Abstract

Air quality forecasting has acquired great significance in environmental sciences due to its adverse affects on humans and the environment. The artificial neural network is one of the most common soft computing techniques that can be applied for modeling such complex problem. This study designed air quality forecasting model using three-layer FFNN's and recurrent Elman network to forecast PM10 air pollutant concentrations 1 day advance in Yilan County, Taiwan. Then, the optimal model is selected based on testing performance measurements (RMSE, MAE, r, IA and VAF) and learning time. This study used an hourly historical data set from 1/1/2009-31/12/2011 collected by Dongshan station. The data was entirely pre-processed and cleared form missing and outlier values then transformed into daily average values. The final results showed that the three-layer FFNN with One Step Secant (OSS) training algorithm achieved better results than Elman network with Gradient Descent adaptive learning rate (GDX) training algorithm. Where, the FFNN required the less training time and achieved better performance in forecasting PM10 concentrations. Also, the testing performance measurements shown that the selected daily average input variables in previous day (PM_2.5), relative humidity, PM10, temperature, wind direction and speed is critical to give better forecasting accuracy. Whereas, the testing measurements RMSE = 6.23 μg/m₃, MAE = 4.75 μg/m₃, r = 0.943, IA = 0.964 and VAF = 88.80 in PM₁₀ FFNN forecasting model that used OSS training algorithm.

Keywords:

Forecasting, neural networks, soft computing,

References

1. Brunekreef, B. and S.T. Holgate, 2002. Air pollution and health. Lancet, 360: 1233-1242.
   CrossRef    
2. Brunelli, U., V. Piazza, L. Pignato, F. Sorbello and S. Vita�bile, 2007. Two days ahead pre�diction of daily maxi�mum concentrations of SO2, O3, PM10, NO2, CO in the urban area of Pa�l�ermo, Italy. Atmos. Environ., 41(14): 2967-2995.
   CrossRef    
3. Chaloulakou, A., G. Grivas and N. Spyrellis, 2003. Neural network and multiple regression models for PM10 prediction in Athens: A comparative assessment. J. Air Waste Manage. Assoc., 53(10): 1183-1190.
   CrossRef    
4. Chan, L.Y., W.S. Kwok, S.C. Lee and C.Y. Chan, 2001. Spatial variation of mass concen�tration of roadside sus�pended particulate matter in metropolitan Hong Kong. At�mos. Envi�ron., 35(18): 3167-3176.
   
5. Chang, S.C., T.Y. Pai, H.H. Ho, H.G. Leu and Y.R. Shieh, 2007. Evaluating Taiwan's air quality variation trends using grey system theory. J. Chinese Inst. Eng., 30(2): 361-367.
   CrossRef    
6. Corani, G., 2005. Air quality prediction in Milan: Feed-forward neural networks, pruned neural networks and lazy learning. Ecol. Model., 185(2): 513-529.
   CrossRef    
7. Cs�ji, B.C., 2001. Approximation with artificial neural networks. M.S. Thesis, Faculty of Sci�ences, E�tv�s Lor�nd University, Budapest, Hungary, pp: 45.
   
8. Demuth, H., M. Beale and M. Hagan, 2008. Neural Net�work Toolbox 6TM User's Guide. Ver�sion 6, The Math Works Inc. (Accessed on: 25th December, 2012).
   Direct Link
9. Elman, J.L., 1990. Finding structure in time. Cognitive Sci., 14(2): 179-211.
   CrossRef    
10. Hagan, M.T., H.B. Demuth and M. Beale, 1996. Neural Network Design. PWS Publishing Co., a Division of Thomson Learning. Boston, U.S.A.
    
11. Hornik, K., M. Stinchcombe and H. White, 1989. Multi-layer feed forward networks are uni�versal approximates. Neural Networks, 2(5): 359-366.
    CrossRef    
12. Jain, A.K., J. Mao and K.M. Mohiuddin, 1996. Artificial neural networks: A tutorial. IEEE Comput., 29(3): 31-44.
    CrossRef    
13. Kaastra, I. and M. Boyd, 1996. Designing a neural network for forecasting financial and eco�nomic time series. Neuro-Comput., 10(3): 215-236.
    
