OF SCI & TECH
- Filing Date
- 2023-03-21
- Publication Date
- 2026-07-17
AI Technical Summary
In the process of manufacturing mechanical equipment, the Distributed Flexible Job Shop Scheduling Problem (DFJSPC) is complex. Existing algorithms are difficult to effectively reduce the maximum completion time and total energy consumption, and do not fully consider the constraints of workpiece transportation resources and energy consumption in the factory.
We employ a Q-Learning-based super-heuristic reinforcement learning scheduling method, combining three-dimensional vector encoding and an "insertion" strategy to design low-level heuristic operations and embed energy-saving and green heuristic operations. We select the best low-level operations through high-level strategies to prevent premature convergence of the algorithm and improve population diversity and search efficiency.
This minimizes the maximum completion time and total energy consumption during the manufacturing process of mechanical equipment, improves resource utilization, reduces production costs and energy consumption, and enhances the economic benefits of enterprises.
Smart Images
Figure CN116300748B_ABST