Tag-based CAD assembly model retrieval for design reuse

Abstract

With increasing Product Data Management Databases' size with thousands of assemblies having hundreds of parts, the search and retrieval of necessary assembly is becoming difficult. Older approaches like filenames or finding similar assemblies are not sufficient in this new scenario. This thesis addresses the problem of Content-based assembly retrieval from the existing database. The algorithm uses a Graph data structure for storing the information about the parts and their kinematic connections in a given assembly. The method uses a unique tag-based assembly retrieval. The graph of any database assembly is searched for tags which are pre-defined graphs for generally used sub-assemblies using exact graph matching algorithm. The assembly database is converted into a document containing information about the presence and frequency of various tags in the assemblies. The end-user can search with the frequency of different tags in the system, giving them direct access to the most relevant database assemblies as per the search parameters. Because of text-based search usage, the proposed method reduces the time required to retrieve the assembly to about a fraction of seconds.