基于时延序列特征的多局放源信号分离方法

Multiple partial discharge sources separation method based on time delay sequence characteristics

现有多局部放电(PD)源信号分离方法多采用PDUHF信号的时频差异作为特征参数进行多源信号分离,但在信噪比较低时分离准确率低。为此,文中提出了基于时延序列分布特征的多PD源信号分离方法,采用两支定向天线组成的旋转检测平台分析出时延序列与天线阵列旋转角度满足余弦函数关系,以此对应关系为特征值进行多PD源信号分离。基于时域有限差分算法仿真了3个模拟PD源在不同信噪比时的分离准确率,与现有多源分离方法进行对比,当信噪比为5dB时,分离准确率从71%提升至95%。在220kV试验变电站内试验,结果表明多个PD源被准确地分离和定位,验证了该分离方法的有效性。

The existing multiple partial discharge (PD) sources separation methods mostly adopt the time-frequency difference of PD UHF signal as the feature parameter for multi-sources signal separation, but the separation accuracy rate will be low when the SNR is low. A multi-PD sources separation method based on the characteristics of time delay sequence distribution is proposed in order to improve this situation. The rotation detection platform based on two directional antennas is used to analyze the relationship between the time delay sequence and the rotation angle of the antenna array, the two ones satisfy the cosine function relation, and this corresponding relationship is adopted as the eigenvalue for multi-PD sources separation. The separation accuracy of 3 analog PD sources at different SNRs is simulated based on the time domain finite difference algorithm. Compared with the existing multi-PD sources separation method, when the signal to noise ratio is 5 dB, the separation accuracy is increased from 71% to 95%. Experiments in a 220 kV test substation show that multi-PD sources are accurately separated and located, and the effectiveness of the separation method is verified.