Recently, there has been an increasing attempt to replace defect detection inspections in the manufacturing industry using deep learning techniques. However, obtaining substantial high-quality labeled data to enhance the performance of deep learning models entails economic and temporal constraints. As a solution for this problem, semi-supervised learning, using a limited amount of labeled data, has been gaining traction. This study assesses the effectiveness of semi-supervised learning in the defect detection process of manufacturing using the MixMatch algorithm. The MixMatch algorithm incorporates three dominant paradigms in the semi-supervised field: Consistency regularization, Entropy minimization, and Generic regularization. The performance of semi-supervised learning based on the MixMatch algorithm was compared with that of supervised learning using defect image data from the metal casting process. For the experiments, the ratio of labeled data was adjusted to 5%, 10%, 25%, and 50% of the total data. At a labeled data ratio of 5%, semi-supervised learning achieved a classification accuracy of 90.19%, outperforming supervised learning by approximately 22%p. At a 10% ratio, it surpassed supervised learning by around 8%p, achieving a 92.89% accuracy. These results demonstrate that semi-supervised learning can achieve significant outcomes even with a very limited amount of labeled data, suggesting its invaluable application in real-world research and industrial settings where labeled data is limited.

In factory automation, efforts are being made to increase productivity while maintaining high-quality products. In this study, a CNN network structure was designed to quickly and accurately recognize a cigarette located in the opposite direction or a cigarette with a loose end in an automated facility rotating at high speed for cigarette production. Tobacco inspection requires a simple network structure and fast processing time and performance. The proposed network has an excellent accuracy of 96.33% and a short processing time of 0.527 msec, showing excellent performance in learning time and performance compared to other CNN networks, confirming its practicality. In addition, it was confirmed that efficient learning is possible by increasing a small number of image data through a rotation conversion method.

This research proposes a novel approach to tackle the challenge of categorizing unstructured customer complaints in the automotive industry. The goal is to identify potential vehicle defects based on the findings of our algorithm, which can assist automakers in mitigating significant losses and reputational damage caused by mass claims. To achieve this goal, our model uses the Word2Vec method to analyze large volumes of unstructured customer complaint data from the National Highway Traffic Safety Administration (NHTSA). By developing a score dictionary for eight pre-selected criteria, our algorithm can efficiently categorize complaints and detect potential vehicle defects. By calculating the score of each complaint, our algorithm can identify patterns and correlations that can indicate potential defects in the vehicle. One of the key benefits of this approach is its ability to handle a large volume of unstructured data, which can be challenging for traditional methods. By using machine learning techniques, we can extract meaningful insights from customer complaints, which can help automakers prioritize and address potential defects before they become widespread issues. In conclusion, this research provides a promising approach to categorize unstructured customer complaints in the automotive industry and identify potential vehicle defects. By leveraging the power of machine learning, we can help automakers improve the quality of their products and enhance customer satisfaction. Further studies can build upon this approach to explore other potential applications and expand its scope to other industries.

To make semiconductor chips, a number of complex semiconductor manufacturing processes are required. Semiconductor chips that have undergone complex processes are subjected to EDS(Electrical Die Sorting) tests to check product quality, and a wafer bin map reflecting the information about the normal and defective chips is created. Defective chips found in the wafer bin map form various patterns, which are called defective patterns, and the defective patterns are a very important clue in determining the cause of defects in the process and design of semiconductors. Therefore, it is desired to automatically and quickly detect defective patterns in the field, and various methods have been proposed to detect defective patterns. Existing methods have considered simple, complex, and new defect patterns, but they had the disadvantage of being unable to provide field engineers the evidence of classification results through deep learning. It is necessary to supplement this and provide detailed information on the size, location, and patterns of the defects. In this paper, we propose an anomaly detection framework that can be explained through FCDD(Fully Convolutional Data Description) trained only with normal data to provide field engineers with details such as detection results of abnormal defect patterns, defect size, and location of defect patterns on wafer bin map. The results are analyzed using open dataset, providing prominent results of the proposed anomaly detection framework.

Recently, the development of computer vision with deep learning has made object detection using images applicable to diverse fields, such as medical care, manufacturing, and transportation. The manufacturing industry is saving time and money by applying computer vision technology to detect defects or issues that may occur during the manufacturing and inspection process. Annotations of collected images and their location information are required for computer vision technology. However, manually labeling large amounts of images is time-consuming, expensive, and can vary among workers, which may affect annotation quality and cause inaccurate performance. This paper proposes a process that can automatically collect annotations and location information for images using eXplainable AI, without manual annotation. If applied to the manufacturing industry, this process is thought to save the time and cost required for image annotation collection and collect relatively high-quality annotation information.

