基于改进拟态物理学算法的配电网故障区段定位

Fault Location in Distribution Network Based on Improved APO

配电网故障定位的本质是一个离散域二进制寻优问题,因此找到一种全局寻优能力强的二进制算法来解决配电网故障定位是十分困难的。本文针对拟态物理学算法(APO,ArtificialPhysicsOptimization)陷入局部最优的缺点,通过引入反向学习原理改进算法初始解生成过程,并在局部最优时利用混沌无序的特点保持算法的多样性,最后构建了配电网故障定位的数学模型,利用改进后的APO算法对配电网故障进行定位处理。仿真结果表明,采用改进类APO算法进行配电网故障区段定位具有较高容错性,能够实现单点和多点故障的准确定位,通过与遗传算法、蚁群算法比较,本文算法在定位准确和容错性方面有较大优势。

The essence of fault location in distribution network is a binary optimization problem in discrete domain, so it is very difficult to 昀nd a binary algorithm for global optimization to solve the fault location in distribution network. Considering the defects that it can be easy for APO to trap into local optimization, the principle of reverse learning is introduced to improve the initial solution of the algorithm and the characteristics of chaos and disorder is used to keep the diversity of the algorithm in this paper. Finally, the mathematical model of fault location in distribution network is established, and we use the improved APO to locate the fault in distribution network. The simulation results show that the improved APO can be used to locate the fault of the distribution network with a high fault tolerance, especially the accurate location of single point and multi point fault.. The comparison between genetic algorithm(GA)、ACO(ant colony optimization) and improved APO in location shows that the proposed method has advantages in accuracy, stability and high fault tolerance.