For many industries, engineering design and manufacturing data needs to be preserved over 50-to-75 year lifespans. Traditional digital data management techniques are usually dependent on the proprietary formats of commercial software systems and cannot guarantee the readability and utility of data over long periods. Hence, while 3D CAD modeling has become indispensable, the engineering part print (i.e., the 2D drawing) still remains a principal method of design knowledge archival. The rich knowledge in 3D CAD about features, manufacturing processes and artifact behavior are simply lost in translation. The research team proposes to develop a representation and algorithmic techniques to archive 3D CAD objects. The approach is to augment shape-based representations with formal models of engineering semantics and domain-driven segmentation algorithms. The rationale is that low-level shape information is repres ntationally straightforward, and easily preserved; whereas native CAD/CAM formats are proprietary and notoriously hard to preserve, even across incremental product versions. Engineering semantics will be based on emerging W3C, Semantic Web, ISO and NIST standards formalisms that will need to be extended to handle new datatypes specific to complex engineering artifacts. These will be mathematical logics described in the non-proprietary syntax of international standards, and thus also easier to preserve than proprietary representations. Lastly, techniques from databases will be extended to extract and populate these representations from existing design tools and engineering documentation.
This approach form the basis of a Digital Archive Toolkit for Engineering (DATE), a set of tools to be developed and tested in collaboration with the government/industrial partner Honeywell FM&T. Working with data from the United States Department of Energy, the PIs will work with the agency-wide Advanced Design and Production Technologies Initiative (ADAPT) team to apply techniques developed in this research to indexing and storage of engineering data used at the Kansas City Plant (KCP). The objective is to extract data to create Digital Engineering Archives, enable answers to meaningful engineering queries on archives of 3D engineering knowledge, and support long-term engineering knowledge preservation.
Intellectual Merits
The relationships among shape and form, structure and function, and behavior and semantics are among the most fundamental questions studied by science and engineering. It is precisely these relationships that must be captured in Digital Engineering Archives. While these problems are large in scope, by focusing on the vital domain of discrete part manufacturing., the proposed project is poised to produce theoretical results, novel techniques, in addition to prototype systems. The aggregate output of the proposed project includes advancements in representation and retrieval algorithms, as well as contributions in basic computer science and engineering design and manufacturing. The scope of the research spans pattern recognition, knowledge representation, database semantics, as well as CAD/CAM/CAPP/PDM/PLM.
Broader Impacts
This research bridges the print information gap by developing a dynamic, self-describing and information-rich alternative to the part print to create Digital Engineering Archives. The availability of such a format will enable the creation of systems for archival, retrieval and reuse of engineering knowledge and activities for design teams operating throughout the product life cycle, which may be 10, 25, or 50 or more years. The proposed DATE system will have an immediate impact on the ability of industry to create data archives. Lastly, the team's collaboration with industrial and government partners position this project to have a significant positive effect on emerging NIST, W3C and ISO standards, and on commercial CAD/CAM/PDM/PLM software systems.
Educational Impact
The project aims to create a novel experience for its graduate and undergraduate students, as well as for the investigators from industry, government and academia. Leveraging Drexel's cooperative education programs (both graduate and undergraduate), the investigators plan extensive personnel exchanges with DOE, NIST and other agencies. Lastly, the team will develop educational and tutorial materials to demonstrate to the engineering community how to apply computer and information science to create and maintain Digital Engineering Archives.
