Engineering Design Knowledge Management for Conceptual Design

The early stages of engineering design are critical, as the highlevel decisions made at this point have the most impact on the final product’s design, cost, and success. However, little software is available to support engineers during the initial, conceptual design phase. In addition, at this and all other stages of design, engineers are increasingly tasked with utilizing unwieldy collections of data such as databases of legacy designs and catalogs. This work addresses both of these issues. A conceptual design interface with several advancements crucial to industrial deployment is developed and used to aid design. Among these are provisions for real-time collaboration and security. A representation of mechanical devices based on capturing intended function is developed and used by the conceptual design interface to capture design semantics. This representation is defined using a description logic, enabling automated reasoning. The descriptions created using the conceptual design interface can thus be employed to annotate designs, create search queries, and to organize collections of designs. Introduction Although much software is available and employed in industry to support detailed geometric design, tolerancing, simulation, process planning, and other aspects of product development, there remains little support for working at the abstract, conceptual level of design semantics. However, the advantages of providing such support have fostered a great deal of research in this area. Conceptual design interfaces for engineers working in the early stages of design are a prominent line of work in computer-aided design (CAD). Similarly, annotation tools for capturing design semantics—the intended function and behavior of elements in a device and their rationale—have received much research attention but are not widely used in industry. Conceptual design is the process of developing a high-level product design. Abstract ideas are generated and evaluated against the product requirements and each other. Detailed design proceeds according to that seen as most favorable. This early stage in product development is critical, as the general solution path chosen at this point is the largest determinant in the product’s final design, cost, and success. The importance of capturing design semantics begins during the early conceptual design stages and continues throughout the life-cycle of a product. Recorded output from early conceptual design serves to guide and direct the entire design proces. Similarly, rationale for choices made at each level of design must be recorded to guide more detailed development. This information also has great value in enabling later troubleshooting and revision, re-use in variant design, and inclusion into a larger design. ∗Also of the Department of Mechanical Engineering Email: [email protected]. Copyright c © 2004, American Association for Artificial Intelligence (www.aaai.org). All rights reserved. If captured appropriately, design knowledge as recorded by software could also be used in a variety of applications beyond recording the design process. In particular, this knowledge could be used in the management of collections of engineering designs. Much of an engineer’s time is spent examining large collections of corporate legacy data for designs which may be modified to solve new problems, or searching through catalogs in search of components to be incorporated into a design (Ullman 1997). Support for these tasks in commercial environments is typically very limited, consisting of browsing manually-managed, sparse categories based on product line or searching on keywords in associated design documentation or attributes. However, design knowledge—the semantics and concepts behind a design, captured by an annotation and conceptual design interface—can be used in a design repository (Szykman et al. 2000; Szykman, Sriram, & Regli 2001) to provide enhanced versions of these services. Design repositories are an evolution of design databases which apply techniques and methods developed in the knowledge representation community to capture and reason on collections of designs. This knowledge can be used to enable such capabilities as search based on complex device descriptions, classifying devices into sophisticated categorization schemes, and automatically developing categorization schemes for a given set of designs, organizing the collection. All of these can be used in managing engineering design knowledge, providing improved support for variant design and other tasks. In this work the three major components of such a system are developed: a conceptual design/annotation interface, a representation of design semantics, and reasoning mechanisms based on that representation. Figure 1 depicts the interaction between these components. The annotation/conceptual design interface is used to record design semantics. These are captured in the form defined by the representation. This information can then be used as input to a repository and other reasoning systems to provide a variety of different services. The components developed in this work feature several prominent aspects. As design is increasingly performed by geographically and organizationally distributed teams, it is essential that secure, collaborative design environments are provided at all stages of design. The conceptual design and annotation interface developed in this work is therefore designed as a collaborative environment and includes several novel security mechanisms. Design semantics are captured via a representation based on engineering function. This representation is defined in a description logic, providing a formal semantics which enables automated reasoning. In addition, the representation makes use of Semantic Web technology, in particular the W3C Ontology Web Language (OWL)1. This provides a standard description logic language with which design knowledge captured in this form can 1http://www.w3.org/TR/owl-features/