Correlation-based Bad Data Detection of PMU Measurements for Improving Network Structure with High Penetration of Renewable Energy

In power grids, phasor measurement units (PMUs) are universally installed, while the existence of bad PMU data endangers the monitoring and control of power systems. Firstly, this paper describes the problem of bad PMU data detection. Then, the PMU time-series measurements from two neighboring buses are presented as a 2D diagram, where the spatio-temporal correlation analysis is conducted to solve the problem of recognizing normal data and outliers. The density-based detector DBSCAN is utilized for data clustering and bad PMU data detection. The diagram-based bad data detection method proposed in this paper is unsupervised and used to detect bad PMU data online with short computation time. Numerical and visualized results of case study on the proposed method performance are shown.