中文摘要：

针对标准遗传算法容易出现早熟收敛现象、全局收敛速度慢等问题,提出了一种改进的遗传算法。该算法使用一个助长算子来对种群中的个体进行一定概率下的助长,其遗传个体具有雄性和雌性两种不同的性别,融合了个体间的亲缘关系,异性个体进行严格的远缘繁殖,从而避免了后代个体性能的消极退化,使得算法的全局寻优能力大大增强。将改进的遗传算法应用于配电网故障定位中,并引入分级处理思想,利用配电网呈辐射状的特点,首先把整个配电网划分为主干支路和若干独立区域,再利用该算法分别对各独立区域进行故障定位,然后进行全局寻优,这样能大大减少可行解的维数,提高定位速度。使用该定位方法对一具有20个节点的配电网系统进行故障定位的仿真实验,它使可行解个数由220个减少到144个。结果表明,该定位方法不仅定位准确,而且定位速度快,对复杂配电网的故障定位尤为有效。

英文摘要：

In order to overcome the limitation of standard genetic algorithm such as premature convergence and low local convergence speed, an improved genetic algorithm was proposed. In this algorithm, a help-operator is used to help the individuals of population according to the given probability. The genetic individuals are separated into male individuals and female individuals, and the consanguinity is fused into individuals. Two individuals with opposite sex can reproduce the next generation if they are distant consanguinity individual. So, the negative degeneration of off- spring individual performance is avoided, and the ability of algorithm to search the global optimal is enhanced greatly. The improved genetic algorithm is used in fault location of distribution network. Firstly, the entire distribution network is divided into a main branch and a number of independent regions by making use of the radiation-like characteristic of distribution network and adopting the thought of stage treatment. Then, the improved genetic algorithm is used to locate faults for all independent regions and search for the global optimal. So, the dimension of feasible solution can be reduced largely and the location speed can be enhanced. In the simulation experiments of fault location by this method for a distribution network system with 20 nodes, the number of feasible solutions is reduced from 220 to 144. The experimental results show that the location method is fast and accurate to loeate faults. It is especially effectual in fault location for complicated distribution network.