Detection of electricity theft in low voltage networks using analytics and machine learning

The objective of the work presented in this paper was to identify and implement a machine learning algorithm, to detect electricity theft using smart meter data. Open-source smart meter consumption data for the year 2015 at a granularity of 15 minutes was used to create the model. A cubic support vector machine classification algorithm was used to train the model, with an optimized value of K. Four test sets with different percentages of fraudulent users namely: 10%, 25%, 50%, and 75% were used to test the proposed solution. Evaluation metrics were used to determine the performance of the proposed solution. An average accuracy of 90.6% and a detection rate of 95.75% was achieved.