Machine Learning Model for the Detection of Electric Energy Fraud using an Edge-Fog Computing Architecture

One of the most relevant applications today in the smart grid is the use of artificial intelligence and analysis techniques over the data generated, allowing both end-users and utility companies new applications. Energy fraud is considered one of the main non-technical losses for utilities that translate into economic losses, so being able to detect and minimize fraud becomes a vital necessity for utilities. This paper presents a machine learning model that allows the detection of abnormal behaviors in readings of consumption and/or production of electrical energy that can be classified as energy fraud. The proposed model is based on regression and classification techniques over an edge-fog computing architecture, and the results obtained show that its implementation in smart metering systems is adequate.