Electricity theft detection in smart grid using random matrix theory

Illegal use of electricity has been a major concern in power system industries for a long time. Fraudulent large-scale consumption of electricity may result in an unbalanced demand-supply gap. This study proposed a data-driven electricity theft detector that is based on random matrix theory with the widespread use of smart meters and advanced metering infrastructure. The application of an augmented matrix as the data source is the key step of the proposed method, indicating the correlations between power consumption and system operating states under abnormal conditions of electricity use. Moreover, the regional/individual theft detection algorithm and the abnormal consumption pattern signal are designed for shortlisting regions with a high probability of theft and selecting suspected fraudulent customers in real time. Numerical case studies using simulated case studies conducted on the IEEE 34-bus system and real data collected from the power system of a Chinese city are used to investigate the correctness and feasibility of the proposed method.