Building models to support synthesis in early stage product design
R. Bharat Rao, Stephen C.-Y. Lu · National Conference on Artificial Intelligence · 1993
Current computer-aided engineering paradigms for supporting synthesis activities in engineering design require the designer to use analysis simulators iteratively in an optimization loop. While optimization is necessary to achieve a good final design, it has a number of disadvantages during the early stages of design. In the inverse engIneering methodology, machine learning techniques are used to learn a multidirectional model that provides vastly improved synthesis (and analysis) support to the designer. This methodology is demonstrated on the early design of a diesel engine combustion chamber for a truck.