Forecasting daily source air quality using multivariate statistical analysis and radial basis function networks.

摘要: It is vital to forecast gas and particle matter concentrations emission rates (GPCER) from livestock production facilities assess the impact of airborne pollutants on human health, ecological environment, global warming. Modeling source air quality a complex process because abundant nonlinear interactions between GPCER other factors. The objective this study was introduce statistical methods radial basis function (RBF) neural network predict daily in Iowa swine deep-pit finishing buildings. results show that four variables (outdoor indoor temperature, animal units, ventilation rates) were identified as relative important model inputs using methods. can be further demonstrated only two factors, environment factor factor, capable explaining more than 94% total variability after performing principal component analysis. introduction fewer uncorrelated would result reduction structure complexity, minimize computation cost, eliminate overfitting problems. obtained RBF prediction good agreement with actual measurements, values correlation coefficient 0.741 0.995 very low systemic performance indexes for all models. indicated could trained these highly relationships. Thus, technology combined multivariate promising tool pollutant emissions modeling.