Análise de dados comerciais para gestão de perdas de água e detecção de fraudes no município de Vitória de Santo Antão

Abstract

Due to the progressive increase in the rate of losses in water supply systems (SAA) in Brazil, apparent losses have become increasingly impacting on the performance and billing of current concessionaires. Apparent losses, in short, encompass the volume consumed, but not recorded by the company providing the service without being billed - caused by reading errors in water meters, fraud, clandestine connections and failures in the commercial register. Thus, it is important to carry out diagnostics in the commercial sector in order to identify, control and reduce such obstacles. Therefore, we seek to analyze the commercial registry database of the city of Vitória de Santo Antão, through the data provided by Companhia Pernambucana de Saneamento (COMPESA), performing a commercial synthesis of the city from the stratification of the available data of the existing connections. In addition, the monthly measurement data and variation of water consumption by connection category, current useful life of the water meter park and consumption ranges in the economies are also evaluated. Therefore, highlighting the fact that frauds are criminal acts and that they must be fought, the importance of analyzing the abnormalities in the connections referring to water consumption is highlighted. Thus, the usability of machine learning algorithms stands out as an alternative for creating models to identify possible fraud arising from water consumption abnormalities.