Abstract: The use of traditional artificial intelligence algorithm in distribution network feeder fault location is wide. But some shortcomings such as large initial population, too much iterations and falling into optimum exist. This paper puts forward the estimation of distribution algorithm to realize the fault location, in which the fault section vector is regarded as the right solution. By sampling from the individual probability model established in the solution space, the probability of the optimal fault section vector is gradually improved. The simulation result shows that the application of estimation of distribution algorithm in the fault section location in multiple source ring-open distribution network has faster speed and good fault tolerance.

Key words: fault location, distribution network, estimation of distribution algorithm

摘要： 传统的人工智能算法在配电网馈线故障定位中的应用广泛，存在初始种群规模大，迭代次数多以及易陷入局部最优等缺陷。提出一种基于分布式估计算法的配电网故障区段定位方法，该方法将故障区段向量作为正确解，通过建立解空间内个体分布的概率模型，对模型采样，逐步提高最优故障区段向量在解空间内出现的概率。仿真结果表明将分布估计算法应用于多源开环条件下的配电网故障区段定位有着较快的故障定位速度和良好的容错性。

关键词: 故障定位, 配电网, 分布估计算法