公开(公告)号：KR1020090019636A

公开(公告)日：2009-02-25

申请号：KR1020070084219

申请日：2007-08-21

Applicant: 재단법인서울대학교산학협력재단

Inventor： 김윤영 ,  정동훈 ,  김대승

IPC: G01N33/48

Abstract: A method for structure design by using multiscale genetic algorithm is provided to efficiently calculate the optimal solution required for the structure design within the shortest time by starting from the initial solution population with the small size, and gradually increasing the number of design variable and the size of solution populations. The method for structure design by using multiscale genetic algorithm comprises the steps of: optionally creating the initial solution population(S310); applying the heredity calculation until the condition of convergence about the entities of the initial solution population is satisfied(S320, S330); producing the initial solution population in which the number of design variables are n times of the design variable of the initial solution population of the former step and the size of the injury population is m times of the initial solution population size of the former step by using the converged solution(S340); applying the heredity calculation until the condition of convergence about the entities of the generated initial solution population is satisfied(S350, S360); and designing the structure by using the solution obtained through repeated performing of the steps in a predetermined times(S370, S380, S390, S400).

Abstract translation: 提供了一种使用多尺度遗传算法进行结构设计的方法，从最小的初始解决方案群体开始，在最短的时间内有效地计算结构设计所需的最优解，并逐渐增加设计变量的数量和大小 的解决方案人口。 通过使用多尺度遗传算法进行结构设计的方法包括以下步骤：可选地创建初始解决方案群体（S310）; 应用遗传算法，直到满足关于初始解群体的实体的收敛条件（S320，S330）; 产生初始解决方案群体，其中设计变量的数量是前一步骤的初始解决方案群体的设计变量的n倍，并且损伤群体的大小是前一步骤的初始解决方案群体大小的m倍，其通过使用 收敛溶液（S340）; 应用遗传算法直到满足生成的初始解群体的实体的收敛条件（S350，S360）; 并且通过使用通过在预定次数中重复执行步骤获得的解决方案来设计结构（S370，S380，S390，S400）。