Research on assembly retrieval based on graph edit distance

Assembly retrieval techniques are an important part of improving the reuse of assembly models and enhancing industrial productivity. In this paper, the similarity between assembly models is measured by introducing a bipartite graph matching method based on graph edit distance (GED). Firstly, the assemblies are encoded according to the main features of the components or parts in the assemblies, and the linkage costs are proposed on the linkage relationships to transform the assembly models into attribute neighborhood graphs. Secondly, drawing on the theory of dichotomous graphs, the GED is used to construct the cost matrices between attribute neighborhood graphs, and the optimal matching is performed on the cost arrays as a solution. Finally, the correctness and effectiveness of our method are verified using instances, and the performance of the two optimal matching algorithms, Hungarian and Volgenant-Jonker, in different assembly retrieval scenarios is compared, which further promotes the application of the GED in the field of assembly retrieval.