Abstract: As the number of 3D CAD model library models increased, the retrieval efficiency could be greatly reduced. Aiming at this problem, a 3D CAD model retrieval optimization scheme based on mechanical structure shape was proposed. Data grouping and aggregation technology was used to analyze the similarity characteristics of 3D models, and clustering methods such as model similarity comparison matrix and model similarity chain were proposed to simplify the model retrieval process. Through the analysis of the working principle of the distributed computing framework and the integration of the 3D CAD model retrieval technology based on the mechanical structure form, a distributed computing method oriented to the characteristics of the mechanical structure model was proposed to construct an efficient 3D CAD model retrieval system. Finally, the basic data of the ESB model library developed by Purdue University was used for experimental verification. The results show that the application of distributed computing technology effectively optimizes the 3D model retrieval method based on the mechanical structure shape, solves the computing power bottleneck problem of 3D model retrieval, and provides a practical method for the distributed retrieval of mechanical structure 3D model.

Key words: mechanical structure form; three-dimensional model retrieval; distributed computing; similarity chain