CADSYN: using case and decomposition knowledge for design synthesis

Design synthesis is a part of the design process during which alternative design solutions are generated. Two process models of design synthesis are case-based reasoning and decomposition. Combining the two approaches allows previous experience to be used directly when available and relevant, and generalized knowledge to be used when direct experience is not appropriate. This hybrid approach to knowledge-based design has been implemented as an extension to an existing knowledge-based design environment, EDESYN. EDESYN provides a domain independent inference mechanism for design by decomposition using a constraint-directed depth-first search through a dynamically defined hierarchy of systems and components. The knowledge base in EDESYN is represented by systems/subsystems, planning rules for decomposing in a specific context, constraints, and mathematical expressions for assigning numerical values to attributes. EDESYN has been extended to accomodate case-based reasoning by adding a case base of design examples and providing a pattern matching algorithm to locate relevant cases. The advantage of using the two approaches in the same system is that the generalized knowledge in EDESYN's knowledge base can be used to transform a previous design situation to the new design context.