Eficiência energética em sistemas pressurizados de distribuição de água com o uso de redes neurais artificiais

Abstract

In order to develop new technologies to improve the operation of water supply systems aiming to reduce water and energy waste, it was carried out in this thesis an experimental study using automation and control via Artificial Neural Networks (ANN) with recurrent architecture. The control technique chosen was applied to control the service pressures in an experimental bench that simulates a water supply system capable of supplying water to two regions with different topography. In order to analyze the performance of the system under the controller action, experimental tests were performed simulating a consumption curve for the two supplied regions, introducing external perturbations and changing the setpoint. In addition, considering the use of the control system to reduce the energy consumption of the pumping systems of the experimental bench, the operational optimization was carried out, by reducing the Specific Energy Consumption (SEC) during the control of the service pressures. In this way, it was guaranteed the reduction of the energy consumption of the system at the same time as the pressure control was performed.