Design search under probabilistic specifications using genetic algorithms

In this paper a specification-based design evaluation method is presented and applied to optimization using genetic algorithms. The evaluation model is intended to emulate how specifications are used by product designers in a concurrent design environment. The approach allows designers to formulate optimization problems in terms of design specifications. The method is intended for a wide variety of design applications, including the location of feasible designs or starting points for other mathematical programming techniques. The model is demonstrated through a computer implementation that uses a genetic optimization and search algorithm. The penalty-based objective formulation strives for designs which are acceptable according to specifications.