Forecasting PM2.5 and PM10 Air Quality Index using Artificial Neural Network

Abstract

The Air Quality Index (AQI) was established in response to the Clean Air Act of the Philippines. A complex correlation exists between air pollution levels and exposure, as indicated by the AQI. Manila is one of the cities with severe air pollution-related environmental problems. The impacts of air pollution exposure are detrimental. Therefore, it is essential to forecast air quality indicators and pollution levels to inform the public, particularly sensitive groups, whether or not air quality is safe and healthy. The study’s objective is to construct an air quality forecasting model using an Artificial Neural Network (ANN), which, to date, is the only air quality forecasting model in the Philippines. A feed-forward neural network is utilized to make the model. The PM2.5 and PM10 pollutant concentration time series are provided from the real-time monitoring station in Mehan garden station in Manila. The best forecasting performance of the model was observed with having minimum values of MSE. MAPE and MAE have a value of R2 nearer to 1 from expected and predicted values.