Reliable Collaboration Chain Mining for Workshop Manufacturing Services Based on Non-Local Graph Convolutional Networks

In order to extract reliable collaboration patterns from the historical data of workshop manufacturing services, the improved graph convolutional network (GCN) method is proposed, and the non-local feature aggregation strategy is introduced to improve algorithm stability. According to the collaboration mechanism of manufacturing service and the actual manufacturing process, the historical manufacturing service collaboration chain is constructed as graph data, where the nodes are manufacturing services, and the edges are service collaborations. GCN is then employed to process the graph of manufacturing service collaboration chains systematically and efficiently, considering the manufacturing service attribute, service collaboration attribute, and collaboration chain topology are considered. Reliability is quantified as the ability to produce sufficient qualified products within a specified time as the label. Experiments with the data of a printed circuit board (PCB) manufacturing enterprise verify the validity and feasibility of the proposed method. The proposed method is stable and has high precision.