离散变量结构形状优化设计的拟满应力遗传算法

为了弥补遗传算法局部寻优能力差的缺点,将拟满应力算法嵌入到遗传算法中构成一种结合二者优点的混合遗传算法,并将该算法应用于离散变量结构形状优化设计问题.将形状设计变量和截面设计变量统一考虑,编码表示在同一染色体中.既解决了两类变量间耦合上的困难,同时也避免了将两类变量分开考虑只能求得局部最优解的问题.算例的结果表明,该方法用于离散变量结构形状优化是有效的.

An Imitative Full-Stress Genetic Algorithm for Structure Shape Optimization Design with Discrete Variables

In order to cover the shortage of the local searching capability,a new generic algorithm in which the merits were mixed together was put forward by inserting imitative full-stress operator into generic algorithm,and at the same time,the new generic algorithm was used in the solving of the problem of shape optimization of structure with discrete variables.In the paper,the shape design variables and cross-section variables were considered uniformly,encodes shows that they are in the same chromosome.This solves both the problems of the two kinds of variables decoupling and individual solving process for two kinds variables.The result by exemplification shows that the imitative full-stress genetic algorithm was an efficient optimal method for shape optimization of structure with discrete variables.