تشخیص خرابی سازه با استفاده از بسته موجک و الگوریتم جنگل تصادفی در سازه آزمایش شده در مرکز تحقیقات لرزه ای دانشگاه بریتیش کلمبیا

نوع مقاله: مقاله پژوهشی

نویسندگان

1 کارشناسی ارشد سازه، دانشکده فنی مهندسی، دانشگاه تبریز.

2 دانشیار، دانشکده فنی مهندسی، دانشگاه تبریز.

چکیده

امروزه روش شناسایی خرابی بر پایه سیگنال، روشی مهم و پرکاربرد است که بسته موجک یکی از جدید ترین شاخه های این روش هاست. از طرف دیگر، در علم مهندسی ابزار زیادی جهت ارزیابی مدل های تحلیل شده وجود دارد؛ که این ابزار به طبقه بندی و یا رگرسیون داده ها می پردازند. در سال های اخیر روش جدیدی به نام جنگل تصادفی توجه محققین را جلب کرده است. در این مقاله به مدلسازی یک سازه پرداخته شده که جابجایی سازه تحت بوسیله بسته موجک به مولفه های انرژی تجزیه شدند و در کلاس های سه گانه طبقه بندی شدند. در نهایت جنگل تصادفی با استفاده از پایگاه داده ها و خرابی های نزدیک، به حدس دیگر حالات خرابی و طبقه بندی آن ها در کلاس مربوطه پرداخت. دقت جنگل تصادفی در این مقاله 83% بدست آمد که مقداری قابل قبول است ومی تواند در کارهای بعدی مورد استفاده بیشتر قرار بگیرد.

کلیدواژه‌ها


عنوان مقاله [English]

Damage Detection Using Wavelet Packet Decomposition and Random Forests Algorithm in Experimental Structure at the UBC (University of British Columbia)

نویسندگان [English]

  • Omid Habibzadeh Azari 1
  • Hosein Ghaffarzadeh 2
1 Master of Science in Civil Engineering, Tabriz University.
2 Associate Professor, Faculty of Engineering, Tabriz University.
چکیده [English]

Damage Detection methods based on signal are principal and widely used methods that contain wavelet packet decomposition, which is one of new methods in this field. On the other hand there are lots of methods and implements for evaluate models which classify data and regression.Random Forests ,which is newly used method has attracted researchers attention. In this paper a experimental structure was designed and analyzed. drifts from time history response were decomposed to energy rate indexes by wavelet packet decomposition. energy rate indexes in each damage conditions were classified in 3 class of damage conditions and they made data base. finally by training the algorithm, R-F tested other conditions with comparing data base and other near damage conditions, and classified in one of 3 classes. Random Forests precision in this research was 83% which is admissible for classifying. this algorithm can be used on other researches in the future time.

کلیدواژه‌ها [English]

  • random forests
  • wavelet packet decomposition
  • damage detection
  • signal

[1] Shahsavari, H., Baghani, M., Sohrabpour, S., & Naghdabadi, R. (2016). “Continuum damage-healing constitutive modeling for concrete materials through stress spectral decomposition”, International Journal of Damage Mechanics25(6), 900-918.

[2] Jiang, X., & Adeli, H. (2005). “Dynamic wavelet neural network for nonlinear identification of highrise buildings”, Computer-Aided Civil and Infrastructure Engineering20(5), 316-330.

[3] Breiman, L. (2001). “Random forests”, Machine learning45(1), 5-32.

[4] Rodriguez-Galiano, V. F., Ghimire, B., Rogan, J., Chica-Olmo, M., & Rigol-Sanchez, J. P. (2012). “An assessment of the effectiveness of a random forest classifier for land-cover classification”, ISPRS Journal of Photogrammetry and Remote Sensing67, 93-104.

[5] Taha, M. R., Noureldin, A., Lucero, J. L., & Baca, T. J. (2006). “Wavelet transform for structural health monitoring: a compendium of uses and features”, Structural Health Monitoring5(3), 267-295.

[6] Wu, N., & Wang, Q. (2011). “Experimental studies on damage detection of beam structures with wavelet transform”, International Journal of Engineering Science49(3), 253-261.

[7] Yi, T. H., Li, H. N., & Sun, H. M. (2013). “Multi-stage structural damage diagnosis method based on”, Smart Structures and Systems12(3_4), 345-361.

[8] Cruz, P. J., & Salgado, R. (2009). “Performance of vibration-based damage detection methods in bridges”, Computer-Aided Civil and Infrastructure Engineering24(1), 62-79.

[9] Sun, Z., & Chang, C. C. (2002). “Structural damage assessment based on wavelet packet transform”, Journal of structural engineering128(10), 1354-1361.

[10] Lotfollahi-Yaghin, M. A., & Koohdaragh, M. (2011). “Examining the function of wavelet packet transform (WPT) and continues wavelet transform (CWT) in recognizing the crack specification”, KSCE Journal of Civil Engineering15(3), 497-506.

[11] Zhou, Q., Zhou, H., Zhou, Q., Yang, F., & Luo, L. (2014). “Structure damage detection based on random forest recursive feature elimination”, Mechanical Systems and Signal Processing46(1), 82-90.

[12] Chandrashekar, G., & Ferat, S. (2014). “A survey on feature selection methods”, Computers & Electrical Engineering40(1), 16-28.

[13] Misiti, M., Misiti, Y., Oppenheim, G., & Poggi, J. M. (2004). Matlab Wavelet Toolbox User's Guide. Version 3.

[14] Ren, W. X., Sun, Z. S., Xia, Y., Hao, H., & Deeks, A. J. (2008). “Damage identification of shear connectors with wavelet packet energy: laboratory test study”, Journal of structural engineering134(5), 832-841.

[15] Prakash, R., Sivakumar, E., & Srinivasan, M. (2013). “Wavelet packet transform based damage identification of GFRP beam”, Journal of Structural Engineering40(1), 44-47.

[16] Biau, G. (2012). “Analysis of a random forests model”, Journal of Machine Learning Research13(Apr), 1063-1095.

[17] Biau, G., Devroye, L., & Lugosi, G. (2008). “Consistency of random forests and other averaging classifiers”, Journal of Machine Learning Research9(Sep), 2015-2033.

[18] Quinlan, J. R. (1986). “Induction of decision trees”, Machine learning1(1), 81-106.

[19] Dyke, S. J., Bernal, D., Beck, J., & Ventura, C. (2003). “Experimental phase II of the structural health monitoring benchmark problem”, In Proceedings of the 16th ASCE engineering mechanics conference.

CAPTCHA Image