模糊动态罚函数遗传算法及其在钢筋混凝土梁优化设计中的应用

Fuzzy dynamic penalty function GAs and its application to optimum design of reinforced concrete beam

在遗传算法中,约束的处理一般采用罚函数法.罚函数有多种形式,通过对其进行分析研究,本文提出一种新的罚函数,即模糊动态罚函数;用模糊动态罚函数遗传算法对钢筋混凝土梁进行了离散变量优化设计.数值实验表明:由于模糊动态罚函数综合考虑了横向放宽约束和纵向加强约束,该方法能从纵向和横向,即时间和空间的组合领域进行搜索,从而有很好的全局搜索能力,并且搜索速度快,能够迅速收敛到约束最优解.

In the Genetic Algorithm, penalty function is often used to deal with constrains. Penalty function is ot several styles, via analysis and study, a new strategy of improving penalty function, that is fuzzy dynamic penalty function, is presented. Fuzzy dynamic penalty function GAs is used to optimize a reinforced concrete beam with discrete variable. The results show that： due to fuzzy dynamic penalty function allowing for both horizontal broadened constrains and longitudinal strengthened constrains, this method can search for the optimized solution in horizontal and longitudinal, or time and space combined domains. So, this method have a good global convergence, and can converge very .soon to the best solution.