14. Kao, J.J. and S.S. Huang, 2000. Forecasts using neural network versus Box-Jenkins method�ology for ambient air quality monitoring data. J. Air Waste Man�age. Asso�c., 50(2): 219-226.
    
15. Krenker, A., J. Be�ter and A. Kos, 2011. Introduction to the Artificial Neural Net Works. In: Suzuki, K. (Ed.), Artificial Neural Net�works-methodological Advances and Biomedical Applica�tions. In Tech, India, pp: 3-18.
    CrossRef    
16. Kr�se, B. and P. Van der Smagt, 1996. An Introduction to Neural Networks. 8th Edn., Univer�sity of Amsterdam, Netherlands.
    PMid:9981955    
17. Kunzli, N., R. Kaiser, S. Medina, M. Studnicka, O. Chanel, P. Filliger, M. Herry, F.H. Jr, V. Puybonnieux-Texier, P. Qu�nel, J. Schneider, R. Seethaler, J.C. Vergnaud and H. Sommer, 2000. Public-health impact of outdoor and traffic-related air pollution: A European assess�ment. Lancet, 356(9232): 795-801.
    CrossRef    
18. Kurt, A. and A.B. Oktay, 2010. Forecasting air pollutant indicator levels with geographic models 3 days in advance using neural networks. Expert Syst. Appl., 37(12): 7986-7992.
    CrossRef    
19. Kurt, A., B. Gulbagci, F. Karaca and O. Alagha, 2008. An online air pollution forecasting system using neural net�works. Environ. Int., 34(5): 592-598.
    CrossRef    PMid:18237781    
20. Lapedes, A. and R. Farber, 1988. How Neural Nets Work. In: Anderson, D.Z. (Ed.), Neural Information Processing Systems. American Institute of Physics, New York, pp: 442-456.
    
21. Leung, Y.K. and C.Y. Lam, 2008. Visibility impairment in Hong Kong-a wind attribution analysis. Bull. Hong Kong Meteorol. Soc., 18: 33-48.
    
22. Ma, J., 2005. Application of neural network approach to acid deposition prediction. M.S. The�sis, Environmental System Engineering, University of Regina, Regina, pp: 112.
    
23. Maraziotis, E., L. Sarotis, C. Marazioti and P. Marazioti, 2008. Statistical analysis of inhal�able (PM10) and fine par�ticles (PM2.5) concentrations in urban region of Patras Greece. Global NEST J., 10(2): 123-131.
    
24. May, R., G. Dandy and H. Maier, 2011. Review of Input Variable Selection Methods for Artifi�cial Neural Networks. In: Suzuki, K. (Ed.), Artificial Neural Networks-methodological Advances and Biomedical Applications. In Tech, India, pp: 3-18.
    CrossRef    PMCid:PMC3049272    
25. M�ller, M.F., 1993. A scaled conjugate gradient algo-rithm for fast supervised learn�ing. Neural Networks, 6(4): 525-533.
    CrossRef    
26. Nikov, A., F. Karaca, O. Alagha, A. Kurt and H. Hakkoymaz, 2005. Air poltool: A web based tool for Istanbul air pol�lution forecasting and control. Proceeding of 3rd International Symposium on Air Quality Management at Urban Regional and Global Scales. Istanbul, 1: 247-255l.
    
27. Niska, H., T. Hiltunen, A. Karppinen, J. Ruuskanen and M. Kolehmaxinen, 2004. Evolving the neural network model for forecasting air pollution time series. Eng. Appl. Artif. Intell., 17(2): 159-167.
    CrossRef    
28. Nygren, K., 2004. Stock prediction-a neural network ap�proach. M.S. Thesis, Royal Institute of Technology, KTH, Stockholm.
    
