A digital twin modeling approach cooperating design and manufacturing using knowledge graph and Multi-physics field modelling

To achieve integrated and intelligent product manufacturing, it is essential to consider the impact of manufacturing on design, and build simulation modeling to evaluate performance accordingly. The digital twin maps the attributes, structure, and performance of the product in the above stages into the virtual world and builds a high-fidelity model. However, the data in the above processes are diverse and heterogeneous, making it difficult to share, flow, and utilize. Knowledge graph, as a multi-source heterogeneous database, can be used to build twin models by managing the above data. Based on this, this paper proposes a digital twin modeling method based on knowledge graph and multi-physical field coupling. Firstly, the knowledge space utilizes the knowledge graph to manage and reuse design and manufacturing knowledge from the physical space. Secondly, a series of multi-physical field simulation models are established in the virtual space, and their coupling is achieved. Based on the measured signals, the model parameters are corrected using the particle swarm optimization (PSO) method. Thirdly, the models are utilized to complete the design task and the generated knowledge is stored in KG. The design of the gearbox verifies the effectiveness of the method in problems such as part selection.