Detection and Estimation of the Invisible Units Using Utility Data Based on Random Matrix Theory
share
Invisible units refer mainly to small-scale units that are not monitored, and thus are invisible to utilities and system operators, e.g., small-scale distributed units like unauthorized roof-top photovoltaics (PVs), and plug-and-play units like electric vehicles (EVs). Massive integration of invisible units into power systems could significantly affect the way in which the distribution grid is planned and operated. This paper, based on random matrix theory (RMT), proposes a data-driven approach for the detection, identification, and estimation of the existing invisible units only using easily accessible utility data. The concatenated matrices and linear eigenvalue statistic (LES) indicators are suggested as the main ingredients of this solution. Furthermore, the hypothesis testing is formulated for anomaly detection according to the statistical characteristic of LES indicators. The proposed approach is promising for anomaly detection in a complex grid--it is able to detect invisible power usage, fraud behavior and even to locate the suspect's location. The case studies, using both simulated data and actual data, validate the proposed method.
READ FULL TEXT