Detection of Pilferage in an AMI-Enabled Low-Voltage Network Using Energy Reading Anomalies

This paper presents an algorithm that may be utilized by electric utilities to detect pilferers, which are sources of their non-technical losses. The algorithm assumes that the customers are equipped with smart meters, and a check meter that records the total consumption of the customers, and these meters are taking measurements that are coincident in time. The algorithm computes for anomaly coefficients, which are scaling factors that are multiplied to individual measurements when they do not add up to the check meter reading. Using a small test system, the anomaly coefficients computed using the algorithm could reveal the identity of the pilferer, and estimate the amount of pilfered energy.