Desenvolvimento e avaliação de um equipamento eletrônico para monitorar odor no contexto de Cidades Inteligentes

Abstract

The increasing urbanization and the expansion of industrial and agro-industrial activities have intensified problems related to odor emissions and air quality degradation. The monitoring and identification of these pollutants remain challenging due to the cost of reference equipment and the limited availability of technologies for environmental regulation and control. In this context, this work presents the development and evaluation of an electronic system for odor monitoring and identification. The proposed system consists of a low-cost sensor array, including electrochemical sensors and a Photo Ionization Detector (PID), capable of detecting relevant odorant gases such as hydrogen sulfide (H2S), ammonia (NH3), sulfur dioxide (SO2), methanethiol (CH3SH), and volatile organic compounds (VOC). The device integrates an embedded system responsible for data acquisition, preprocessing, and storage, as well as remote data transmission using the LoRaWANprotocol. The experimental methodology included laboratory tests to evaluate the influence of temperature and humidity on sensor response, odor sample tests for odor profile analysis, experiments with known concentrations of H2S, and field campaigns in different scenarios, including composting environments, urban areas, and a dairy cattle barn. Additionally, the influence of the sampling regime (active and passive flow) was evaluated, highlighting the impact of the air intake system on sensor response dynamics. The results demonstrated that the system is capable of indicatively distinguishing odor profiles. The application of machine learning techniques, such as Kernel PCA and Random Forest, enabled pattern recognition, reinforcing the multivariate analysis approach typical of electronic noses. As an additional contribution, representative datasets of different odor emission scenarios were generated, including an urban dataset in the city of Toledo and an agricultural dataset in a dairy cattle environment. The results indicate that systems based on low-cost sensors, integrated with machine learning techniques and IoT infrastructure, represent a promising alternative for environmental monitoring and odor identification, particularly in distributed applications within the context of smart cities.