In the case of a die-casting process, defects that are difficult to confirm by visual inspection, such as shrinkage bubbles, may occur due to an error in maintaining a vacuum state. Since these casting defects are discovered during post-processing operations such as heat treatment or finishing work, they cannot be taken in advance at the casting time, which can cause a large number of defects. In this study, we propose an approach that can predict the occurrence of casting defects by defect type using machine learning technology based on casting parameter data collected from equipment in the die casting process in real time. Die-casting parameter data can basically be collected through the casting equipment controller. In order to perform classification analysis for predicting defects by defect type, labeling of casting parameters must be performed. In this study, first, the defective data set is separated by performing the primary clustering based on the total defect rate obtained during the post-processing. Second, the secondary cluster analysis is performed using the defect rate by type for the separated defect data set, and the labeling task is performed by defect type using the cluster analysis result. Finally, a classification learning model is created by collecting the entire labeled data set, and a real-time monitoring system for defect prediction using LabView and Python was implemented. When a defect is predicted, notification is performed so that the operator can cope with it, such as displaying on the monitoring screen and alarm notification.

In this study, surface roughness and interfacial defect characteristics were analyzed after forming a high-k oxide film on the surface of a prime wafer and a test wafer, to study the possibility of improving the quality of the test wafer. As a result of checking the roughness, the deviation in the test after raising the oxide film was 0.1 nm, which was twice as large as that of the Prime. As a result of current-voltage analysis, Prime after PMA was 1.07 × 10 A/cm2 and Test was 5.61 × 10 A/cm2, which was about 5 times lower than Prime. As a result of analyzing the defects inside the oxide film using the capacitancevoltage characteristic, before PMA Prime showed a higher electrical defect of 0.85 × 1012 cm2 in slow state density and 0.41 × 1013 cm2 in fixed oxide charge. However, after PMA, it was confirmed that Prime had a lower defect of 4.79 × 1011 cm2 in slow state density and 1.33 × 1012 cm2 in fixed oxide charge. The above results confirm the difference in surface roughness and defects between the Test and Prime wafer.

연구대상판결은 크게 두 부분으로 구성되어 있다. 첫째, 민법 제663조 제2항과 제3항에 의해 도급인의 ‘하자보수청구권’과 ‘하자 보수에 갈음하는 손해배상채권’은 특별한 사정이 없는 한 수급인의 도급인에 대한 ‘공사대금채권’과 동시이행의 관계에 있고, 이로 인해 도급인이 하자보수나 이에 갈음하는 손해배상청구권을 보유하고 이를 행사하는 한 도급인의 공사비 지급채무는 이행지체에 빠지지 않는다. 둘째, 수급인의 공사비채권의 변제기는 건물의 준공, 인도일이고, 도급인의 하자 보수에 갈음하는 손해배상채권의 변제기는 도급인이 그 권리를 행사한 때인데, 도급인이 하자보수에 갈음하는 손해배상채권을 자동채권으로 하고, 수급인의 공사비잔대금채권을 수동채권으로 하여 상계의 의사표시를 한 경우, 상계적상일 다음날이 아니라 “상계의 의사표시를 한 다음날”부터 이행지체에 빠진다. 연구대상판결은 동시이행의 범위에서 전부거절설에 기초하고 있다. 그러나 전부 거절설을 취한 이유에 대해서는 밝히지 않고 있다. 그러나 연구대상판결에서 동시이행의 항변권에 기해 전부거절설을 취한 것은 i) 하자보수에 갈음한 손해배상채권과 보수채권 사이의 동시이행의 항변권은 비쌍무계약상 동시이행의 항변권에 해당하므로, ii) 제667조 제3항에서 제536조 를 준용한 취지를 살펴 그 동시이행의 항변권에 기한 이행거절의 범위를 결정할 필요가 있는데, iii) 그 준용의 취지가 하자보수에 갈음하는 손해배상채권에 기한 상계권을 대금감액적으로 행사할 수 있도록 특별히 동시이행관계를 설정한 것 이라고 보는 것이 가장 적절한 설명이 될 수 있을 것으로 보인다. 또한 상계적상일 다음날이 아니라 상계의 의사표시 다음날부터 잔여보수채무의 이행지체책임을 지도록 한 것은 i) 하자보수에 갈음한 손해배상청구에서 하자 보수비용의 확정이 현실적으로 용이하지 않다는 현실적인 문제를 고려하여 그 하자보수비용 확정 전에 보수채무의 변제기가 도래하는 상황에서 도급인이 하자보수에 갈음하는 손해배상채권에 기한 상계권을 대금감액을 위해 행사할 수 있도록 제667조 제3항에서 제536조를 준용하여 하자보수에 갈음하는 손해 배상채권에 기한 동시이행의 항변권의 행사로 보수채무 전부를 거절할 수 있다는 현실적 관점과 ii) 상계의 소급효가 전부거절설에 의한 이행지체책임의 면제이익을 뒤집을 수 없다는 이론적 관점에서 정당화될 수 있다고 본다.

