Self-Organization in Smart Manufacturing— Background, Systematic Review, Challenges and Outlook

IEEE Access, Journal Year: 2023, Volume and Issue: 11, P. 10107 - 10136

Published: Jan. 1, 2023

The concept of smart manufacturing has attracted huge attention in the last years as an answer to increasing complexity, heterogeneity, and dynamism ecosystems. This vision embraces notion autonomous self-organized elements, capable self-management self-decision-making under a context-aware intelligent infrastructure. While dealing with dynamic uncertain environments, these solutions are also contributing generating social impact introducing sustainability into industrial equation thanks development task-specific resources that can be easily adapted, re-used, shared. A lot research context self-organization been produced decade considering different methodologies developed contexts. Most works still conceptual or experimental stage have application scenarios. Thus, it is necessary evaluate their design principles potentiate results. objective this paper threefold. First, introduce main ideas behind manufacturing. Then, through systematic literature review, describe current status terms technological implementation details, mechanisms used, some potential future directions. Finally, presentation outlook summarizes results work interrelation facilitate solutions. By providing holistic overview field, we expect used by academics practitioners guide generate awareness possible requirements, challenges, opportunities self-organizing towards transition.

Language: Английский

---

Combining Reinforcement Learning Algorithms with Graph Neural Networks to Solve Dynamic Job Shop Scheduling Problems

Processes, Journal Year: 2023, Volume and Issue: 11(5), P. 1571 - 1571

Published: May 21, 2023

Smart factories have attracted a lot of attention from scholars for intelligent scheduling problems due to the complexity and dynamics their production processes. The dynamic job shop problem (DJSP), as one problems, aims make an optimized decision sequence based on real-time environment. traditional reinforcement learning (RL) method converts with Markov process combines its own reward obtain sequences in different states. However, definition states often relies experience model constructor, which undoubtedly affects optimization capability model. In this paper, we combine graph neural network (GNN) deep (DRL) algorithm solve DJSP. An agent state analysis rules is constructed, thus avoiding that methods rely artificially set feature vectors. addition, new function defined, experimental results prove our proposed more effective. effectiveness feasibility demonstrated by comparing general algorithms minimizing earlier later completion time, also lays foundation solving DJSP later.

Language: Английский

---

Joint multi-objective dynamic scheduling of machine tools and vehicles in a workshop based on digital twin

Journal of Manufacturing Systems, Journal Year: 2023, Volume and Issue: 70, P. 345 - 358

Published: Aug. 8, 2023

Language: Английский

---

Multirobot collaborative task dynamic scheduling based on multiagent reinforcement learning with heuristic graph convolution considering robot service performance

Journal of Manufacturing Systems, Journal Year: 2023, Volume and Issue: 72, P. 122 - 141

Published: Nov. 30, 2023

Language: Английский

---

Designing and modeling of self-organizing manufacturing system in a digital twin shop floor

The International Journal of Advanced Manufacturing Technology, Journal Year: 2023, Volume and Issue: 131(11), P. 5589 - 5605

Published: Jan. 30, 2023

Language: Английский

---