Feasibility study of Gusty wind prediction using data mining and regression based on the sum of limited field data

Document Type : Research Article

Author

Assistant Professor, Institute of Geophysics, University of Tehran, Tehran, Iran

Abstract

This research has investigated the possibility of predicting the direction and speed of Gusty wind by using the information of Mehrabad Airport runway automatic station during the period of January 2013 to June 2013, the Metar report of Mehrabad Synoptic Station in the period of 2013, and regression method.
    The data of the automatic station is taken from three sensors located in the band with a length of 4000 meters and a width of 45 meters, in southeast-northwest direction.The sensor number 29 and the sensor number 11, are at the northwestern end of the band and southeast edge of the band respectively. The Mid sensor location is at the middle of the band which distance from the band is 600 meters to the north direction.
    First, all data (except the data of Gusty wind direction and speed measured by the sensors) was normalized to intervals 0.1-0.9. Second, all the data of sensors were randomly divided into three unequal parts: 70% of the data was stored for training, 50% of the remaining data was used for testing and the rest was used for validation. During the calculations, they were used instead of the original data. Third, the quantities were processed by using the three methods of feature selection: Sequential Forward Feature Selection(SFS); Backward(SBS) and Mutual Information(MI) with the method of the Maximum-Relevance and Minimum-Redundancy criterion. At this stage, selective features by every method were separately used in the linear regression method to predict the speed and direction of Gusty wind in the winter and spring seasons. The results were then compared with each other.
    The results show that the selected features by SBS method for wind speed in winter are similar to spring, but their wind direction is slightly different. Selected features for winter Gusty wind with SFS method are a subset of the set of the selected features for spring. Selected features with MI are similar for the two seasons but with different weights.
    The performance of the selected features for wind speed are better than for wind direction. The SFS method is optimal for selecting features of Gusty wind in the Mid runway. On runway 11, the SBS method and the SFS are optimized for predicting the Gusty wind speed and direction respectively. On runway 29, the SBF method is very suitable for selecting features related to Gusty wind speed and direction.
    Finally, by examining the output of the models for each of the runways, an equation is provided to predict the direction and speed of the Gusty wind in each runway.
The quantity of predicted wind direction in runway 29 and 11 depends on the mean wind direction in 2 minutes, the minimum and mean direction in 10 minutes and the wind speed component along the runway, but the quantity of predicted wind speed depends on the maximum wind speed in 10 minutes on the runway, the instantaneous pressure of the station, and the pressure of the station relative to sea level.
    The quantity of predicted wind direction in Mid runway depends on the minimum and mean wind direction in 2 minutes, minimum, mean and maximum wind direction in 10 minutes in runway Mid, maximum wind speed in 2 minutes and components of wind direction along with the runway 11. The quantity of predicted wind speed depends on the maximum and minimum wind speed in 10 minutes on the runway, deviation of wind direction during the last 10 minutes in the runways 29, Mid and 11.
 

