Smart grid meter analytics for revenue protection

With more and more installed smart meters, there is strong need from utilities to use the fast increasing data for different analytics applications. One of the important applications is revenue protection due to electricity theft, meter installation errors or meter malfunction. A data set with real data of more than 6,000 installed meters is applied in the study for revenue protection. The data is preprocessed to report unusual meter readings by examining unusual long missing/zero data and outliers. The temperature effect is normalized, followed by the algorithm of change detection. Two major change detection methods are implemented and integrated to report out meters with the most suspicious electricity usage patterns with scores.