Artificial Intelligence

Detection of Non-Technical Losses

Among its several benefits, the roll-out of smart meters can support the reduction of network losses, including non-technical losses, i.e., energy that is consumed but not measured or billed. Along with the introduction of smart meters, however, energy fraud has evolved by adopting more sophisticated techniques, calling for the development of new fraud detection methods.

In France, Enedis has developed and industrially implemented various innovative fraud detection algorithms based on the analysis of consumption data from smart meters, customer segmentation, cross-referencing with non-energy datasets and clustering. These algorithms have been optimised to reduce false positives, achieving a reliability rate up to 80%.

Related Success Cases