Digital twin-based assembly process framework utilizing STEP and knowledge graph

Effective information organization and data analysis are foundational for achieving assembly process digital twin. However, the development of the assembly digital twin technology faces significant challenges due to “information silos” resulting from ineffective data exchange among multi-source heterogeneous data involved in assembly processes. This study proposes a Digital Twin-based Assembly Process Framework (DT-APF) designed to systematically organize and manage multi-stage manufacturing data. Through object instantiation, the framework establishes a Digital Twin-based Assembly Process Information Model (DT-APIM) that enables standardized transformation of heterogeneous data. By converting the instance of the DT-APIM into a graph-based structure, the framework generates a Digital Twin-based Assembly Process Knowledge Graph (DT-APKG), achieving deep integration of multi-source data. Furthermore, the integration of Autoregressive Structured Prediction (ASP) algorithm constructs a knowledge reasoning engine, establishing a comprehensive data organization and knowledge reasoning system that spans the entire assembly lifecycle. Experimental validation was conducted using an aero-engine casing assembly case. Results demonstrate that the ASP-based reasoning mechanism validates the effectiveness in identifying assembly performance correlations, while the DT-APF framework improves data integration efficiency by 31.6% compared with traditional knowledge graph construction approaches. This research provides a systematic solution for overcoming information barriers in complex product assembly and enhances the implementation of DT technology in industrial applications.