Detecting and Estimating Amount of Energy Theft in the Distribution Network Using Linear Regression

Non-technical losses (NTL) in the distribution network has been an issue for a lot of electric utilities for a long time. Recent developments for NTL detection had focused on using energy meter consumption for pattern or anomaly detection in identifying suspicious consumption behavior. In this study, meter bypass and unmetered load tapping were modeled in OpenDSS to simulate the effect of NTL in the low voltage network. With the assumption that smart meter data from houses and check meter data from a distribution transformer is known, the voltage and power readings were used to build a linear regression model for estimating power consumption of each customer. A significant mismatch between the estimate and actual meter reading had been found to correspond to an anomalous consumption in the network. The algorithm had been able to identify the locations for meter bypass and narrow down the area for possible inspection for unmetered load connection.