• The Journal of the Convergence on Culture Technology
• /
• v.9 no.2
• /
• pp.569-574
• /
• 2023

For maintenance of concrete structures, it is necessary to identify and maintain various defects. With the current method, there are problems with efficiency, safety, and reliability when inspecting large-scale social infrastructure, so it is necessary to introduce a new inspection method. Recently, with the development of deep learning technology for images, concrete defect classification research is being actively conducted. However, studies on contamination and spalling other than cracks are limited. In this study, a variety of concrete defect type classification models were developed through transfer learning on a pre-learned deep learning model, factors that reduce accuracy were derived, and future development directions were presented. This is expected to be highly utilized in the field of concrete maintenance in the future.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2023.07a
• /
• pp.7-10
• /
• 2023

최근 기후 변화가 심각해짐에 따라 수소 에너지에 대한 관심이 집중되고 있으며 이를 안전하게 운송/보관할 수 있는 용기에 대한 연구도 활발히 진행되고 있다. 특히 고압 가스를 저장하는 TypeIII 용기의 노후화 및 안전과 관련되어 결함을 인지하는 연구가 활발하다. 그러나 이 용기의 외각층을 이루는 CFRP 소재는 탄소 섬유와 에폭시가 복잡한 구조로 구성되어 결함별 탐지가 매우 어렵다. 본 논문에서는 음향방출시험과 딥러닝을 활용하여 CFRP 결함 데이터셋을 구축하고 이를 분류할 수 있는 모델을 제안한다. 특히 CFRP 시편을 직접 제작하여 AE 센서를 부착하고 파괴하여 파형 데이터를 수집하였다. 이후 표현 학습을 통해 데이터의 특징을 압축/추출하고 유사도를 비교해 결함별 데이터를 판별하는 알고리즘을 개발하였다. 구축된 데이터셋의 실루엣 계수는 0.86으로 높은 군집도를 보였다. 마지막으로 구축된 데이터셋을 실시간으로 분류할 수 있는 1D-CNN 딥러닝 모델을 개발하였으며 99.33%의 높은 분류 정확도를 보였다.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2019.05a
• /
• pp.442-444
• /
• 2019

Fabric defect classification is an important issue in fabric quality control. However, automated classification is difficult because it is hard to identify various types of defects in images. classification of fabric defects mostly rely on human ability. In this paper, to solve this problem we apply Convolutional Neural Networks (CNN) for fabric defect classification. To make training CNN easier, we propose a method that is manually attentive defect regions in images. we compare the proposed method with the original image and confirm that the proposed method is effective for learning.

• Journal of the Korean Society for Nondestructive Testing
• /
• v.21 no.5
• /
• pp.510-517
• /
• 2001

This paper describes an initial study for the application of eddy current pattern recognition approaches to more realistic flaw characterization in steam generator tubes. For this purpose, finite-element model-based theoretical eddy current testing (ECT) signals are simulated from 5 types of OD flaws with the variation in flaw size parameters and testing frequency. In addition, three kinds of software are developed for the convenience in the application of steps in pattern recognition approaches such as feature extraction feature selection and classification by probabilistic neural networks (PNNs). The cross point of the ECT signals simulated from flaws with non-symmetric cross-sections shows the deviation from the origin of the impedance plane. New features taking advantages of this phenomenon are added to complete the feature set with a total of 18 features. Then, classification with PNNs are performed based on this feature set. The PNN classifiers show high performance for the identification of symmetry in the cross-section of a flaw. However, they show very limited success in the interrogation of the sharpness of flaw tips.

• Proceedings of the Korean Information Science Society Conference
• /
• 2006.10a
• /
• pp.295-300
• /
• 2006

DVD 비디오와 차세대 DVD의 한 진영인 HD-DVD 비디오 디스크는 다른 광 디스크와 마찬가지로 스크래치와 같은 결함이 발생할 수 있다. 결함은 물리적/논리적 원인에 따라 분류 가능하다. 결함은 디스크의 내부구조(네비게이션 데이터와 프리젠테이션 데이터)의 위치에 따라 각각 다른 이상 동작 현상을 발생 시킨다. 본 논문에서는 결함으로 인한 이상 동작 현상을 분류하고, 복구 또는 회피 할 수 있는 알고리즘을 제안하여 DVD/HD-DVD 비디오 디스크를 사용하는데 있어 결함으로 인한 불편함을 최소화하는데 목적이 있다.

