Abstract
With the rise of social manufacturing resources and cross-enterprise collaboration, reliable and timely resource scheduling becomes essential for social manufacturing systems. However, dynamic resource availability and uncertain production state complicate scheduling decisions. This paper proposes a deep reinforcement learning-based resource scheduling method in a social manufacturing system with blockchain-supported state management, in which cross-enterprise production data, logistics records, and resource responses are validated and confirmed before being mapped to the scheduling inputs used by the off-chain scheduler. The resource scheduling model is established with the objectives of minimizing time, cost, and resource unresponsiveness, while maximizing the competitive-cooperative effects among manufacturing resources. Afterwards, an adaptive reward backtracking mechanism-based dueling double deep Q-network (ARB-D3QN) algorithm is proposed to solve the proposed scheduling model. Finally, an industrial case of air-conditioner compressor production in the Guangdong-Hong Kong-Macao Greater Bay Area is conducted to evaluate the proposed method. The results show that the ARB-D3QN algorithm achieves the highest mean converged reward among the tested deep reinforcement learning baselines and provides a comparatively balanced trade-off among the scheduling objectives, together with shorter solution time than the multi-objective metaheuristics in the tested scenarios.
| Original language | English |
|---|---|
| Article number | 116646 |
| Journal | Engineering Applications of Artificial Intelligence |
| Volume | 186 |
| DOIs | |
| State | Published - 15 Dec 2026 |
| Externally published | Yes |
Keywords
- Blockchain
- Deep reinforcement learning
- Resource scheduling
- Social manufacturing system
Fingerprint
Dive into the research topics of 'Deep reinforcement learning-based resource scheduling in a blockchain-supported social manufacturing system'. Together they form a unique fingerprint.Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver