Journal of Manufacturing Systems, Год журнала: 2024, Номер 78, С. 351 - 369
Опубликована: Дек. 25, 2024
Язык: Английский
Journal of Manufacturing Systems, Год журнала: 2024, Номер 78, С. 351 - 369
Опубликована: Дек. 25, 2024
Язык: Английский
Journal of Manufacturing Systems, Год журнала: 2025, Номер 80, С. 824 - 840
Опубликована: Апрель 25, 2025
Язык: Английский
Процитировано
0Applied Sciences, Год журнала: 2025, Номер 15(10), С. 5648 - 5648
Опубликована: Май 19, 2025
In view of the Flexible Job-shop Scheduling Problem (FJSP) under multi-product and variable-batch production modes, this paper presents an intelligent scheduling approach based on a heterogeneity-enhanced graph neural network combined with deep reinforcement learning. By constructing incidence to dynamically represent state, proposed method effectively captures both dependencies among operations interaction features between machines. Moreover, Proximal Policy Optimization (PPO) algorithm is leveraged achieve end-to-end optimization decisions. Specifically, FJSP formulated as Markov Decision Process. A heterogeneous enhanced architecture designed extract from operation nodes, machine their relationships. Then, policy generates joint actions for assignment selection, while PPO iteratively refines policy. Finally, validated in aerospace component machining workshop scenario benchmark dataset. Experimental results demonstrate that, compared traditional dispatching rules existing learning techniques, not only achieves superior performance but also maintains excellent balance response efficiency quality.
Язык: Английский
Процитировано
0Journal of Manufacturing Systems, Год журнала: 2024, Номер 78, С. 351 - 369
Опубликована: Дек. 25, 2024
Язык: Английский
Процитировано
0