Uma abordagem baseada em conhecimento para apoio ao combate às perdas comerciais na distribuição de energia elétrica

Abstract

Frauds and measure failures in Power Distribution cause great losses to Power Companies. A traditional way to combat these problems is performing field inspections on customers. However, selecting customers to be inspected is a difficult task for experts on such domain. The goal of the present work is to analyze such knowledge and propose a knowledge based system able to acquire, modify and apply heuristics obtained from the experts on the customer selection process. The system used a real customer database of the local power distribution company for selecting customers for inspection. The evaluation of the results was carried out through two different strategies: executing real field inspections on the selected customers and comparing the results with previously performed field inspections.