A Novel Anomaly Localization Method on PMU Measure System Based on LS and PCA

Abstract

Real-time and robust location of anomaly in distribution system plays an important role. However, the data collected by the micro-phasor measurement unit (µPMU) is massive and diversified, and it is urgent to take effective method to excavate the massive data to realize the location of abnormal signals. This paper proposes a novel method to locate the anomaly based on linear shrinkage (LS) and principal component analysis (PCA). The data matrix constructed from µPMU is compressed effectively. By the remained key features, we obtain the corresponding principal component scores. The simulation results have showed that the anomaly can be located exactly according to the scores. Besides, we use LS to avoid the influence of the huge noise. A large number of simulation results show that the method can quickly and accurately realize the anomaly localization in distribution network, and the method has a strong robustness. Even under low SNR environment, the reliability of most abnormal source localization is more than 93%