PARALLEL POPULATIONS GENETIC ALGORITHM FOR MINIMIZING ASSEMBLY VARIATION IN SELECTIVE ASSEMBLY

Selective assembly is a means by which high-precision assemblies are made from relatively low-precision components. This is accomplished by partitioning produced components into groups prior to random assembly. The mating components in the selective groups are then assembled at random. In this work, a parallel population genetic algorithm is developed to find the best combination of selective groups which will lead to overall minimum variation in the assembly tolerance, with minimum number of generation cycles during the GA search process. An attempt is also made to further speed up the convergence and diversification process of the GA by maintaining more number of concurrent parallel populations in the proposed methodology. It is proved that the proposed parallel populations genetic algorithm is much faster than the normal GA with single population.