Energy theft detection in advanced metering infrastructure

Abstract

Energy theft is one of the key concern in Advanced Metering Infrastructure (AMI). In developed and developing countries, the financial losses due to energy theft are billions of dollar per year. In this paper, we have proposed a Principal Component based Theft Detection scheme to detect energy theft in AMI. Principal components have been found using consumers' consumption data. Mahalanobis distance is calculated between transformed testing samples and historical consumption data. If the Mahalanobis distance is outside the predefined threshold range, the testing sample is considered as malicious. The proposed method is tested under different attack scenarios using real smart meter data. The experimental results show that the proposed scheme detects energy theft attacks with high detection rate.