Computational Intelligent Approaches for Non-Technical Losses Management of Electricity

This paper presents an intelligent system for the detection of non-technical losses of electrical energy associated with the fraudulent behaviors of system users. This proposal has three stages: a non-supervised clustering of consumption profiles based on a hybrid algorithm between self-organizing maps (SOM) and genetic algorithms (GA). A second stage for demand forecasting is based on ARIMA (autoregressive integrated moving average) models corrected intelligently through neural networks (ANN). The final stage is a classifier based on random forests for fraudulent user detection. The proposed intelligent approach was trained and tested with real data from the Colombian Caribbean region, where the utility reports energy losses of around 18% of the total energy purchased by the company during the five last years. The results show an average overall performance of 82.9% in the detection process of fraudulent users, significantly increasing the effectiveness compared to the approaches (68%) previously applied by the utility in the region.