离散变量结构优化设计的复合形遗传算法

对离散复合形法提出了一种新的初始点产生办法,并基于满应力思想,对离散复合形法的优化结果进行进一步搜索,提高了离散复合形法的局部寻优能力·为了弥补遗传算法自身的不足,把改进的复合形算法作为复合形算子嵌入到遗传算法中,以提高遗传算法的局部寻优能力;同时对遗传操作过程做了改进,如在进化初期采用大的交叉率,以尽快筛选出最优个体;对最差个体采用大的变异率,使其向最优解逼近,从而建立了一种离散变量结构优化设计的混合遗传算法·算例表明这种混合遗传算法优于基本遗传算法和改进的复合形法,是可行和有效的·

Genetic Algorithm in Compound Form for Structural Optimization with Discrete Variables

A new approach to the generation of an initial point is proposed for discrete compound form. In view of the theory of imitative full-stress, the optimization results of discrete compound form are searched further, so as to improve fully the local searching capability of discrete compound form. On the other hand, a compound form operator is imbedded in genetic algorithm so as to improve the local searching capability of genetic algorithm to make up for the shortage of genetic algorithm. Moreover, the operation of standard genetic algorithm is also improved, e.g., a high cross rate shall be taken in the early stage of evolution to pick out the best individual as fast as possible with a high mutation rate taken for the worst individual to approach to optimal solution. A hybrid genetic algorithm is thus set up for structural optimization with discrete variables. The results by exemplification show that the hybrid genetic algorithm, as an efficient optimal method having the advantages of both standard genetic algorithm and improved compound form approach without their disadvantages, is superior to either one.