Diagnóstico de perdas comerciais de energia elétrica na distribuição usando redes bayesianas.

Abstract

The following work introduces a methodology for diagnosing non-technical losses by using Bayesian networks and research relating to compliance assessment. This proposed methodology allows estimating the most significant non-technical losses and their spatial distribution, identifying their external or internal causes. The diagnosis makes it possible to evaluate cost-benefit ratios for actions aiming to minimize non-technical losses and, consequently, define an efficient reduction plan. A case study exemplifies each stage of the diagnosis, describing actions to be taken and arranging them according to their economic impact. It also chooses an efficient program for non-technical loss reduction. Brazil, as well as in several other developing countries, presents high non-technical energy loss rates, particularly the distribution companies. An indicator relating non-technical and technical losses is also proposed as a comparative tool for electrical power distribution companies. So the indicator can be useful for controlling the process. It is applied, in this work, to describe the situation of energy losses in Brazilian distribution companies.