Evolutionary Design of Planar Kinematic Chains

Abstract A procedure is developed to optimize planar mechanism type. A Genetic Algorithm is used to cycle populations of kinematic chain link adjacency matrices, through selection, crossover, and mutation. During this optimization, fit kinematic chains survive while unfit kinematic chains do not. Upon convergence, synthesized kinematic chains of high fitness remain. This technique was lead to be called the Genetic Algorithm for Type Synthesis (GATS). GATS introduces four new ideas for the type synthesis of mechanisms. First, it does not permute all possible kinematic chains. It searches for the best kinematic chains depending on a designer’s specifications. Second, larger size mechanisms can be generated because of the genetic algorithm’s evolutionary naturalness. Third, a novel approach was applied to genetic algorithms to allow the encodings to mutate in size. This allowed for addition or elimination of links in kinematic chains during evolution. Forth, a new property was deduced from mechanism topography that describes the mechanism design flexibility.