基于频域阈值处理广义互相关和空间筛选的多源局部放电定位方法

Multi-source Partial Discharge Localization Based on Frequency Domain Threshold Processing Generalized Cross-correlation and Spatial Screening

该文针对电力设备的多源局部放电定位问题,提出了基于频域阈值处理广义互相关(frequency domain threshold processing generalized cross-correlation,FDTP-GCC)和空间筛选的多源局部放电定位方法。为实现多源局部放电超声信号的同源时间差提取问题,该文提出了FDTP-GCC算法,该算法直接应用于超声传感设备采集的原始信号,在频域内通过设置幅度谱阈值排除干扰频段,幅度谱归一化突出多源信号时间差的冲激函数特征,直接得到对应同源信号的时间差。通过仿真证明了FDTP-GCC算法进行时间差估计的高准确性、高稳定性、高抗噪能力。为解决FDTP-GCC算法带来的同源时间差分离的问题,该文提出了基于空间筛选的多源局部放电定位方法,并分析了采用基于解析解的算法和智能算法进行空间筛选的定位性能。最后采用双局部放电源进行了定位试验,试验结果表明该文方法提取的时间差误差均在1μs以内,两个局部放电源的定位误差为20.53cm和4.93cm。

In this paper,a method of multi-source partial discharge localization based on the frequency domain threshold processing generalized cross-correlation(FDTP-GCC)and the spatial screening for the multi-source partial discharge localization problem of the power equipment is proposed.In order to realize the same-source time difference extraction of multi-source partial discharge ultrasonic signals,the FDTP-GCC algorithm is suggested in this paper.This algorithm is directly applied to the original signals collected by ultrasonic sensing devices,excluding the interfering frequency bands in the frequency domain by setting the amplitude spectrum threshold.Then the amplitude spectrum is normalized to highlight the impulse function characteristics of the multi-source signal time difference,directly obtaining the time difference of the corresponding same-source signals.The simulation tests verify the high accuracy,stability and noise immunity of the FDTP-GCC algorithm for the time difference estimation.In order to solve the problem of the same-source time difference separation brought by the FDTP-GCC algorithm,this paper puts forward a multi-source partial discharge localization based on the spatial screening and analyzes the localization performance of the spatial screening using the analytical solution-based algorithm and the intelligent algorithm.Finally,a localization test is conducted using the dual partial discharge sources,and the experimental results show that the time difference errors extracted by the method in this paper are all within 1μs,and the localization errors of the two partial discharge sources are 20.53 cm and 4.93 cm.