混合变量遗传算法在预应力网架结构中的应用

Application of mixing variable genetic algorithm in prestressed space lattice work

针对预应力网架结构设计变量的特点,提出了采用混合变量编码、基于遗传算法的优化设计方法。该方法将杆件截面对应的离散变量和预应力对应的连续变量置于同一水平变量空间,个体编码中同时体现两种变量的遗传基因,避免了割裂设计空间、分级优化而带来的误差影响。为改进标准遗传算法易于发生早熟现象、局部寻优能力差等缺点,优化方法采取了适应度变换、最优保留及自适应遗传等改进措施。算例分析表明,该优化方法合理有效,优化结果满足设计要求,符合工程实践。

Directed at the feature of structural design variable of prestressed space lattice work,the paper presents an optimum design method based on mixing variable code and genetic algorithm.The method consists of putting discrete variable corresponding to cross section and continuous variable corresponding to prestress of member bar in the same level variable space,incarnating genetic gene of two variables in individual codes and avoiding error due to rending design space and cascading optimum.The optimal method ...