Wavelet-Networks for Prediction of Ozone Levels in Puebla City Mexico

Wavelet-networks are inspired by both the feed forward neural networks and the theory underlying wavelet decompositions. This special kind of networks has proved its advantages over other networks schemes, particularly in approximation and prediction problems. In this paper a novel approach, based on a wavelet neural network structure with correlation-based initialisation and training algorithm, is introduced in order to face with the problem of pollutant estimation in a metropolitan area. In particular a short-term prediction of the maximum ozone pollutant value has been performed. Ozone gas is considered one of the most common and damaging air contaminants. The results reported in this work show clearly that wavelet networks have good prediction properties and seriously represent a novel alternative to the traditional ozone forecasting methods