Prediction of Ambient Air Quality Based on Neural Network Technique

Air quality index (AQI) system lays an important role in conveying to both decision-makers and the general public the status of ambient air quality, ranging from good to hazardous. Five types of air pollutants will be studied which consists of ozone (O₃), carbon monoxide (CO), nitrogen dioxide (NO₂), sulphur dioxide (SO₂) and suspended particulate matter less than 10 micron in size (PM₁₀). The objective of this paper were to investigate the effectiveness of Artificial Neural Network (ANN) model with Back Propagation Neural Network (BPNN) for predicting the ambient air quality for air quality monitoring in states of Malaysia. The measurement activities are carried at Jalan Tasek in Perak, Nilai in Negeri Sembilan and Jerantut in Pahang. The data collected comprises of data for the previous two months, beginning from November 2004. The ambient air quality plays an important role in evaluating the air quality. The artificial neural network simplifies and speeds up the computation of the ambient air quality, as compared to the currently existing method. For this purposes, neural network model provides an interesting alternative to air quality monitoring. The comparison between data from model predictions and actual observations is coherent which shows that promising result based on the developed ANN model in predicting ambient air quality (AAQ) is effective and accurate.