• 스마트미디어저널
• /
• 제11권11호
• /
• pp.92-98
• /
• 2022

섬유산업에서 생산된 직물의 결함을 식별하는 것은 품질관리를 위한 핵심적인 절차이다. 본 연구는 직물의 이미지를 분석하여 결함을 검출하는 모델을 만들고자 하였다. 연구에 사용된 모델은 딥러닝 기반의 VGGNet 과 ResNet이었고, 두 모델의 결함 검출 성능을 비교하여 평가하였다. 정확도는 VGGNet 모델이 0.859, ResNet 모델이 0.893으로 ResNet 모델의 정확도가 더 높은 결과를 보여주었다. 추가적으로 딥러닝 모델이 직물의 이미지 내에서 결함으로 인식한 부분의 위치를 알아보기 위하여 XAI(eXplainable Artificial Intelligence)기법인 Grad-CAM 알고리즘을 사용하여 모델의 관심영역을 도출하였다. 그 결과 딥러닝 모델이 직물의 결함으로 인식한 부분이 육안으로도 실제 결함이 있는 것으로 확인되었다. 본 연구의 결과는 직물의 결함 검출에 있어서 딥러닝 기반의 인공지능을 활용함으로써 섬유의 생산과정에서 발생하는 시간과 비용을 줄일 수 있을 것으로 기대된다.

• 비파괴검사학회지
• /
• 제33권1호
• /
• pp.34-39
• /
• 2013

선행 연구에서 적외선열화상기법을 이용하여 원전 배관의 감육 결함을 측정하기 위하여, 4 inch 배관에 인공결함을 가공하여 이에 대한 결함 검출을 도출하였다. 본 논문에서는 선행연구에서 도출된 조건을 이용하여 원전 소구경 배관의 결함 검출 조건에 관한 연구를 수행하였다. 결함의 가공은 감육 길이, 원주방향 각도, 감육 깊이를 변화시켜서 결함 조건을 가공하였다. 사용된 장비는 IR camera와 1 kW용량의 halogen lamp 2개를 사용하였으며, halogen lamp와 대상 배관과의 거리를 1 m, 1.5 m, 2 m 순으로 변화시켜 실험을 수행하였다. 실험 결과의 분석을 위하여 온도분포데이터를 확보하고, 이를 분석하여 결함 길이를 측정하였다. 4 inch 배관의 인공결함은 2 m에서 측정 결과의 신뢰도가 높았으나, 소구경 배관은 1.5 m에서 결함이 명확하게 검출되었다.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2001년도 하계학술대회 논문집 C
• /
• pp.1673-1675
• /
• 2001

In this paper, Partial Discharge(PD) patterns are compared by means of Low Frequency PD(LFPD, based on the IEC 270) and recently proposed High Frequency PD(HFPD). For this purpose, three different types of artificial defects are provided such as corona in air, void in epoxy insulator and needle defect in XLPE cable insulation. PD were generated from each defect and then detected respectively by two different methods such as LFPD and HFPD. As a result, remarkable resemblance in PD pattern for differ detecting method have been observed from each defect. Accordingly, it could be deduced that the pattern recognition by LFPD could be regarded as the reference for the investigations by HFPD.

• 한국추진공학회:학술대회논문집
• /
• 한국추진공학회 2006년도 제27회 추계학술대회논문집
• /
• pp.289-292
• /
• 2006

본 논문에서는 항공기용 터보 축 엔진의 결함진단 알고리즘으로 지지 벡터 장치(Support Vector Machine) 과 인공신경망(Artificial Neural Network) 을 복합으로 이용하였다. 인공신경망 알고리즘의 특성상 데이터 수에 따라 정확성과 수렴속도 등에서 차이가 나므로 탈설계 영역에서의 효용성여부를 판단하기 위해서 연료유량과 마하수에 따른 탈설계 영역 진단 결과를 지상정지 상태와 비교하였다.

• 한국추진공학회지
• /
• 제10권2호
• /
• pp.102-109
• /
• 2006

본 논문에서 항공기용 터보 축 엔진의 결함 진단 알고리즘을 개발하기 위해 Support Vector Machine(SVM)과 인공신경망(ANN)을 이용하였다. 신경망을 이용한 시스템은 비선형성이 과도한 데이터를 학습할 때 지역 최소점(Local Minima)에 빠져 분류 정확률이 낮아질 수 있다. 이러한 위험성을 보안하기 위해 SVM에 의한 ANN의 분할 학습 알고리즘(SLA)을 제안하였다. 이것은 SVM을 이용하여 결함 위치를 판별 한 후 신경망이 선택적으로 학습을 하는 방법으로 학습 데이터의 비선형성을 줄여 분류 정확률을 높이기 때문에 신경망을 단독으로 사용할 때보다 개선된 성능을 보여주었다.