29. Peirce, J.J., R.F. Weiner and P.A. Vesilind, 1998. Meteor�ology and Air Pollution. In: Envi�ronmental Pollution and Control, 4th Edn., Butterworth-Heinemann, Boston, pp: 271-286.
    CrossRef    
30. Perez, P. and J. Reyes, 2006. An integrated neural network model for PM10 forecast�ing. Atmos. Environ., 40(16), 2845-2851.
    CrossRef    
31. Petchsuwan, T., 2000. A neural network approach for the prediction of ambient SO2 concen�trations. M.S. Thesis, En�vironmental System Engineering, University of Regina, Can�ada, pp: 188.
    
32. Riedmiller, M. and H. Braun, 1993. A direct adaptive me�thod for faster back propagation learn�ing: The RPRO Palgo�rithm. Proceedings of IEEE International Confe�rence on Neural Networks (ICNN 93), pp: 586-591.
    
33. Roy, S., 2012. Prediction of particulate matter concentra�tions using artificial neural net�work. Resour. Envi�ron., 2(2): 30-36.
    
34. Rumelhart, D.E., G.E. Hintont and R.J. Williams, 1986. Learning representations by back-propagating er-rors. Nature, 323: 533-536.
    CrossRef    
35. Saffarinia, G. and S. Odat, 2008. Time series analysis of air pollution in Al-Hashimeya town Zarqa, Jordan. Jordan J. Earth Environ. Sci. (JJEES), 1(2): 63-72.
    
36. Srinivasan, D., A.C. Liew and C.S. Chang, 1994. A neural network short-term load fore�cas�ter. Elect. Power Syst. Res., 28: 227-234.
    CrossRef    
37. Taiwan Air Quality Monitoring Network (TAQMN), 2012. Taiwan En�vironmental Protection Administra�tion (TEPA). (Accessed on: 23th Decem�ber 2012).
    Direct Link
38. Ul-Saufie, A.Z., A.S. Yahya, N.A. Ramli and H.A. Hamid, 2011. Comparison between multiple linear regression and feed forward back propagation neural network models for pre�dicting PM10 concentration level based on gaseous and meteorological parame�ters. Int. J. Ap�pl., 1(4): 42-49.
    
39. USEPA, 2003. Guidelines for Developing an Air Quality (Ozone and PM2.5).Forecasting Program, U.S. Environmental Pro�tection Agency Office of Air Quality Planning and Standards Infor�mation Transfer and Program Integration Di�vision AIRNow Program Research Triangle Park. North Car�olina, EPA-456/R-03-002.2.
    
40. WHO, 2005. Air Quality Guidelines for Particulate Matter, Ozone, Nitrogen Dioxide and Sulfur Diox-ide. Global Update 2005, Summary of Risk Assessment (2005). World Health Organi�zation (WHO), Geneva.
    
41. Willmott, C.J., 1982. Some comments on the evaluation of model performance. Bull. Am. Meteorol. Soc., 63: 1309-1369.
    CrossRef    
42. Yang, K.L., 2002. Spatial and seasonal variation of PM10 mass concentrations in Tai�wan. Atmos. Environ., 36(21): 3403-3411.
    CrossRef    
43. Yildirim, Y. and M. Bayramoglu, 2006. Adaptive neuro-fuzzy based modeling for prediction of air pollution daily levels in city of Zonguldak. Chemosphere, 63(9): 1575-1582.
    CrossRef    PMid:16310825    
44. Yilmaz, I., N.Y. Erik and O. Kaynar, 2010. Different types of learning algorithms of Artificial Neural Network (ANN) models for prediction of gross calorific value (GCV) of coals. Acad. J., 5(16): 2242-2249.
    
45. Zadeh, L.A., 1994. Fuzzy logic, neural networks and soft computing. Commun. ACM, 37(3): 77-84.
    CrossRef    
46. Zhang, G., B.E. Patuwo and M.Y. Hu, 1998. Forecasting with arti?cial neural networks: The state of the art. Int. J. Forecasting, 14(1): 35-62.
    CrossRef    

Competing interests

The authors have no competing interests.

Open Access Policy

This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.

Copyright

The authors have no competing interests.