Caracterização do Consumo de Água Residencial utilizando Redes Neurais Artificiais

Abstract

: Water leaks in residential pipes are usually noticed only after 30 days, when the consumer receives the bill regarding the use of water and realizes that there was an atypical consumption on site. In such scenario, this article proposes the use of an artificial neural network to propose a methodology for identifying possible water leaks in residential buildings, thus allowing a better understanding and awareness of water consumption in the property. The proposed water monitoring system is based on four main stages: data acquisition; exploration and manipulation of data; data characterization and classification. Through the proposed system, from the classification / forecast of one week, it was possible to identify dates with abnormal consumption, which could be caused by excess water consumption or failures in water lines. dos dados e classificação. Por meio do sistema proposto, a partir da classificação/previsão de uma semana, foi possível identificar datas com consumos anormais, podendo ser causados por excesso de consumo de água ou avarias relacionadas a vazamentos.