Prévision du NO2 en utilisant la méthode du réseau de neurones

Abstract

This article describes a procedure for modeling the phenomenon of dispersion of the of nitrogen dioxide concentration (NO2) using the technique of multilayer perceptron (MLP). Our goal is to prove that the concentration of NO2 is an autoregressive variable explained by meteorological variables. To do this, we proceeded through three stages: in the first stage we used only NO2 concentration variable, in the second stage we used the meteorological variables only, in the third step we used the concentration of NO2 in combination with the meteorological variables. The results showed that the third case is better than the other cases that prove our hypothesis.