Prediction of Ozone Concentration using Feed Forward Back Propagation Neural Network (FFBP-NN)

Abstract

Air pollution has been an ongoing problem in Malaysia. One of the major air quality issue in Malaysia is high concentrations of ozone in urban area. Rapid increase in vehicles number and fossil fuel consumption in Malaysia cause the emission of ozone and their precursors especially nitrogen oxides increasing sharply. This research focus on daytime and nighttime ozone concentration at Kuala Terengganu, Malaysia. The aim of this study is to predict ozone concentration using feed forward back propagation neural network (FFBP-NN) with two hidden layers. Five performance indicators were used to evaluate the models performances which are normalized absolute error (NAE), root mean squared error (RMSE), index of agreement (IA), prediction accuracy (PA) and coefficient of determination (R2). Result show that FFBP-NN with 2 hidden layers model gives good performance for prediction of ozone concentration with high accuracy measures (IA=0.9551, PA=0.8453, R2=0.8402) and small error measures (NAE=0.1642, RMSE=4.4958) for daytime and nighttime (IA=0.9541, PA=0.8429, R2=0.8358, NAE=0.2160, RMSE=3.2485). The result from this study provides a reference for city council to improve the existing guidelines and to plan an effective mitigation measures to monitor the status of air quality towards a sustainable environment.