Linear and Nonlinear Schemes in the Prediction of Urban NO and NO2 Concentrations

Abstract

ISEE-105 Objective: Because of the harmful effects of the traffic-induced nitric oxide on human health, it is important to have reliable methods enabling the prediction of its concentrations several hours in advance, so that the public authorities could avoid the harmful consequences of severe air pollution episodes. The aim of the paper was to predict NO and NO2 concentrations 4 days in advance comparing 2 artificial intelligence learning methods, namely, multilayer perceptron (MLP) and support vector machines (SVM) on 2 kinds of spatial embedding of the temporal time series. Material and Methods: Hourly values of NO and NO2 concentrations, as well as meteorological variables were recorded in a cross-road monitoring station with heavy traffic in Szeged to build a model for predicting NO and NO2 concentrations several hours in advance. Two effective learning methods, namely, MLP and Support Vector Regression (SVR), were used to provide efficient nonlinear models for NO and NO2 time-series predictions. MLP is widely used to predict these time series, but SVR has not yet been applied for predicting NO and NO2 concentrations. Results: According to the experiments, the applied forecasting techniques can perform well for the prediction of NO and NO2 concentrations. Although MLP improved the results of the best reference algorithms by 5% to 11% for NO prediction, it could not improve those for NO2 prediction. SVR showed 20% to 30% improvement for NO prediction and 2% to 14% for NO2 prediction. Conclusions: Undoubtedly, the application of machine-learning techniques mentioned above can be relatively simple and is worth using. Concerning the NO predictions, the nonlinear learning methods give significantly better predictions than the reference linear methods. In the case of NO2 the improvement of the prediction is considerable; however, it is less notable than for NO.