Automated exploration of design concepts

Modern product engineering faces the challenge of making numerous design tradeoffs for high-dimensional, interdisciplinary nonlinear systems. This paper presents a new method for exploring product concepts, including an insightful definition of design concept, an algorithmic approach to evaluating high-level design options, and a concise way to communicate simulation and optimization results to product designers and architects. The feasible landscape of architectural alternatives is described in terms of a list of optimized designs, unique design concepts, and implementation rules. Our implementation of automated concept exploration (ACE), which includes very broad mapping of the design space, gradient-based optimization, and rule mining, is demonstrated for the design of a high-performance liquid crystal display (LCD).