Variables determined for the detection of fraud in companies of electrical service

Abstract

Losses suffered by power utilities represent a problem affecting its financial and economic sustainability. In Venezuela, they reached 30; 68% in 2010. Among them, non-technical losses can be detected and controlled, contributing to the financial recovery of the company, and thus using such surplus of money to improve the quality of service. The aim of this paper is to identify variables used by supply inspection planners in power utilities to allow detection of users that may be using the service under illegal or fraudulent conditions. The methodology used for knowledge acquisition is based upon the principle of the Delphi Method, that is, independent experts in the area answer questionnaires and interviews, variables are identified, and then answers are shared among all of them to select those that receive higher scores on a pre-defined scale. Selection of variables provided by experts is supported by statistical tools. A set of consumption, location and knowledge variables that characterizes fraudulent behavior is then established. They in turn allow construction of a future intelligent system to automate detection of such customers, thus contributing to the reduction of non-technical losses and financial recovery of the company.