Particulate Matter Forecasting Using Neural Networks: A Case Study of Delhi

There is a continuous addition of unwanted particles in the environment due to an increase in industrialization and urbanization in metro cities. These unwanted particles contaminate the natural resources and hence lead to environmental pollution. As per WHO 2018, Delhi is the 6th most polluted city in the world. Major sources of pollution in Delhi are various industries and heavy vehicular density. In the present study, an effort has been made to predict PM10 concentration for different hotspot sites. Four Artificial Neural Networks, NARX, RNN, FFBP, ELMAN were used to predict the air pollutant concentration at these hotspots. The results depicted that the NARX network provides better results among the other neural networks. The R and MSE values obtained for the study were 0.94795 and 0.0088474 respectively for PM10. This study will facilitate better prediction models for accurate PM10 concentration.