Prediction of Ambient PM₁₀ and Toxic Metals Using Artificial Neural Networks

Abstract

In this study, an artificial neural network is employed to predict the concentration of ambient respirable particulate matter (PM 10 ) and toxic metals observed in the city of Jaipur, India. A feed-forward network with a back-propagation learning algorithm is used to train the neural network the behavior of the data patterns. The meteorological variables of wind speed, wind direction, relative humidity, temperature, and time are taken as input to the network. The results indicate that the network is able to predict concentrations of PM 10 and toxic metals quite accurately. INTRODUCTIONRecently, artificial-neural-network-based prediction techniques have become more popular because of their ability to generalize the nonlinear patterns in data sets. Their most important advantage is that they can solve problems that are too complex for conventional technologies such as statistical methods. These problems include pattern recognition and forecasting. Their applicability is increasing in airquality predictions because of their ability to handle uncertainties and complex relationships in the data.

05/00450 Correlation of air pollution and meteorological data using neural networks

“…For example, the sine and cosine functions of the wind direction were created, leading to minor improvements of the models. However, the index WD= 1+sin(θ+π/4), suggested by Melas, D., I. Kioutsioukis, and I. Ziomas (2000) and adapted by Chelani, A.B., D.G. Gajghate and M.Z.…”

“…Other advantages include greater fault tolerance, robustness, and adaptability especially compared to expert systems, due to the large number of interconnected processing elements that can be trained to learn new patterns (Lippman, R.P., 1987). These features provide NN the potential to model complex non-linear phenomenon like air pollution (Kolhmainen, M., H. Martikainen and J. Ruuskanen, 2001;Perez, P. and A. Trier, 2001;Chelani, A.B., D.G. Gajghate and M.Z.…”

Analysis of Traffic and Meteorology on Airborne Particulate Matter in Münster, Northwest Germany

“…MLP brought better results than the RBF neural networks and has been proven to be applicable and reliable for predicting air quality applications. 17,18,21 The final network was built by MLP with modification by Broydon-Fletcher-Goldfarb-Shanno (BFGS) with 237 cycles, which is a quasi-Newton, second-order training algorithm accelerating the convergence. 28 Each of the input neurons equaled one input variable.…”

“…Artificial neural network models were generated to predict the PM 10 concentration in Münster. These networks are an alternative to multivariate linear regression, 17 especially if the data cannot be characterized with an approximation of a known theoretical distribution such as the Gaussian normal distribution. 18,19 Originally, the networks are models that try to image the structure and the architecture of exchange of information of human neurons.…”

“…17,20,21 The aim of this work is not only to forecast the PM 10 mass concentration but also to get information on the impact weight of each input parameter such as temperature or traffic.…”