A Neuro-fuzzy System for Fraud Detection in Electricity Distribution

The volume of energy loss that Brazilian electrical utilities have to deal with has been ever increasing. Electricity distribution companies have suffered significant and increasing losses in the last years, due to theft, measurement errors and other irregularities. Therefore there is a great concern to identify the profile of irregular customers, in order to reduce the volume of such losses. This paper presents a combined approach of a neural networks committee and a neuro-fuzzy hierarchical system intended to increase the level of accuracy in the identification of irregularities among low voltage consumers. The data used to test the proposed system are from Light S.A., the distribution company of Rio de Janeiro. The results obtained presented a significant increase in the identification of irregular customers when compared to the current methodology employed by the company. Keywords— neural nets, hierarchical neuro-fuzzy systems, binary space partition, electricity distribution, fraud detection.