Search

Air pollution, especially particulate matter (PM) pollution, has a significant impact on India. PM pollution is due to roadside dust, fossil fuel use, vehicular population, and industrial emissions. PM10 forecasting model development is essential because it permits the experts and the citizens to take appropriate actions to restrict their exposure and execute protective measures to improve air quality. This study aimed to develop a specialized computational intelligence methodology that uses principal component (PC) based artificial neural networks (ANN). The model was used to forecast PM10 in ambient air using meteorological data. This application is demonstrated for monitoring data from the urban area of Belagavi city of Karnataka state, India. Principal component analysis (PCA) was applied to understand the interactions between PM10 concentration and meteorological data. The analysis found that the PCANN model is better than the principal component regression (PCR) model, based on using various performance indexes (MAE, MSE, MAPE, RMSE, R, and R2). The PM10 predictive model performance was satisfactory, with a MAPE of 0.069. The overall predictive capability of PM10 was 89.59% in terms of R.

Air pollution, especially particulate matter (PM) pollution, has a significant impact on India. PM pollution is due to roadside dust, fossil fuel use, vehicular population, and industrial emissions. PM10 forecasting model development is essential because it permits the experts and the citizens to take appropriate actions to restrict their exposure and execute protective measures to improve air quality. This study aimed to develop a specialized computational intelligence methodology that uses principal component (PC) based artificial neural networks (ANN). The model was used to forecast PM10 in ambient air using meteorological data. This application is demonstrated for monitoring data from the urban area of Belagavi city of Karnataka state, India. Principal component analysis (PCA) was applied to understand the interactions between PM10 concentration and meteorological data. The analysis found that the PCANN model is better than the principal component regression (PCR) model, based on using various performance indexes (MAE, MSE, MAPE, RMSE, R, and R2). The PM10 predictive model performance was satisfactory, with a MAPE of 0.069. The overall predictive capability of PM10 was 89.59% in terms of R. [ABSTRACT FROM AUTHOR]