INTELLIGENT SYSTEM OF TECHNOLOGICAL ROUTE SELECTION BASED ON GRAPH NEURAL NETWORK AND MACHINE-READABLE PRODUCT MODEL

The aim of the article is to develop an intelligent system for automated technological route selection based on graph neural networks and an interface for encoding part attributes. The objective is to formalize the product structure in a machine-readable format and build a GCN model capable of classifying parts by structure and selecting technological templates. Research methods include manual attribute entry through an interface, construction of a directed graph, parameter encoding in one-hot format, training the neural network on labelled data, and implementing a self-learning mode. Novelty lies in combining an interface for detailing part structure without requiring 3D models; a graph model with intersection nodes treated as separate entities; a GCN model for route selection; incorporation of user-defined solutions into the learning process. Results state that the model is trained on 100 graphs and achieves an accuracy of 78.6%. Loss and accuracy dynamics demonstrate stable convergence. Findings show that the system enables representation of parts as graphs, automatic template selection, and accumulation of technologists’ expertise. The solution is geared towards flexible manufacturing tasks and can be integrated into digital CAPP platforms.