基于零序电流分布的有源配电网接地故障定位

Grounding Fault Location of Active Distribution Network Based on Zero Sequence Current Distribution

有源配电网单相接地故障特征弱、特性复杂,更容易导致测量点采集到的故障信息发生畸变,基于就地或局部特征进行故障定位易受信息畸变的影响,导致定位困难。借鉴智能算法在配电网相间短路故障定位中的应用,提出了基于暂态零序电流广域分布的有源配电网接地故障定位方法,受信息畸变影响小且定位准确。根据有源配电网发生接地故障时各个馈线测量点采集到的零序电流信号,利用闭合打开差运算(closing opening difference operation,CODO)提取暂态零序电流方向。基于暂态零序电流的分布规律并应用图论中的最短路理论确定疑似故障区域,减小用于故障定位解空间的规模,并提出了一个新的适应度函数进行故障定位,使其对畸变信息更敏感。在解空间中采用改进二进制海洋捕食者算法(improvedbinarymarinepredators algorithm,IBMPA)搜索故障馈线,通过混沌映射、自适应t分布以及分组学习策略提高其跳出局部最优的能力以及全局搜索的速度,实现了准确故障定位。最后在IEEE119有源配电网上进行仿真测试,结果表明该方法能够有效提升配电网接地故障定位准确性和速度,同时对于畸变信息有着较高的容错性。

The single-phase grounding faults of active distribution network are usually weak and complicated,easily contributing to the distortion of fault information collected at the measurement point.Therefore,it is difficult to position a fault locally.Taking into consideration the application of intelligent algorithms in the location of inter-phases short circuit faults in distribution network,a grounding fault location method of active distribution network based on wide area distribution of transient zero-sequence current is proposed,which is later proved less affected by information distortion and has more accurate location.The transient zero sequence current direction is extracted by the Closing Opening Difference Operation(CODO)calculation based on the zero sequence current signals collected at the feeder measurement points when there is a ground fault in the active distribution network.Given the distribution characteristics of transient zero sequence current and the application of shortest path theory in the graph theory,the suspected fault area is determined,reducing the size of solution space for fault location.Then a new fitness function for fault location is proposed,which makes the proposed method more sensitive to the distorted information.An improved binary ocean predator algorithm(IBMPA)is used to find out the fault feeders in the solution space,and the chaotic mapping,adaptive t-distribution,and group learning strategies are used to improve its ability to jump out of the local optimization and the global searching speed,succeeding in accurate fault location.Finally,simulation tests are carried out on the IEEE119 active distribution network.The results indicate that this method improves the accuracy and speed of grounding fault location in the distribution network effectively.Meanwhile,it has a higher fault tolerance for the distortion information.