Development of a method for evaluating corrosion defect by acoustic emission signals using machine learning

Name: Keisuke Yamamoto

Organization : Meiji University. School of Science and Technology

Name: Takuma Matsuo

Organization : Meiji University. School of Science and Technology

Email : ce222090@meiji.ac.jp

LCSH: Acoustic emission

LCSH: Acoustic emission testing

LCSH: Stress corrosion

Abstract: Since the steel pipe has a risk of leaking the contents due to corrosion, it is necessary to evaluate the soundness regularly. A conventional non-destructive inspection method has the problem that it requires a lot of cost to inspect a large area. Therefore, the acoustic emission (AE) method, which enables real-time monitoring of corrosion conditions, is the preferred method. The AE method can evaluate damage conditions based on waveform parameters, however, changes in AE parameters due to differences in damage conditions are small. Therefore, visual identification requires a lot of time for identification, and the accuracy varies depending on the skill of the engineer. In this study, it aims at development of the method to evaluate corrosion defect by AE signals using machine learning. At first, nine pipes which was subjected to thickness reduction with different depths was corroded. A 5% NaCl solution was dropped periodically in the pores to accelerate corrosion. Each steel pipe was conducted a 100-minute thermal cycle test to generate AE by short term measurement. Next, wavelet transforms were applied to AE waveforms to extract the wavelet coefficient of specific frequency, and the intensity ratio of the L(0, 1)-modes and F(1, 1)-modes, which are the fundamental modes of cylindrical waves were used as features for supervised learning. As a result, it was possible to classify corrosion defect with high accuracy when the damage level is divided into 3 groups.

King Mongkut's University of Technology North Bangkok. Central Library

Address: BANGKOK

Email: library@kmutnb.ac.th

Created: 2024

Modified: 2024-12-16

Issued: 2024-12-16

บทความ/Article

application/pdf

BibliograpyCitation : In Thai Society of Mechanical Engineers (TSME) and Chiang Mai University. The 13th TSME International Conference on Mechanical Engineering (TSME-ICoME 2023) (pp.267-274). Chiang Mai : Chiang Mai University, 2023

eng

©copyrights King Mongkut's University of Technology North Bangkok

ลำดับที่.	ชื่อแฟ้มข้อมูล	ขนาดแฟ้มข้อมูล	จำนวนเข้าถึง	วัน-เวลาเข้าถึงล่าสุด
1	TSME-ICoME13 2023pp.267-274.pdf	813.66 KB

ใช้เวลา

0.032933 วินาที