输入阻抗谱分析的电力电缆故障自动测距方法

Automatic Power Cable Fault Ranging Method for Input Impedance Spectrum Analysis

现有的测距方法相对误差大,为此,本文提出了一种输入阻抗谱分析的电力电缆故障自动测距方法。首先,通过测量电缆阻抗谱,诊断电缆运行状态及阻抗幅值频谱极大值附近的过零点相位频谱表征电缆状态,计算电缆阻抗相位谱第一个过零点的频率;其次,单独取样测量每层绝缘介质的介电参数,获取电缆分布电容结果,并在反射波开始前对信号去噪;再次,划分信号时频空间,分解高低频,计算固定阈值,对最优小波包基系数进行软阈值处理,使得故障行波信号去噪后重构;然后,在电缆的一端注入脉冲形成反射脉冲,在不同脉冲之间产生时延,提取信号中的奇异点,将奇异值点带入拟合曲线;最后,计算2个脉冲之间的时间差,代入测距公式得到故障点之间的距离,完成测距。结果表明,实验组相对误差为0%,为6组的最小相对误差,达到了精准测距,提升了测距方法的有效性。

The large relative error of the existing ranging method,therefore,this paper studied the automatic ranging method of power cable fault for input impedance spectrum analysis.Firstly,by measuring the impedance spectrum of the cable,the operating status of the cable and the zero crossing phase spectrum near the maximum value of the impedance amplitude spectrum are diagnosed to characterize the cable status,and the frequency of the first zero crossing point of the cable impedance phase spectrum is calculated.Secondly,separately sample and measure the dielectric parameters of each layer of insulation medium,obtain the results of cable distributed capacitance,and denoise the signal before the reflection wave begins.Once again,divide the signal time-frequency space,decompose high and low frequencies,calculate fixed thresholds,and perform soft threshold processing on the optimal wavelet packet basis coefficients to denoise and reconstruct the fault traveling wave signal.Then,a pulse is injected at one end of the cable to form a reflection pulse,which generates time delay between different pulses,extracts singular points in the signal,and brings singular value points into the fitting curve.Finally,calculate the time difference between two pulses and substitute it into the distance measurement formula to obtain the distance between fault points,completing the distance measurement.The results show that the relative error of the experimental group is 0%,which is the minimum relative error of 6 groups,which achieves accurate ranging and improves the effectiveness of the ranging method.