• 소성∙가공
• /
• 제32권5호
• /
• pp.276-286
• /
• 2023

The field of sheet metal forming using press technology has become essential in modern mass production systems. Draw bead is often used to enhance formability. However, optimal draw bead design often requires excessive time and cost due to iterative experimentation and sometimes results in some defects. Given these challenges, there is a need to enhance formability by introducing segmented blank holders without draw beads. In this paper, the feasibility of a localized holding strategy using segmented blank holders is evaluated without the use of draw beads. The possibility for improving the formability was evaluated by utilizing a combination of the forming limit diagram and the wrinkle pattern-based defect indicators. Artificial neural networks were used for predicting defect indicators corresponding to arbitrary input holding forces and the NSGA-II optimization algorithm is used to find optimum blank holder forces yielding better defect indicators than the original process with drawbeads. Using optimum holding forces obtained from the proposed procedure, the stamping process with the segmented blank holders can yield better formability than the conventional process with drawbeads.

• 한국멀티미디어학회논문지
• /
• 제19권3호
• /
• pp.633-641
• /
• 2016

Automatic defect inspection system is composed of the step in the pre-processing, defect candidate detection, and classification. Polarizing films containing various defects should be minimized over-detection for classifying defect blobs. In this paper, we propose a defect detection algorithm using a skewness of histogram for minimizing over-detection. In order to detect up defects with similar to background pixel, we are used the characteristics of the local region. And the real defect pixels are distinguished from the noise using the probability density function. Experimental results demonstrated the minimized over-detection by utilizing the artificial images and real polarizing film images.

• 동력기계공학회지
• /
• 제15권4호
• /
• pp.37-41
• /
• 2011

This paper introduces the special eddy current sensor and its characteristic for bolt hole defect evaluation in gas turbine rotor. In the past, Fluorescent penetration inspection method was used for qualitative defect evaluation in gas turbine rotor bolt hole. This method can defect the bolt hole defect but can not evaluate the defect size. Nowadays, eddy current method is used quantitative defect evaluation due to advanced sensor design technology. And eddy current method is more time and cost saving than the old method. We developed bolt shape eddy current sensor for the rotor bolt hole defect detection and evaluation. The eddy current sensor moves to the bolt hole guided by screw nut and detects the defect on the bolt hole. The bolt hole mock-up and artificial defects were made and used for the signal detection & resolution analysis of eddy current sensor. The results show that signal detection capability is enough to detect 0.2 mm depth defect. And the resolution capability is enough to differentiate 02, 0.5, 1.0 and 2.0 mm depth defect.

• 한국해양공학회지
• /
• 제10권1호
• /
• pp.53-64
• /
• 1996

In this paper, rotating bending fatigue tests have been carried out to investigate the growth behabiors of surface fatigue crack initiated from a small artificial surface defect, that might exist in real structures, on 2024-T3 and 6:4 brass. The test results are analysed in the viewpoints of both strength of materials and fracture mechanics, it can be concluded as follows. The effect of a small artificial surface defect upon the fatigue strength is very large. The sensitivity of 2024-T3 on the defect is higher than that of 6:4 brass. The growth behavior of the surface fatigue crack of 2024-T3 is different from that of 6:4 brass. The growth rate of the surface fatigue crack of 2024-T3 is considerably rapid in the early stage of the fatigue life and apt to decrease in the later stage. It was impossible to establish a unifying approach in the analysis of crack growth begabior of 2024-T3 and 6:4 brass using the maximum stress intensity factor because of their dependence on stress level. But if the elastic strain and cyclic total strain intensity factor range were applied to obtain the growth rate of surface fatigue cracks of the materials, the data were found to be nearly coincided.

• Journal of Chest Surgery
• /
• 제10권1호
• /
• pp.143-147
• /
• 1977

Atrial septal defect is the most common of the congenital heart disease in the adult. Since the first description of atrial septal defect by Rokitansky in 1857, this anomaly has been studied by many workers in past one century. In 1953, Lewis had first corrected the atrial septal defect under direct vision with deep hypothermia, and Gibbon [1954] had first done the atrial septal defect under direct vision with extracorporeal circulation. In our college [May 2’ 1977], we have first repaired the A.S.D. under direct vision with artificial heart-lung machine and, the defect was 4x5cm in size which was closed by Dacron patch. This patient was 12 year old girl who is well now. [postoperative 13 days]