WDM(Wavelength Division Multiplexing) is called a wavelength division multiplexing optical transmission method and is a next-generation optical transmission technology. Case company F has recently developed and sold PLC(Planar Lightwave Circuit), a key element necessary for WDM system production. Although Chinese processing companies are being used as a global outsourcing strategy to increase price competitiveness by lowering manufacturing unit prices, the average defect rate of products manufactured by Chinese processing companies is more than 50%, causing many problems. However, Chinese processing companies are trying to avoid responsibility, saying that the cause of the defect is the defective PLC Wafer provided by Company F. Therefore, in this study, the responsibility of the PLC defect is clearly identified through estimating the defect rate of PLC using the sampling inspection method, and the improvement plan for each cause of the PLC defect for PLC yeild improvement is proposed. The result of this research will greatly contribute to eliminating the controversy over providing the cause of defects between global outsourcing companies and the head office. In addition, it is expected to form a partnership with Company F and a Chinese processing company, which will serve as a cornerstone for successful global outsourcing. In the future, it is necessary to increase the reliability of the PLC yield calculation by extracting more precisely the number of defects.

I propose an algorithm to detect defects in the production of wire mesh using computer image processing. The process is explained as follows, First reading consecutive frames coming through the camera, then the preprocessing process is performed. Second calculate the absolute difference between the two images to distinguish the moving wire mesh from the unnecessary background image. Third based on the past moving data of the welded wire mesh, predict and track future movement. As a result of observing the samples of some defective welded wire mesh products, it was confirmed that the horizontal line of the defective wire mesh had a higher height value of the tracked wire netting. Therefore it is possible to judge whether there is a defect or not at the same time without any additional process to judge. Finally, shear strength test were performed on the welds determined to be normal products by the algorithm proposed in this paper, so that I could verify the reliability and validity of the proposed algorithm.

The domestic CANDU nuclear power plants have been operated for a long time and various unforeseen spent fuel defects have been discovered. As the spent fuel defects are important factors in the safety of the nuclear power plant, a study on the analysis of the spent fuel defects to prevent their recurrence is necessary. However, in cases where the fuel rods inside the fuel assembly are defected, it is difficult to dismantle the fuel assembly owing to their welded structure and the facility conditions of the plant. Therefore, it is impossible to analyze the spent fuel defect because it is difficult to visually check the shape of the fuel defect. To resolve these problems, an analysis technology that can predict the number of defected fuel rods and defect size was developed. In this study, we developed a methodology for investigating the root cause of spent fuel defects using a database of the earlier fuel defects in the plants. It is anticipated that in the future this analysis technology will be applied when spent fuel defects occur.

NDF (No Defect Found) is a phenomenon in which defects have been found in the manufacturing, operation and use of a product or facility, but phenomenon of defects is not reproduced in the subsequent investigation system or the cause of the defects cannot be identified. Recently, with the development of the fourth industrial revolution, convergence of hardware and software technologies in various fields is spreading to products such as aircraft, home appliances, and mobile devices, and the number of parts is increasing due to functional convergence. The application of such convergence technologies and the increase in the number of parts are major factors that lead to an increase in NDF phenomena. NDF phenomena have a significant negative impact on cost, reliability, and reliability for both manufacturers, service providers and operators. On the other hand, due to the nature of NDF phenomena such as difficult and intermittent cause identification and ambiguity in judgment, it is common to underestimate the cost of NDF or fail to take appropriate countermeasures in corporate management. Therefore, in this paper, we propose a methodology for estimating NDF costs by the PAF model which is a quality cost analysis model and ABC (Activity Based Costing) technique. The methodology of this study suggests a detailed procedure and the concept to accurately estimate the NDF costs, using ABC analysis, accounting system information, and IT system data. In addition case studies have validated the methodology. We think this could be a valid methodology to refer to when estimating the cost of other parts. And, it is meaningful to provide important judgment information in the decision-making process based on quality management and ultimately reduce NDF costs by visualizing them separately by major variable factors.