Detection of Electricity Theft based on Compressed Sensing

Electricity Theft, a major non-technical loss during transmission and distribution systems, is a cause of serious concern in developing countries. Detecting and dealing with energy thefts have become a challenging task for the utility companies. Electricity thefts impair the power supply quality, increase the load on generating stations and impact the tariff. This paper proposes methods for detection of electricity thefts based on compressed sensing and sparse representation techniques. Compressed sensing is a promising signal processing technique which is used to accurately reconstruct signals and information that are sparse, from small number of random measurements. Since electricity thefts are infrequent, the difference between power consumed actually and the power measured in the meters results in a set of equations which has sparse solutions. This sparse structure enables to detect electricity thefts with minimal number of sensors.