• 산업경영시스템학회지
• /
• 제46권2호
• /
• pp.1-12
• /
• 2023

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.

• Journal of information and communication convergence engineering
• /
• 제9권1호
• /
• pp.21-26
• /
• 2011

In this paper, we suggest the object-oriented methodology for the design and implementation scheme for the program development in the application of control and instrumentation such as the detection system of RFID defective tags which needs the embedded programming. We apply the design methodology of UML in the system design phase, and suggest the implementation scheme of LabVIEW programs using LVOOP(LabVIEW Object Oriented Programming)in which make it possible to write the object-oriented programming. We design the class diagram and the sequence diagram using UML, and write the classes of LVOOP from the designed class diagram and the main VI from the sequence diagram, respectively. We show that it is possible to develop the embedded programs such as the RFID application through the implementation example of the detection system of RFID defective tags in this paper. And, we obtain the advantages based on the object-oriented design and implementation using the LVOOP approach such as the development of LabVIEW programs by adding the classes and the concept of object of the object-oriented language to LabVIEW.

• 대한전자공학회논문지SD
• /
• 제44권4호
• /
• pp.55-62
• /
• 2007

본 논문에서는 이미지 센서에서 불량 화소를 자동으로 검출하기 위한 알고리듬을 제안하고, 그에 따른 하드웨어 구조를 제시하였다. 기존에 제안된 방법은 영상의 특징을 고려하지 않고 단순히 주위 화소들 값과의 차이가 일정 이상이면 불량 화소로 간주하였다. 그러나 이러한 방식은 영상에 따라서 불량 화소가 아닌 화소를 불량 화소로 간주하거나, 불량 화소를 정상 화소로 판단하는 일이 발생한다. 이러한 단점을 보완하기 위해 여러 프레임에 걸쳐 확인하는 방법도 제안되었으나, 불량 화소 검출시간이 오래 걸리는 단점이 있다. 이러한 기존 방식의 단점을 해결하기 위해, 제안된 불량 화소 검출 기법은 단일화면 내에서는 경계 영역을 고려하여 불량 화소를 검출하고, 여러 프레임에 걸친 확인 과정을 거치되, 화면 전환 여부를 확인하여 화면 전환이 일어날 때마다 검출된 화소의 불량 화소 여부를 판단하고 확인한다. 실험 결과, 단일 화면 내에서의 검출률은 기존 대비 6% 향상되었고, 100%의 불량화소 검출까지 걸리는 시간은 평균적으로 3배 이상 단축되었다. 본 논문에서 제안된 알고리듬은 하드웨어로 구현되었고, 하드웨어 구현 시 색 보간 블록에서 사용되는 경계 영역 표시자를 그대로 활용함으로써 0.25um 표준 셀 라이브러리를 이용하여 합성했을 때, 5.4K gate의 낮은 복잡도로 구현할 수 있었다.

• 한국컴퓨터정보학회논문지
• /
• 제16권2호
• /
• pp.47-53
• /
• 2011

국내 대부분의 RFID 태그 생산업체에서는 칩 본딩 작업 후에 태그 불량품 검출 작업이 수작업으로 수행되기 때문에 태그 생산 시간과 비용을 감소시키기 위한 요구가 산업계에서 요구되어 왔다. 이에 따라 이 논문에서는 태그 불량 검출 기능을 수행하는 시스템을 개발하여 산업 현장에서 요구되는 RFID 태그 부적합품 검출 자동화 시스템을 구축할 수 있는 소프트웨어 분야의 기초 기반 기술을 제공한다. 프로그램의 개발 비용을 낮추고 유지보수를 용이하게 하기 위하여 시스템의 모델링에는 UML을 사용하고 프로그램의 구현에는 JAVA 언어를 사용하여 개발하였다. 개발된 시스템은 태그 불량 검출 작업을 수작업으로 처리하는 방법과 비교하여, 매우 뛰어난 처리속도의 성능 향상과 부적합품 검출에 대한 완벽한 인식률을 보여준다.

• 한국기계가공학회지
• /
• 제9권3호
• /
• pp.76-82
• /
• 2010

A laborer always watched a process of production carefully but defective manufactures were inspected after press process. These inspections made a waste of human power and defective manufactures could make a serious damage of press mold. Therefore, AE(Acoustic Emission) system was introduced to prevention of the damage of the press molds, to a real time detection of defective manufactures and to save human power. AE system was introduced to solve this problem which is a detecting defective manufacture on real time and to prevent the damage of the press mold. In this research we get acoustic emission signal in accordance with weight and processing method of press by using AE sensor, Preamplifier, AE board signal board which occurs press processing and it analyzed various signal through using CMD8 software on the time. From the result, we found that the intensity and shape of the signal were changed according to the weight and processing type of the press. By using this special algorithm, it can judge the acoustic signal which occurs from press on real time.

