基于高频分量的配电网高阻接地故障识别

High impedance fault detection for distribution networks based on high-frequency components

配电网发生高阻接地故障(HIF)时,接地介质的电阻较大,故障电流比负荷电流要小得多,故障特征不明显。现有的电流或电压保护装置均难以应对HIF,寻找方法对其进行准确快速的辨识意义重大。本文首先通过连续小波变换(CWT)对故障线路的零序电流进行时频域分析,选定特征频段内的时频矩阵作为识别特征量,随后使用改进的LeNet-5网络对识别特征量进行辨识以判定HIF是否发生。在PSCAD/EMTDC中进行了大量的仿真研究,最终结果表明,本文所提方法能够准确识别发生在不同中性点接地系统中的HIF与各类干扰,且抗噪声能力较强。

In case of high impedance fault in distribution network,the resistance of grounding medium is large,the fault current is much smaller than the load current,and the fault characteristics are not obvious.Therefore,the existing current or voltage protection devices are difficult to operate,so it is of great significance to find a method to identify them accurately and quickly.Firstly,the zero sequence current of fault line is analyzed in time-frequency domain by continuous wavelet transform(CWT),and the time-frequency matrix in the characteristic frequency band is selected as the identification characteristic quantity.Then,the improved LeNet-5 network is used to identify the identification characteristic quantity to determine whether HIF occurs.A large number of simulations are carried out on PSCAD/EMTDC.The final results show that the proposed method can accurately identify HIF and various disturbances in different neutral grounding systems,and has strong anti noise ability.