基于多代理系统的含DG辐射状配电网故障定位

A Fault Location Algorithm in Radial Distribution Networks with Distributed Generators Based on Multi-Agent Technology

分布式电源(DG)的高渗透接入使配电网结构发生了改变,导致传统的配电网故障定位算法失效。基于含多DG辐射状配电网拓扑结构的特点,构建多代理系统的配电网故障定位框架,该系统主代理根据子代理上传的电源端故障量测量建立RBF神经网络故障测距模型,估计各电源端到故障点的故障距离,考虑上传量测量误差对相邻线路分界点故障定位的影响,应用相邻线路故障信息修正故障线路定位信息;子代理根据主代理下传的故障线路定位结果,利用并行线路首端过流信息校验故障线路。基于MATLAB/SIMULINK搭建32节点配电系统仿真模型,分析不同故障距离、不同故障位置(包含相邻线路分界点附近、分支线路)的故障测距结果和故障定位结果的正确性,结果表明,RBF模型的测距精度高,基于多代理的故障定位方法能够精确定位故障位置,且定位结果不受量测量误差与分支线路的影响。

The accession of large-scale distributed generation changes the distribution network structure,leading to the invalidation of conventional fault location methods. Based on the topology structure of radial distribution networks with distributed generators,a multi-agent architecture for fault location which contains master agent and sub-agents is proposed. First,the master agent builds an RBF neural network model for fault location according to the fault information at the head end of power supply unloaded by the sub-agent,and estimates the distance between fault point and the head end of each power supply. Considering the influence of the uploaded measurement error on the location near the cutoff point between the adjacent lines,a revising approach for the fault section location information by using fault information of the adjacent line is proposed. Then,based on the fault section location result received from master agent,sub-agents use the overcurrent information at the head end of the parallel line to check each fault section. A distribution network model of 32 bus system is built by using MATLAB/SIMULINK system to analyze the effect of different fault distance and position near the cutoff point or on branch line on the fault location results. The results show that the fault location accuracy of the RBF model is high and the proposed fault location method based on RBF neural network can correctly locate the fault position and the fault-locating results are not affected by the measurement error and branch lines.