*** 9700040 Barton This grant provides funding for research on the use of metamodels to allow system-level computer-aided design using disparate computer-based subsystem models. Metamodels are approximating models of engineering product or subsystem performance models. The research has several goals: (1) to define an architecture needed to fit the approximating metamodels and integrate these subsystem metamodels to allow efficient examination of system design alternatives; (2) to develop useful measures of the adequacy of a given metamodel for a given design task; (3) to develop sequential methods to improve the adequacy of the metamodel, providing additional accuracy only where it is needed as the design is refined; and (4) to test the methodology on an application to magnetic levitation transport design. If successful, this research will allow increased complexity for systems which are amenable to computer-aided design. For many complex products such as aircraft, computers, automobiles, and telecommunications networks, it is necessary to design subsystems separately because it is too difficult to integrate the computer-based analysis modules to create a model of the entire system. Metamodel-based integration technology will allow the construction of fast, approximate models for entire systems. As a result, designers will be able to consider interaction among subsystems, and develop systems with better performance at lower cost, in a shorter period of time. These advantages will be particularly important to US firms competing in the markets mentioned above. ***
