Forecasting Particulate Matter Concentrations: Use of Unorganized Machines

Yara de Souza Tadano, Hugo Valadares Siqueira, Thiago Antonini Alves, Manoel Henrique de Nobrega Marinho

air pollution, neural networks, particulate matter.

Air pollution is an environmental issue studied worldwide, as it has serious impacts on human health. Therefore, forecasting its concentration is of great importance. Then, this study presents an analysis comprising the appliance of Unorganized Machines – Extreme Learning Machines (ELM) and Echo State Networks (ESN) aiming to predict particulate matter with aerodynamic diameter less than 2.5 m (PM2.5) and less than 10 m (PM10). The databases were from Kallio and Vallilla stations in Helsinki, Finland. The computational results showed that the ELM presented best results to PM2.5, while the ESN achieved the best performance to PM10.

10.22161/ijaers.4.4.27