论文部分内容阅读
当前统计公差分析与综合中常采用遗传算法对装配件中各零件公差进行优化,以平衡产品成本与合格率的比例.利用遗传算法进行优化虽然可以获得全局最优解,但计算量过大.文中提出利用基于数论网格点集的数论序贯优化算法进行公差优化.与MonteCarlo方法产生的随机点集相比,数论网格点集的分布更为均匀.实例计算结果表明:序贯优化算法在保证获得全局最优解的同时,还显著地降低了计算量,计算量为遗传算法的1/5或更少
The current statistical tolerance analysis and synthesis often use genetic algorithms to optimize the tolerances of the parts in the assembly, in order to balance the ratio of product cost and pass rate.Although the global optimal solution can be obtained by genetic algorithm, but the computation is too large. This paper proposes to optimize the tolerance by using number theory sequential optimization algorithm based on number-theoretic grid point set.Compared with the random point set generated by Monte Carlo method, the distribution of number-theoretic grid point set is more uniform.Example calculation results show that the sequential optimization algorithm In addition to guaranteeing the global optimal solution, the computational cost is also reduced significantly, which is one-fifth or less than the genetic algorithm