Volume 15, Issue 5 (2015)                   Modares Mechanical Engineering 2015, 15(5): 41-48 | Back to browse issues page

XML Persian Abstract Print


Download citation:
BibTeX | RIS | EndNote | Medlars | ProCite | Reference Manager | RefWorks
Send citation to:

Ziaiefar H, Amiryan M, Ghodsi M, Honarvar F, Hojjat Y. Ultrasonic Damage Classification in pipes and plates using Wavelet Transform and SVM. Modares Mechanical Engineering. 2015; 15 (5) :41-48
URL: http://journals.modares.ac.ir/article-15-3906-en.html
Abstract:   (2942 Views)
In this paper, we proposed a practical method for classifying damages in pipes and plates using ultrasonic guided waves. The A-scan Pulse-Echo lamb wave ultrasonic tests used in this study. Tests accomplished on isotropic 1050 Aluminum with 0.4 mm thickness. Damages studied here were corrosion and crack which is common in pipe lines and steel structures like vehicles body or aerospace structures. This investigation is done in three steps. First step, experimental testing (making standard sample, lamb wave tests), second step, signal processing (window function, normalizing, wavelet function), third step, using the proper algorithm for classification. In first step, 206 ultrasonic lamb wave tests are measured on standard damaged samples (on pipe and plate) and the signals digitalized. After that, these signal processed and classified by classification algorithm. In this the classification algorithm is the support vector machine (SVM). In machine learning, support vector machines are supervised learning models with associated learning algorithms that analyze data and recognize patterns, used for classification and regression analysis. The results show that the corrosion damage can be distinguished from crack damages with 99% accuracy by proposed algorithm.
Full-Text [PDF 1016 kb]   (2546 Downloads)    
Article Type: Research Article | Subject: Non Destvuctive Test
Received: 2014/10/23 | Accepted: 2014/04/4 | Published: 2015/02/21

Add your comments about this article : Your username or Email:
CAPTCHA