Cited 0 time in
Automation of Ultrasound Nondestructive Testing and Improvement of Defect Detection Performance using Principal Component Analysis and ResNet
| DC Field | Value | Language |
|---|---|---|
| dc.contributor.author | Kim, Hyun Su | - |
| dc.contributor.author | Kim, Min Seok | - |
| dc.contributor.author | Jo, Hyeong Geun | - |
| dc.contributor.author | Park, Kwan Kyu | - |
| dc.contributor.author | Kim, Min Ho | - |
| dc.contributor.author | Jeong, Gi Ho | - |
| dc.date.accessioned | 2025-06-12T06:01:57Z | - |
| dc.date.available | 2025-06-12T06:01:57Z | - |
| dc.date.issued | 2025-04 | - |
| dc.identifier.issn | 1225-7842 | - |
| dc.identifier.issn | 2287-402X | - |
| dc.identifier.uri | https://scholarworks.bwise.kr/hanyang/handle/2021.sw.hanyang/207525 | - |
| dc.description.abstract | This study proposes a method for effectively extracting and classifying features from time-series data obtained during ultrasonic nondestructive testing (NDT) to detect defects. Traditional ultrasonic NDT is a technique used to inspect internal defects in semiconductor devices and other materials. It generates 2D images by analyzing ultrasonic reflection signals from various points on the specimen, which are then manually examined by an inspector to determine the presence and location of defects. However, this approach is susceptible to subjective judgment and is challenging to automate. To overcome these limitations, this study presents a method that minimizes inspector influence, automates the inspection process, and achieves an improvement of approximately 40% in detection sensitivity and an enhancement of 20% in defect detection performance compared to conventional methods. | - |
| dc.description.abstract | 본 연구에서는 초음파 비파괴검사 과정에서 획득한 시계열 데이터의 특징을 효과적으로 추출하고 분류하여 결함을 검출하는 방법을 제안한다. 전통적인 초음파 비파괴검사는 반도체 소자 등의 내부 결함을 확인하는 기술 중 하나로, 시편의 각 지점에 조사된 초음파 반사 신호를 이용해 2D 이미지를 생성한 뒤, 검사자가 직접 분석하여 결함의 유무와 위치를 판단한다. 그러나 이러한 방식은 검사자의 주관적 판단에 영향을 받을 수 있으며, 자동화가 어렵다는 한계를 가진다. 본 연구에서는 이러한 한계를 극복하기 위해 검사자의 영향을 최소화하고, 검사 과정을 자동화하며, 기존 방법보다 약 40% 향상된 검출 감도와 20% 향상된 검출 성능을 확보하는 방법을 제시한다. | - |
| dc.format.extent | 9 | - |
| dc.language | 한국어 | - |
| dc.language.iso | KOR | - |
| dc.publisher | 한국비파괴검사학회 | - |
| dc.title | Automation of Ultrasound Nondestructive Testing and Improvement of Defect Detection Performance using Principal Component Analysis and ResNet | - |
| dc.title.alternative | 주성분 분석과 ResNet을 활용한 초음파 비파괴검사 자동화 및 결함 검출 성능 개선 | - |
| dc.type | Article | - |
| dc.publisher.location | 대한민국 | - |
| dc.identifier.doi | 10.7779/JKSNT.2025.45.2.100 | - |
| dc.identifier.wosid | 001485822600003 | - |
| dc.identifier.bibliographicCitation | 비파괴검사학회지, v.45, no.2, pp 100 - 108 | - |
| dc.citation.title | 비파괴검사학회지 | - |
| dc.citation.volume | 45 | - |
| dc.citation.number | 2 | - |
| dc.citation.startPage | 100 | - |
| dc.citation.endPage | 108 | - |
| dc.type.docType | Article | - |
| dc.identifier.kciid | ART003198189 | - |
| dc.description.isOpenAccess | N | - |
| dc.description.journalRegisteredClass | esci | - |
| dc.description.journalRegisteredClass | kci | - |
| dc.relation.journalResearchArea | Materials Science | - |
| dc.relation.journalWebOfScienceCategory | Materials Science, Characterization & Testing | - |
| dc.subject.keywordAuthor | Defect Detection | - |
| dc.subject.keywordAuthor | Ultrasound | - |
| dc.subject.keywordAuthor | Principal Component Analysis (PCA) | - |
| dc.subject.keywordAuthor | Signal to Noise Ratio (SNR) | - |
| dc.subject.keywordAuthor | 결함 검사 | - |
| dc.subject.keywordAuthor | 초음파 | - |
| dc.subject.keywordAuthor | 주성분분석 | - |
| dc.subject.keywordAuthor | 신호 대 잡음 | - |
| dc.identifier.url | https://www.dbpia.co.kr/journal/articleDetail?nodeId=NODE12153416&language=ko_KR&hasTopBanner=true | - |
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.
222, Wangsimni-ro, Seongdong-gu, Seoul, 04763, Korea+82-2-2220-1366
COPYRIGHT © 2024 HANYANG UNIVERSITY.
Certain data included herein are derived from the © Web of Science of Clarivate Analytics. All rights reserved.
You may not copy or re-distribute this material in whole or in part without the prior written consent of Clarivate Analytics.
