Optimal tolerance allotment using a genetic algorithm：（使用遗传算法最优公差分配）.pdfVIP

Optimal tolerance allotment using a genetic algorithm and truncated Monte Carlo simulation J Lee and G E Johnson As is typical of stochastic-optimization problems, the multi- yield by modifying the tolerance domain. In problems variate integration of the probability-density function is the with low dimensions, original tolerance domains are most difficult task in the optimal allotment of tolerances. In approximated as simple regions such as spheres, cubics, this paper, a truncated Monte Carlo simulation and a genetic or simplices 3-5. Other approaches for approximating algorithm are used as analysis (i.e.multivariate-integration) and discrete distributions have also been tried 6. synthesis (i.e. optimization) tools, respectively. The new method Because approximate methods change the domains has performed robustly in limited experiments, and was able and/or distributions, the calculated approximate yield is to provide a significant reduction in optimal cost when not the same as the yield for the original domain. As a compared with results published in a previous study. consequence, when approximate methods are used to Keywords: tolerance allotment, tolerance optimization, stochastic solve the tolerance-optimization problem, the reported optimization, Monte Carlo simulation, genetic algorithms yield may deviate significantly from the spec yield~. Since the spec yield is directly related to