Assignment of Customers to Registered Meters for Energy Theft Detection

Abstract

Power utilities are constantly seeking strategies to reduce electrical energy losses, technical and non-technical. Technical losses are intrinsic to the network operation, and non-technical losses are all the remaining losses. Several methods focus on mitigating non-technical losses, but most of them are costly or ineffective. This paper proposes a robust and efficient methodology to detect energy theft based on the assignment of the coordinates of customer units to the respective energy meters registered at the distribution utility. Two datasets are used to locate the centroid of the lots (polygons) in a city, representing buildings detected through artificial intelligence in orthoimages. The resulting lots are then associated with utility energy meters. Lots left with no associated meter are marked as potentially illegal consumption. The method is validated with real and updated data (from 2022) and results indicate that it can effectively detect possible energy theft locations, increasing the effectiveness of utility inspections.