NEURAL NETWORK PERFORMANCE COMPARE TO EULARIAN DISPERSION MODEL FOR CO FORECASTING

• Main Authors: Y. H. , Noorfidza, T. S. Y. , Colleen, M . Z. O. , Zakaria, S, Yusup, A. S., Azmi, H., Zabiri
• Format: Conference or Workshop Item
• Published: 2007
• Subjects: TP Chemical technology
• Online Access: http://eprints.utp.edu.my/3760/1/ECOMOD2007a.pdf; http://eprints.utp.edu.my/3760/

Numerical model, such as Eulerian Dispersion Model is principally used in a deterministic air quality model to forecast pollutant concentrations. Kv, vertical diffusity coefficient is identified by using this functional model. Here, users of the model have to rely on some sort of sensitivity analysis to decide the suit value to their needs. Artificial Neural Network is proposed can act as universal approximations of non-linear functions and consequently can be used in assessing the dynamics of complex non-linear systems such as atmosphere. In this study carbon monoxide concentrations were forecasted utilizing both the neural network and Eulerian dispersion model. The predictions were performed over Jalan Tasek Ipoh using meteorological data from the year 2001 to 2005. Results show that the neural network produces more accurate forecasts, has greater flexibility and would thus be extremely useful in air pollution control.