Genetic Algorithms Versus Simulated Annealing: Satisfaction of Large Sets of Algebraic Mechanical Design Constraints

Abstract

The embodiment phase of mechanical design is, in part, a constraint specification and satisfaction problem. A set of tools to specify and satisfy sets of inequality and equality non-linear constraints has been developed and is presented in this paper. A combination of constraint-based methods and optimization techniques—a direct search method and two adaptive search methods—was applied to the problem of constraint satisfaction of large sets of highly coupled, non-linear, equality and inequality constraints. The constraint-based methods were used to reduce the constraint space size through algebraic manipulation of the constraints. Adaptive search techniques, genetic algorithms and simulated annealing, were used to find designs that satisfy all the imposed constraints.