针对配电网现场环境复杂，馈线分接点极多，准确故障定位难度较大的问题，拟从现场实测配电网多维度数据进行分析，探讨配电网系统中的冲击性谐波规律，研究不同时间尺度下各类故障与扰动状态的波形差异，探寻有效波形图像表达方式和恰当的特征提取手段，实现冲击性负荷、谐振和故障之间的区分，提炼典型的配电网拓扑结构，分析增加测点的可能性与最优方案，实现多测点信息聚合条件下的故障区段划分，在配电网系统精细建模的基础上探讨实测数据与仿真波形之间差异的产生原因，考虑高采样率下配电网高次谐波的产生原因，实现负荷、线路等模型的调整和修正，利用确定的线缆接头、分支点、线路首末端点等固定的波阻抗不连续点所决定的行波时序特征来缩短测距区间的方法。进而通过稳健回归算法实现故障精确定位，形成当前亟需的基于数据驱动的不依赖精确模型结构的配电网故障定位体系，为配电网的安全运行提供重要保障。

The distribution system is under the complex electromagnetic circumstances and the number of T-connection points is a great quantity, so the fault location for a distribution system is much more difficult than a transmission line. The measured multi-dimension data and the valid image feature is used to analyze the transient harmonics on distribution system for distinguishing of faults and other transient disturbances. The classic topology of distribution system is extracted to study the traveling wave transmission rule. The possibility and stationing optimization of measurement point number increasing are studied to select fault section on distribution lines based on different measure station. The reason of difference of measured data and simulation data is probed to accomplish the models of load and line. The disconnections of surge impedance on connection of line and cable and so on are used to analyze the features of sequence of arrival time for traveling waves and accurate fault location is performed by robust regression. The proposed fault location system based on data-driven independent on precise system model will support the security and stability of distribution system.