• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.20 no.3
• /
• pp.171-179
• /
• 2020

Recently, studies have shown that convolution neural networks are achieving the best performance in image classification, object detection, and image generation. Vision based defect inspection which is more economical than other defect inspection, is a very important for a factory automation. Although supervised anomaly detection algorithm has far exceeded the performance of traditional machine learning based method, it is inefficient for real industrial field due to its tedious annotation work, In this paper, we propose ADGAN, a unsupervised anomaly detection architecture using the variational autoencoder and the generative adversarial network which give great results in image generation task, and demonstrate whether the proposed network architecture identifies anomalous images well on MNIST benchmark dataset as well as our own welding defect dataset.

• Journal of the Korean Society for Nondestructive Testing
• /
• v.32 no.1
• /
• pp.12-19
• /
• 2012

This study conducted infrared (IR) thermography tests using pipe and plate specimens with artificial wall-thinning defects to find an optimal condition for IR thermography test on the wall-thinned nuclear piping components. In the experiment halogen lamp was used to heat the specimens. The distance between the specimen and the lamp and the intensity of halogen lamp were regarded as experimental parameter. When the distance was set to 1~2 m and the lamp intensity was above 60 % of full power, a single scanning of IR thermography detected all artificial wall-thinning defects, whose minimum dimension was $2{\Theta}=90^{\circ}$, d/t=0.5, and $L/D_o=0.25$, within the pipe of 500 mm in length. Regardless of the distance between the specimen and the lamp, the image of wall-thinning defect in IR thermography became distinctive as the intensity of halogen lamp increased. The detectability of IR thermography was similar for both plate and pipe specimens, but the optimal test condition for IR thermography depended on the type of specimen.

• Journal of the Korean Society for Nondestructive Testing
• /
• v.24 no.3
• /
• pp.253-258
• /
• 2004

To get accurate information about the defect from the test signal, NDT engineers should have a good knowledge on forward problems. Such knowledge is usually obtained by a lot of testing experiences. Another why of obtaining such knowledge is to build a database containing lots of defect information and their corresponding signals. However, the archiving of raw test data would require a lot of storage space. In this paper, an economic way of storing signals is studied by using Fourier descriptors. Instead of saving raw signal data, Fourier descriptors are saved and the storage spare is reduced. Of course, the defect signal can be reconstructed from the stored descriptors. By using differential ECT signals produced by numerical modeling and experiment, the savings of 85% from the original signal and $57{\sim}65%$ from the filtered signal in the storage space were confirmed. The similarity of the reconstructed signal and the original signal was also demonstrated. This Fourier descriptor approach could contribute significantly in building differential signal databases.

• Proceedings of the KWS Conference
• /
• 2010.05a
• /
• pp.61-61
• /
• 2010

The purpose of this study is to prevent the unstable fracture at the FCA butt weldment of hatch coaming deck in the ultra large containership during service life. In order to do it, the behavior of the embedded crack at the weldment under design loading conditions was evaluated in accordance with BS7910. Here, the level of primary stress induced by ship motion was evaluated by the design code of classification society and FEA. The level of residual stress as secondary stress was calculated in consideration of the restraint degree of weldment and welding heat input by using the predictive equation proposed by authors in the previous study. The fatigue crack growth rate at the weldment was evaluated using CT specimen in accordance with ASTM E647. According to the results, although the allowable defect for embedded crack specified in the classification society exists at the weldment, the occurrence possibility of unstable fracture at the weldment could not be negligible, regardless of CTOD value given in this study. So, in this study, the effect of initial defect size, welding heat input, restraint degree and CTOD on the fracture mechanical characteristics of embedded crack at the weldment was evaluated by the comprehensive fracture assessment. Based on the results, the design criteria including allowable defect, residual stress level and CTOD value was established to prevent the unstable fracture at the FCA butt weldment of hatch coaming deck in an ultra large containership during service life of 20years.

• Journal of Korean Society of Water and Wastewater
• /
• v.34 no.4
• /
• pp.251-258
• /
• 2020

Sewer deterioration models are needed to forecast the remaining life expectancy of sewer networks by assessing their conditions. In this study, the serious defect (or condition state 3) occurrence probability, at which sewer rehabilitation program should be implemented, was evaluated using four probability distribution functions such as normal, lognormal, exponential, and Weibull distribution. A sample of 252 km of CCTV-inspected sewer pipe data in city Z was collected in the first place. Then the effective data (284 sewer sections of 8.15 km) with reliable information were extracted and classified into 3 groups considering the sub-catchment area, sewer material, and sewer pipe size. Anderson-Darling test was conducted to select the most fitted probability distribution of sewer defect occurrence as Weibull distribution. The shape parameters (β) and scale parameters (η) of Weibull distribution were estimated from the data set of 3 classified groups, including standard errors, 95% confidence intervals, and log-likelihood values. The plot of probability density function and cumulative distribution function were obtained using the estimated parameter values, which could be used to indicate the quantitative level of risk on occurrence of CS3. It was estimated that sewer data group 1, group 2, and group 3 has CS3 occurrence probability exceeding 50% at 13th-year, 11th-year, and 16th-year after the installation, respectively. For every data groups, the time exceeding the CS3 occurrence probability of 90% was also predicted to be 27th- to 30th-year after the installation.

