Key points are not available for this paper at this time.
In highly customized production within Smart Discrete Manufacturing Workshops (SDMW), material feeding and scheduling are critical research areas, as they influence not only the seamless flow of the production line but also inventory and logistics costs. Production data in SDMW serve as a crucial foundation for supporting material feeding and scheduling decisions. However, this data is multi-source and heterogeneous, with complex semantic relationships, making direct application in production decision-making challenging. To address this, this paper proposes an efficient decision-making method for material feeding and scheduling based on knowledge extraction from SDMW data and machine learning algorithms. First, a Temporal Knowledge Graph (TKG) model is established to structure field data and extract relevant knowledge. Next, materials are categorized into three types based on their general utilization ratio, demand rate, and space occupancy ratio. Tailored material feeding strategies are then designed to generate feeding tasks. Additionally, an efficient logistics scheduling decision method is developed using a dual neural network framework, supported by a multi-dimensional knowledge matrix transfer mechanism. Finally, the decision quality of the dual neural networks is tested, and the feasibility of the proposed material feeding and scheduling method is verified through a case study in a turbocharger manufacturing workshop.
Jing et al. (Wed,) studied this question.