• Journal of information and communication convergence engineering
• /
• 제21권1호
• /
• pp.24-31
• /
• 2023

The solid-state drive (SSD) possesses higher input and output speeds, more resistance to physical shock, and lower latency compared with regular hard disks; hence, it is an increasingly popular storage device. However, tiny components on an internal printed circuit board (PCB) hinder the manual detection of malfunctioning components. With the rapid development of artificial intelligence technologies, automatic detection of components through convolutional neural networks (CNN) can provide a sound solution for this area. This study proposes applying the YOLOv5 model to SSD PCB component detection, which is the first step in detecting defective components. It achieves pioneering state-of-the-art results on the SSD PCB dataset. Contrast experiments are conducted with YOLOX, a neck-and-neck model with YOLOv5; evidently, YOLOv5 obtains an mAP@0.5 of 99.0%, essentially outperforming YOLOX. These experiments prove that the YOLOv5 model is effective for tiny object detection and can be used to study the second step of detecting defective components in the future.

• 한국기계가공학회지
• /
• 제18권12호
• /
• pp.38-44
• /
• 2019

Various automation studies have been conducted to detect defective products based on product images. In the case of machine vision-based studies, size and color error are detected through a preprocessing process. A situation may arise in which the main features are removed during the preprocessing process, thereby decreasing the accuracy. In addition, complex systems are required to detect various kinds of defects. In this study, we designed and developed a system to detect errors by analyzing various conditions of defective products. We designed the deep learning algorithm to detect the defective features from the product images during the automation process using a convolution neural network (CNN) and verified the performance by applying the algorithm to the checker-switch failure detection system. It was confirmed that all seven error characteristics were detected accurately, and it is expected that it will show excellent performance when applied to automation systems for error detection.

• 한국소음진동공학회:학술대회논문집
• /
• 한국소음진동공학회 2006년도 추계학술대회논문집
• /
• pp.392-395
• /
• 2006

In this paper, Maharanobis distance was used to distinguish defective motors from good motors. Maharanobis distance was calculated from the noise data of good motors and the test motor that were measured in 1/3 octave hand from 25 Hz to 20 kHz frequency range. The suggested method was applied to the detection of defective air-conditioner motors.

• 대한원격탐사학회:학술대회논문집
• /
• 대한원격탐사학회 2005년도 Proceedings of ISRS 2005
• /
• pp.130-133
• /
• 2005

This paper presents the efficient method to detect the defective detectors of the SWIR band of SPOT 4. The key of this method are to flatten the baseline of the data using high pass band filter instead of differentiation. This method is made up six steps. First step is to apply image enhancement techniques to enhance the lines imaged by defective detector and improve the quality of an image. Second step is processed by summing the enhanced image in line direction. These summed data have the peaks that represent the defective detectors and the curved baseline characterized by the reflectivity of Earth surface. In order to exactly detect these peaks, third step is to flatten the curved baseline using high pass filtering in the frequency domain. In fourth step, the data with flat baseline is normalized to have zero mean and unity standard deviation. In fifth step, the defective detectors are detected using $99.9\%$ confidence interval. Finally, after removing the detected ones in summed data, the steps from third to five are iterated. Three SPOT 4 images, which have different reflectivity of Earth surface and different sensor, were used to validate this method. The overall accuracy of detection for three images was $97.9\%$. This result shows that this method can detect efficiently the lines made by defective detectors.

• 스마트미디어저널
• /
• 제11권4호
• /
• pp.46-55
• /
• 2022

과일 품질 자동 선별 시스템에서 흠집이나 부패한 부위가 존재하는 불량 과일을 우선적으로 검출하여 제거하는 작업은 매우 중요하다. 본 연구에서는 기존의 영상처리 기법을 이용하여 불량 부위가 있는 과일 검출하는 방법의 한계점을 극복하기 위하여, 최신 인공지능 기술인 Yolo V4 딥러닝 지능기술을 이용하여 과일 불량 부위를 검출하는 방법을 제안한다. 본 연구에서는 흠집 또는 부패 부위가 존재하는 1,100개의 불량 사과 및 1,300개의 불량 배를 포함한 총 2,400개의 불량 과일에 대하여 Yolo V4 딥러닝 모델을 사용하여 학습하고 불량 부위 검출 실험을 하였다. 성능 실험 결과에 따르면 사과의 정확률은 0.80, 재현율은 0.76, IoU는 69.92%, mAP는 65.27%이고, 배의 정확률은 0.86, 재현율은 0.81, IoU는 70.54%, mAP는 68.75%의 성능을 나타내었다. 본 연구에서 제안한 방법은 기존 영상처리 기법을 이용한 방법보다 불량 부위가 있는 과일을 실시간으로 정확하게 선별하여 기존 과일 자동 품질 선별시스템의 성능을 획기적으로 개선할 수 있다.