电压无功优化的强多样性遗传算法

Diversity enhancing genetic algorithm for voltage and reactive power optimization

在求解大规模复杂组合最优化问题的应用中,遗传算法往往因为个体多样性不足存在早熟收敛倾向。引入免疫系统“亲和度”的概念和生物进化过程中的“灾变”概念,提出一种强多样性的遗传算法来克服早熟收敛。首先降低初始群体中各个体之间的亲和度,使得搜索尽可能遍历整个解空间。然后在进化过程基本停滞时采用灾变手段,除当前最好解保留以外,其他个体重新随机产生,以挖掘解空间中当前局部最优解以外的其他更好的解。由于该算法能在群体规模小的情况下充分提高解的多样性,具有很好的全局收敛性能以及较快的收敛速度,能够成功地用于电力系统电压无功优化控制中。

In the application to solving the complicated large-scale combination optimization prob-lems ,GAs usually tend to converge prematurely because of the insufficient diversity of individuals.Diversity Enhancing Genetic Algorithm(DEGA)is presented to prevent premature convergence of GA,which introduced the ideas of affinity from immune system and cataclysm from biologic evolution process.DEGA searches all over the solution space as far as possible by lowering the affinities among the individuals of the initial population.Cataclysm is adopted to explore other better solu-tions than the local optimization in the solution space when the evolution process is halted,by which individuals are generated randomly again besides the best chromosome at present .As the proposed algorithm enhances the diversity of individuals in the solution space with a small size population well,it has a good global convergence performance and faster convergence speed.It has been successfully applied to optimal reactive power and voltage control of power systems.