中文摘要：

在研究已有小电流接地系统单相接地故障定位方法优缺点的基础上，提出了一种小波优化神经网络的故障定位算法。首先在分析单相接地故障暂态电气量特征的基础上，利用小波奇异性检测进行特征提取。然后构建故障定位的优化神经网络模型，以遗传算法良好的全局寻优能力为基础，根据神经网络良好的非线性拟合能力，建立起故障特征与故障点位置之间的映射，实现故障定位。最后仿真测试表明，故障定位的相对误差不超过1．5％，并且不受故障点位置、故障点电阻和相角的影响。

英文摘要：

After researching the advantages and disadvantages of existed methods for the fault line location in small current grounded system, this paper presents a method of fault location by combining wavelet transformation and neural network opti- mized by genetic algorithm. Based on the analysis of the single-phase ground of the system, the features are extracted accord- ing to singularity detection by using wavelet transform modulus maximum. And then optimized network model of fault location is constructed. Genetic algorithm has good global search ability. The capacity of non-linear fitting of neural network is well. These studies establish correspondence between features and fault location, and realize the fault location. The simulation result shows the fault location＇s relatively error is less than 1.5%, and it is independent of fault distance, power supply phase angle and transient resistance.