Aplicação de algoritmo de aprendizado de máquina para a melhoria de processos em aterros sanitários

Abstract

Solid waste management in Brazil faces significant challenges that demand techno logical innovations for optimization and sustainability. Artificial intelligence emerges as a transformative vector, with the capacity to enhance the efficiency of involved pro cesses. This study aims to develop an artificial intelligence application targeted at the management of sanitary landfills. The research focuses on creating a computational model capable of estimating the progression of the working front using convolutional neural networks, a deep learning modality specialized in computer vision tasks. Aerial imagery collected by Unmanned Aerial Vehicles (UAVs) constitutes the database for training the proposed model. The results demonstrate the neural network's ability to identify surface waste through semantic segmentation, employing classification masks. The measurement of operational expansion is carried out using the XOR logi cal operator. The developed solution promises to contribute to the enhancement of operational control of sanitary landfills, positively impacting the environmental man agement of solid waste.