• Journal of the Korean Society for Nondestructive Testing
• /
• v.17 no.4
• /
• pp.248-261
• /
• 1997

Ultrasonic pattern recognition is the most effective approach to the problem of discriminating types of flaws in weldments based on ultrasonic flaw signals. In spite of significant progress in the research on this methodology, it has not been widely used in many practical ultrasonic inspections of weldments in industry. Hence, for the convenient application of this approach in many practical situations, we develop an intelligent ultrasonic signature classification software which can discriminate types of flaws in weldments based on their ultrasonic signals using various tools in artificial intelligence such as neural networks. This software shows the excellent performance in an experimental problem where flaws in weldments are classified into two categories of cracks and non-cracks. This performance demonstrates the high possibility of this software as a practical tool for ultrasonic flaw classification in weldments.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2019.07a
• /
• pp.9-10
• /
• 2019

본 논문은 구동장치의 진동에서 특징 데이터를 추출하고 인공신경망에 학습을 시킨 후, 구동 장치의 결함을 분류하는 시스템을 구현하였다. 딥러닝 기술을 이용함으로써 특정 장치에 종속되지 않고 학습할 데이터의 특징에 따라 쉽게 변경 가능하다. 또한, 실제 적용될 현장에서 발생할 수 있는 예측외의 진동 환경에 유연하게 대처하기 위해 딥러닝 모델 중 CNN을 적용한 시스템을 설계하였으며, 본 연구팀의 이전 연구에서 제안된 DNN 기반의 진단시스템을 학습데이터의 환경과 다른 처리배제가 필요한 진동 환경에서 비교 실험하여 제안된 시스템이 새로운 환경적응 성능향상에 대하여 우수한 결과를 얻었음을 확인하였다.

• Journal of the Korean Society for Nondestructive Testing
• /
• v.29 no.1
• /
• pp.27-35
• /
• 2009

In this paper, we study the classification of steam generator tube defects using an improved feature extraction. We consider 4 axisymmetric defect patterns of tube: I-In type, I-Out type, V-In type, and V-Out type. Through numerical analysis program based on finite element modeling, 400 ECT signals are generated by varying width and depth of each defect type. From those generated ECT signals, we propose new feature vectors that include an angle between the two points where the Maximum impedance and half the Maximum impedance, and angles between Maximum impedance point and 10%, 20%, 30%, 40% of Maximum impedance points. Also, multi-layer perceptron with one hidden layer is used to classify the defect patterns. Through the computer simulation study, it is shown that the proposed method achieves an improved defect classification performance in terms of Maximum Error and mean square Error.

• The Korean Journal of Applied Statistics
• /
• v.35 no.3
• /
• pp.407-419
• /
• 2022

The semiconductor fabrication process is complex and time-consuming. There are sometimes errors in the process, which results in defective die on the wafer bin map (WBM). We can detect the faulty WBM by finding some patterns caused by dies. When one manually seeks the failure on WBM, it takes a long time due to the enormous number of WBMs. We suggest a two-step approach to discover the probable pattern on the WBMs in this paper. The first step is to separate the normal WBMs from the defective WBMs. We adapt a hierarchical clustering for de-noising, which nicely performs this work by wisely tuning the number of minimum points and the cutting height. Once declared as a faulty WBM, then it moves to the next step. In the second step, we classify the patterns among the defective WBMs. For this purpose, we extract features from the WBM. Then machine learning algorithm classifies the pattern. We use a real WBM data set (WM-811K) released by Taiwan semiconductor manufacturing company.

• Proceedings of the KIEE Conference
• /
• 2007.07a
• /
• pp.1746-1747
• /
• 2007

본 논문에서는 다층 퍼셉트론(Multi-Layer Perceptron)을 이용한 하드 디스크 결함 분포의 패턴 인식 기법을 제시한다. 결함 분포로부터 5 가지의 특징들을 추출하고, 이를 이용하여 퍼셉트론의 입력을 구성하였으며, 미리 분류된 표준 패턴 클래스를 이용하여 퍼셉트론의 출력을 구성하였다. 테스트 결과, 제시된 신경망은 하드 디스크의 패턴 분류에 만족할 만한 성능을 나타내었다.