Otimização do sistema de iluminação pública por meio de visão computacional

Abstract

This work demonstrates the optimization of the public lighting system through computer vision, for which pedestrian detection algorithms are used to reduce energy costs in which the current scenario of demand for electricity has rates that constantly increase, due to taxes, urban expansion, among others. Therefore, it is extremely important to look for alternative ways to minimize costs. One of the segments to be explored with great economic potential is the management of public lighting, which in recent times has undergone numerous changes in this area, in which governments are replacing sodium vapor lighting with LED (Light-Emitting Diode) lamps. which are already able to reduce energy consumption. In this context, computer vision systems can help to further reduce this consumption, controlling the power of these LED lamps according to the flow of people on the roads. This work demonstrates the implementation in Python using the OpenCV library, applied to a Raspberry Pi 4. Fuzzy Logic was also used to calculate the power that the Lamps should be adjusted depending on the number of people detected and natural ambient lighting present, adjusting the lighting power adequately ensuring compliance with the NBR 5101 standard. proposed possibilities. With the actual application of this project, a saving of 45% in public lighting consumption was observed, compared to the use of conventional LED lighting.