Improved Genetic Algorithm for structure shape optimization design of mixed discrete variables

Owing to the weak local searching power and slow iteration in the Genetic Algorithm, some strategies such as optimum individual conservation, infeasible individual transformation and local hesitation strategies are used to impove the Genetic Algorithm, when structure shape optimization design of mixed discrete variables is conducted, if the shape variables are continuous and the cross-section variables are discrete, this two kinds of variables should be considered comprehensively. And two-step searching strategies are put forward. The two-step strategies means that in the early stage of evolution,rough searching is carried through the whole searching range, the searching range is reduced to the adjacency of the optimum solution and then local fine searching is done in the late stage of evolution, thus the searching process is accelerated. The result of the exemplification indicates that the Genetic Algorithm for structure shape optimization design of mixed discrete variables is effective.