An Efficient Assembly Retrieval Approach based on Compressed Part Parameter Vector

Abstract

Abstract Assembly retrieval techniques enable product-level model reuse, and are relevant for accelerating product design and improving industrial efficiency. Previous work involved describing the assembly using part parameter vectors and solving the problem using the modified Hausdorff distance (MHD). However, this assembly description approach does not provide an in-depth consideration of the part composition of the actual product, nor does it provide a convincing description of the choice of distance paradigm for Hausdorff distance. We proposed a compressed part-parameter vector model to represent the assembly. By searching and removing identical and standard parts that exist within the assembly and between matched assemblies, the original part parameter vector model is compressed into an assembly model that requires less representation space, thereby directly reducing the time consumption of the retrieval process. At the same time, we used a correction coefficient to correct the retrieval results to prevent the model distortion caused by the deletion of parts in the assembly and to affect the retrieval accuracy. Regarding the choice of distance metrics, we studied the commonly used metric paradigms of Euclidean and Manhattan distances in a unified MHD framework. Relevant experimental verifications have demonstrated that our method can effectively improve the efficiency of assembly retrieval.