CONTENT BASED SEARCH OF MECHANICAL ASSEMBLIES

Abstract

The increased use of 3D CAD systems by product development organizations has resulted in large databases of assemblies; this explosion of assembly data will continue in the future. Currently, there are no effective content-based techniques to search these databases. Ability to perform content-based searches on these databases is expected to help the designers in the following two ways. First, it can facilitate reuse of existing assembly designs, thereby reducing the design time. Second, a lot of useful Design for Manufacturing and Assembly (DFMA) knowledge is embedded in existing assemblies. Therefore a capability to locate existing assemblies and examine them can be used as a learning tool by the designers to learn from the existing assembly designs and hence transfer the best DFMA practices to new designers. This thesis describes a system for performing content-based searches on assembly databases. It lists the templates identified for comprehensive search definitions and describes algorithms to perform content-based searches for mechanical assemblies. The characteristics of mechanical assemblies were identified and categorized based on their similarity and computational complexity to perform comparison. The characteristics were extracted from the CAD data to prepare a CAD independent signature of the assembly. The search methodology consists of exact and approximate string matching, number matching and computing graph compatibility. Various research groups have solved the former two problems. This thesis describes a new algorithm to solve graph compatibility problem using branch and bound search. The performance of this algorithm has been experimentally characterized using randomly generated graphs. This search software provides a CAD format independent tool to perform content based search of assemblies based on the form of assemblies. The capabilities of the search software have been illustrated in this thesis through several examples. This search tool can contribute to significantly reduce the design time and reuse of the knowledge in existing designs.