The use of constraint-based design knowledge to improve the search for feasible designs

Abstract

Design is a process of constraint identification and satisfaction. The process of constraint satisfaction is complex, because coupling between the constraints hinders the process of searching for a feasible design. This paper presents computational methods to aid designers by automatically searching for a feasible design, subject to a set of highly coupled, non-linear, equality and inequality constraints. The tools combine symbolic methods, adaptive searches (genetic algorithms and simulated annealing), and methods to incorporate knowledge of the constraints into the search process. The paper concludes with a comparison of two approaches: one using simulated annealing and one using genetic algorithms. The author found that including knowledge of the constraint space into the adaptive searches significantly improved the efficiency of the search process over the simple, untailored application of the methods.