Keywords

Main Subjects


حبیبی، آ.، 1395، آموزش کاربردی نرم‌افزار  SPSS، ویرایش چهارم: پایگاه اینترنتی پارس‌مدیر،  200 صفحه.
حبیبی، ف.، 1395، پیش­بینی بارش ایستگاه بندرعباس با استفاده از روش رگرسیون لجستیک: مجموعه مقالات هفدهمین کنفرانس ژئوفیزیک ایران، 13-17.
حبیبی، ف.، 1398، انتخاب ویژگی و پیش­بینی باد گاستی با شبکه عصبی پرسپترون چند لایه­ای در ایستگاه خودکار فرودگاهی: مجله ژئوفیزیک ایران ،13(3)، 33–52.
رضائی یوسفی، م.، 1386، انتخاب متغیرهای ورودی در شناسایی سیستم‌ها و کاربرد آن در پیش‌بینی سری‌های زمانی: پایان نامه کارشناسی ارشد در مهندسی برق-گرایش کنترل، پردیس فنی دانشگاه تهران.
رضایی، م.، حسن زاده، م. ت.، قاسمی، س.، 1387، استفاده از روش­های رگرسیون خطی و رگرسیون غیرخطی در تخمین روند تغییر جریان نشتی مقره­های شبکه توزیع در مناطق آلوده سواحل جنوبی کشور: سیزدهمین کنفرانس شبکه­های توزیع نیروی برق، 11 و 12 اردیبهشت 1387- گیلان.
عرب عامری، م.، حبیبی، ف.، کلهر، ا.، 1393، پیش‌بینی سرعت باد با استفاده از شبکه عصبی پرسپترون چند لایه در فرودگاه مهراباد: مجموعه مقالات شانزدهمین کنفرانس ژئوفیزیک ایران، 117-122.
Agbo, G. A., Alfa, B., Ibeh, G. F., and Adamu, I. S., 2013, Application of regression and multiple correlation analysis to morning hours solar radiation in Lapai: International Journal of Physical Sciences, 8(27), 1437-1441, DOI: 10.5897/IJPS12.323, ISSN 1992-1950 © 2013 Academic Journals, http://www.academicjournals.org/IJPS.
Bluestein, H., 1992, Synoptic-Dynamic Meteorology in Midlatitudes, Volume I: Principles of Kinematics and Dynamics, Oxford University Press.
Fattahi, S., 2011, A comparative study of parametric and nonparametric regressions: Iranian Economic Review, 16(30),19-43.
Gladysz, B., and Kuchta, D., 2008, Application of regression trees in the analysis of electricity load: Badania Operacyjne i Decyzje (Operations Research and Decisions), 4, 19–28.
Han, J., ‏Kamber, M., and Pei, J.‏, 2012, Data Mining: Concepts and Techniques, 3rd edition: Morgan Kaufmann Publishers, ISBN 978-0-12-3814791.
Hazra, A., and Gogtay, N., 2017, Biostatistics series module 10: Brief overview of multivariate methods: Indian Journal of Dermatology, 62(4), 358–366, doi: 10.4103/ijd.IJD_296_17.
Maksood, F. Z., and Achuthan, G., 2016, Analysis of data mining techniques and its applications: Journal of Computer Applications, 140(3), 6-14.
Moustris, K. P., Nastos, P. T., Larissi, I. K., and Paliatsos, A. G., 2012, Application of multiple linear regression models and artificial neural networks on the surface ozone forecast in the greater Athens area, Greece: Advances in Meteorology, Article ID 894714, 8 pages, doi:10.1155/2012/894714.
Murphy, K., 2012, Machine Learning: A Probabilistic Perspective: The MIT Press Cambridge, Massachusetts, ISBN 978-0-262-01802-9.
 
Patlakas, P., Drakaki, E., Galanis, G., Spyrou, C., and Kallos, G., 2017, Wind gust estimation by combining a numerical weather prediction model and statistical post-processing: Energy Procedia, 125, 190-198, www.elsevier.com/locate/procedia.
Prasanna, V., Choi, H. W., Hong S. O. K., Kim, G. H., Lee, Y. G., and Kim, B. J., 2020, Surface wind gust prediction over Incheon international airport using the unified model: Natural Hazards, 103, 1499–1535, DOI:10.1007/s11069-020-04047-z.
Roundy, P. E., and Frank, W. M., 2004, Applications of a multiple linear regression model to the analysis of relationships between eastward- and westward-moving intraseasonal modes: Journal of Atmospheric Science, 61(24), 3041-3048.
Steuer, R., Daub, C. O., Selbig, J., and Kurths, J., 2005, Measuring distances between variables by mutual information, in Baier, D., Wernecke K. D. (eds), Innovations in Classification, Data Science, and Information Systems. Studies in Classification, Data Analysis, and Knowledge Organization. Springer, Berlin, Heidelberg, https://doi.org/10.1007/3-540-26981-9_11.
Vergara J. R., and Este´vez, P. A., 2014, A review of feature selection methods based on mutual information: Neural Computing and Applications, 24(1), 175–186, DOI 10.1007/s00521-013-1368-0.
Ververidis, D., and Kotropoulos, C., 2005, Sequential forward feature selection with low computational cost: 2005 13th European Signal Processing Conference, http://www.eurasip.org/Proceedings/Eusipco/Eusipco2005/defevent/papers/cr1411.pdf.