Estação de baixo custo para monitoramento da qualidade do ar
| Ano de defesa: | 2023 |
|---|---|
| Autor(a) principal: |
Roncaglio, Mateus Maruzka
 |
| Orientador(a): | |
| Banca de defesa: | , , |
| Tipo de documento: | Dissertação |
| Tipo de acesso: | Acesso aberto |
| Idioma: | por |
| Instituição de defesa: |
Universidade Estadual do Oeste do Paraná
Cascavel |
| Programa de Pós-Graduação: |
Programa de Pós-Graduação em Ciência da Computação
|
| Departamento: |
Centro de Ciências Exatas e Tecnológicas
|
| País: |
Brasil
|
| Palavras-chave em Português: | |
| Palavras-chave em Inglês: | |
| Área do conhecimento CNPq: | |
| Link de acesso: | https://tede.unioeste.br/handle/tede/7090 |
Resumo: | Air pollution is one of the main health problems causing various diseases. According to the World Health Organization, about 4 million deaths are due to air pollution every year. For this reason, air quality monitoring is an important management tool that makes it possible to guide public policy in relation to this environmental problem. However, there are few government monitoring stations and their high cost makes investment in new stations to expand the monitoring network unprofitable. Low-cost air quality monitoring sensors can overcome this problem, but also pose new challenges. For this reason, the study of low-cost sensors has been combined with machine learning calibration techniques. This paper presents the development of a low-cost air quality monitoring device using Alphasense sensors. The station records the following measurements using the appropriate sensor models: Carbon Monoxide (CO-B4), Nitrogen Dioxide (NO2-B43F), Sulfur Dioxide (SO2- B4), Ozone (OX-B431) and Particulate Matter (HM3301 and PMS5003). The collected data is sent to the internet in real time via the LoRaWAN protocol. After the field test, data correction was performed using a linear regression model, a random forest algorithm and the conversion model provided by Alphasense. The mean square error (MSE), root mean square error (RMSE), mean absolute error (MAE), linearity coefficient (R2 ) and Pearson r were used as performance metrics to evaluate the models. The results were compared using the proposed device with a commercial station, the Thermo-Scientific GM5000, and a reference station. The preliminary results of the comparison with the GM5000 show that low-cost sensors are susceptible to influences from environmental variables such as humidity. When the same sensors are used in a more controlled environment and the relative humidity does not exceed the operating limits of the sensors, the quality of the readings improves significantly compared to the commercially available GM5000 devices. As for the reference station data, the results show that linear models cannot describe the nuances of the low-cost sensor’s response to the reference gas sensor, while the RF model performs better in every metric. The performance of the RF model shows the potential to improve air quality monitorin |