Artificial Inteligence Based IoT Systems and Data Processing for Monitoring Flow in a Water Supply Network

The technological development of water distribution systems is essential to improve the population's living conditions, as well as improving management within companies in the sanitation sector. In this context, seeking efficient ways of monitoring and forecasting demand constitute promising alternatives to detect hydroenergy losses and reduce system operating costs. In this work, a methodology based on Typicality and Eccentricity Data Analytics and artificial neural networks is proposed for use in an experimental monitoring node, to estimate, indirectly, the water flow in a laboratory water distribution system. As a result, proposed system was able to forecast future flows with MAPE under 5% and MSE under 0.0003.