Detection of Electricity Theft in Customer Consumption Using Outlier Detection Algorithms

Advanced Metering Infrastructure (AMI) is a core part of Smart-grid, which is responsible for collecting, measuring and analyzing energy usage data of customers. The development of this network has been possible thanks to the emergence of new information and communication technologies. However, with the arrival of these technologies, new problems have arisen in the AMI. One of these challenges is the energy theft, which has been a major concern in traditional power systems worldwide. To face these challenges, datasets of electricity consumptions are analyzed to detect intruders. Traditional techniques to detect intruders include the use of machine learning and data mining approaches. In this paper, we analyze the feasibility of applying outliers detection algorithms for enhancing the security of AMI through of the detection of electricity theft. We explore the performances of various existing outlier detection algorithms on a real dataset (consumer energy usage). The results show the feasibility of use outliers algorithms in the security of AMI and also the effectiveness of the use of these methods in the electricity consumption datasets for theft detection.