Optimization Design Method Based On Swarm Intelligence Product Tolerance

Manufacturing is a very important industry in national economy; its condition shows national industrialization level. Manufacturing is facing fierce market competition, manufacturing cost, product quality and time these reflect the manufacturing ability of an enterprise. Some advanced design method was presented by some researchers to meet higher consumers'requests. Design for Quality (DFQ) is one of advanced design method, which believes that design stage has effect to product quality intensively. Tolerance optimal design is a main point in DFQ for tolerance is related to product design and manufacturing at the same time. A tolerance design multi-objective model and a hybrid swarm intelligence algorithm were given in the paper by researching tolerance optimal design.To solve the problem that product quality is ignored in cost-tolerance model, a multi-objective model of tolerance design is presented, which is based on Taguchi's quality view and the conception of Pareto optimum set. Manufacturing cost and quality loss are taken as design objective at the same time, the objectives are subject to assembly success rate of statistical tolerance, and the tolerance zone obtained is looser than the worst tolerance method. The traditional particle swarm optimization algorithm is improved, the particle is redefined according to the conception of Pareto optimum, and then the fast non-dominant sorting technology is adopted to sequence the particles by their fitness values, so multi-objective model of tolerance design can be solved in the improved algorithm. Used to engineering example, only for one running good Pareto optimum set was obtained. Solutions can be selected according to manufacturing reality and market demand. By the analysis of Pareto front, the tolerance design characteristic of this kind of part can be got; the general laws of tolerance design are also validated by the results.Traditional tolerance design includes two sequenced steps which are design tolerance and manufacturing tolerance, Serial design mode of traditional tolerance is changed by concurrent tolerance optimal design, which is virtually hybrid variable combination optimization problem; it was defined as a special kind of TSP, so solving becomes simpler. Making the best of ant colony optimization and particle swarm optimization in solving discrete problem and continuous problem, a hybrid swarm intelligence algorithm of them was presented, which was applied to an example of concurrent tolerance optimization design and satisfied results were gained. The algorithm is high efficient and shows strongly searching ability compared with genetic algorithm and simulation annealing algorithm, It's better when multi-working procedure combination explosion is happened, for it's concurrent characteristic and information social share, and a new way is also given for solving hybrid variable optimization.