• Journal of the Korean Society for Nondestructive Testing
• /
• v.36 no.6
• /
• pp.483-489
• /
• 2016

The typical methods used for inspecting ferromagnetic pipes include the ultrasonic testing (UT) contact method and the following non-contact methods: magnetic flux leakage (MFL), electromagnetic acoustic transducers (EMAT), and remote field eddy current testing (RFECT). Among these methods, the RFECT method has the advantage of being able to establish a system smaller than the diameter of a pipe. However, the method has several disadvantages as well, including different sensitivities and difficult-to-repair coil sensors which comprise its array system. Therefore, a giant magneto-resistance (GMR) sensor was applied to address these issues. The GMR sensor is small, easy to replace, and has uniform sensitivity. In this experiment, the GMR sensor was used to measure remote field and defect signal characteristics (in the axial and radial directions) in a ferromagnetic pipe. These characteristics were measured in an effort to investigate standard defects at changing depths within a pipe. The results show that the experiment successfully demonstrated the applicability of the GMR sensor to RFECT signal detection in ferromagnetic pipe.

• Journal of the Korea Institute of Building Construction
• /
• v.15 no.2
• /
• pp.209-215
• /
• 2015

This study was conducted to analyze the actual utilization rate of the deposit for defect repair of apartment complexes and its influential factors. We analyzed the data on enforced defect deposit cases by using one-sample t-test, Pearson's correlation analysis, and simple linear regression analysis. The difference between the deposit amount specified and the enforcement amount actually disbursed was found to about 0.51%. The change rates for short-term costs amounted approximately to $839KRW/m^2$ and 130,000 KRW/household per year, and those for long-term costs were $647KRW/m^2$ and 123,207 KRW/household per year. The results warrant the need for further research on establishing a deposit amount based on actual statistical data.

• Journal of the Korean Society for Nondestructive Testing
• /
• v.30 no.2
• /
• pp.91-97
• /
• 2010

This paper proposes method to evaluate the location and size of the internal defects of metallic specimens by the shearing phase lock-in infrared thermography. Until now, infrared thermography test for metal specimen of STS304 and Cu-Zn were conducted to find the best test conditions. However, In unspecified situation of the form and existence of defects, there was a disadvantage to takes a long time for finding the optimal experimental conditions. The defect detection and evaluation was performed at 60 MHz signal using lock-in and shearing-phase method under limited heating conditions. By shearing-phase distribution method, Defects for the maximum, minimum and zero points were quantitatively detected at the size and location of the subsurface. As results, application of the proposed technique was verified for STS304 and Cu7-Zn3 with artificial defect and factors affected defect evaluation were searched and analyzed.

• Journal of the Korea Institute of Building Construction
• /
• v.21 no.2
• /
• pp.149-156
• /
• 2021

This study empirically verified whether aggressive order intake affect occurrence of defects of Korean construction companies. In order to compensate for the shortage of construction costs due to aggressive low-cost orders of contractors, self-response measures such as design changes are followed, and construction defects may be induced in this process. It aims to verify relevance between aggressive order intake and defect in construction empirically using financial accounting data. As a result, it was confirmed that an abnormal increase in order volume and a deterioration in profitability such as operating margin due to aggressive order taking behavior leads to a decrease in construction quality and occurrence of defects. In addition, it was also confirmed that low construction ability and intentional changes in construction design or contract conditions after construction commences were a direct causes of defects. The results of this study can serve as an opportunity to come up with measures to prevent defects by confirming the relationship between defects in construction and deterioration of profitability due to aggressive low-cost order takings.

• The Journal of the Convergence on Culture Technology
• /
• v.9 no.2
• /
• pp.561-568
• /
• 2023

Various defects occur in concrete structures due to internal and external environments. If there is a defect, it is important to efficiently identify and maintain it because there is a problem with the structural safety of concrete. However, recent deep learning research has focused on cracks in concrete, and studies on exfoliation and contamination are lacking. In this study, focusing on exfoliation and contamination, which are difficult to label, four models were developed and their performance evaluated through unlabelling method, filtering method, the convergence of transfer learning based k-means clustering. As a result of the analysis, the convergence model classified the defects in the most detail and could increase the efficiency compared to direct labeling. It is hoped that the results of this study will contribute to the development of deep learning models for various types of defects that are difficult to label in the future.