• 전기학회논문지P
• /
• 제51권1호
• /
• pp.52-54
• /
• 2002

This research presented in this paper focuses on a system development for checking stator's wedge of large generator in heavy industry. The system is composed of impactor, A/D converter and digital signal processing board. Also this paper proposes a method for detection of stator's insulating state using analysis algorithm. As developed impact system is acting, impact wave is generated real time and transferred data to microprocessor. Using communication program this signal move microprocessor to hard directory of note book with 1000 data/sec. The developed system and analysis program performed very well by real stator's wedge in large generator.

• 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.

• 방송공학회논문지
• /
• 제25권2호
• /
• pp.200-207
• /
• 2020

CNN 기반 인공신경망은 영상 분류, 객체 인식, 화질 개선 등 다양한 분야에서 뛰어난 성능을 보이고 있다. 그러나, 많은 응용에서 딥러닝(Deep Learning) 모델의 복잡도 및 연산량이 방대해짐에 따라 IoT 기기 및 모바일 환경에 적용하기에는 제한이 따른다. 따라서 기존 딥러닝 모델의 성능을 유지하면서 모델 크기를 줄이는 인공신경망 압축 기법이 연구되고 있다. 본 논문에서는 인공신경망 압축기법을 통하여 원본 CNN 모델을 압축하고, 압축된 모델을 임베디드 시스템 환경에서 그 성능을 검증한다. 성능 검증을 위해 인공지능 지원 맞춤형 칩인 QCS605를 내장한 임베디드 보드에서 카메라로 입력한 영상에 대해서 원 CNN 모델과 압축 CNN 모델의 분류성능과 추론시간을 비교 분석한다. 본 논문에서는 이미지 분류 CNN 모델인 MobileNetV2, ResNet50 및 VGG-16에 가지치기(pruning) 및 행렬분해의 인공신경망 압축 기법을 적용하였고, 실험결과에서 압축된 모델이 원본 모델 분류 성능 대비 2% 미만의 손실에서 모델의 크기를 1.3 ~ 11.2배로 압축했을 뿐만 아니라 보드에서 추론시간과 메모리 소모량을 각각 1.2 ~ 2.1배, 1.2 ~ 3.8배 감소함을 확인했다.

• 정보처리학회논문지:소프트웨어 및 데이터공학
• /
• 제4권2호
• /
• pp.65-70
• /
• 2015

PCB(Printed Circuit Board) 제조공정에서의 수율은 제품의 원가와 품질을 결정하는 중요한 관리 요인이다. PCB 제조공정은 일반적으로 많은 단계의 미세공정을 거쳐서 제품인 칩(Chip)이 생산되기 때문에 높은 수율을 보장하기가 현실적으로 어렵다. 제품의 수율을 향상시키기 위해서는 저수율의 원인이 되는 불량요인을 분석하고, 불량요인에 영향을 미치는 중요공정 및 설비를 찾아서 관리해야 한다. 본 연구는 로지스틱 회귀분석 및 변수선택법을 이용하여 혐의공정 및 설비를 찾는 방법을 제안하였다. 데이터는 실제 현장의 로트 데이터를 사용하였고, 각 로트는 진행한 설비 및 불량유형별 불량수를 갖고 있다. 또한 분석 결과는 실제 현장 확인을 통하여 수율에 미치는 영향을 확인하였다.

• Archives of Reconstructive Microsurgery
• /
• 제16권2호
• /
• pp.133-143
• /
• 2007

For the better success in the society, it is important to collect opinions from members. The progress was slow before because it was hard to collect the opinions from each member. But since we have opened the web site it gets easier to evolve with the society for microsurgery, and expect to have active participations from the members. We make 20 questions in three different groups. The researches are made on participation, development in science council that starting from 2005 October to 2007 June, education on training with member's preference to thesis and individuality and suggestion. Thus we have results following to that research. 1.75% of the members vote for the lecture has to be substantial, and also want the atmosphere of the council to be more sincere. 46.2% of members want the council to be not only exchange information but also the place to experience the social intercourse. 2. When the thesis is present and submitted that will also connect to main society. It will be counted as equal as board thesis, only if the thesis can count in both plastic & reconstructive surgery and orthopedic surgery. Then it will be scored as same as examination of board, and will be a receipt of paper by home page 3. The information-oriented project on our council is slower than what members expected. 100% of members voted yes to online thesis examination, and 66.7% of members voted for preference of web site. We need to make improvements on our web pages as soon as possible such as online payment with credit card. 4. According to the result of questionnaire, our web site helps member's participation in council and eventually become a bridge between members. We hope the questionnaires would help us to develop the council and eventually would help us to society with each other in the council. We also hope this council would be the one that connect all of us in society.

• 전기전자학회논문지
• /
• 제20권1호
• /
• pp.89-93
• /
• 2016

본 논문에서는 다양한 영상 전처리 알고리즘들을 적용할 수 있고, 영상 인식과 같이 ISP 응용 프로그램을 병렬로 처리 가능한 SIMT(Single Instruction Multiple Threads) 기반 Image Signal Processor를 제안하였다. 기존의 ISP는 입력 영상의 품질 개선을 위하여 영상 개선 알고리즘이 하드웨어로 설계되어 처리 속도는 빠르지만 다양한 영상 처리 알고리즘에 따라 성능 최적화에 어려움이 있었다. 제안한 ISP는 명령어를 기반으로 한 프로세서로서 다양한 영상 처리 알고리즘을 수행하고 SIMT 구조를 적용하여 알고리즘을 병렬로 수행해 성능을 개선하였다. 제안하는 ISP를 검증하기 위해 Xilinx Virtex-7을 탑재한 VC707 Board를 사용하였으며 cell multicore processor와 비교했을 경우 수행시간이 약 71%, ARM Cortex-A9과 ARM Cortex-A15와 비교하였을 경우 각각 63%, 33% 성능을 개선하였다.

• 천문학논총
• /
• 제32권1호
• /
• pp.197-199
• /
• 2017

Using the InfraRed Camera (IRC) on board the infrared astronomical satellite AKARI we study the ${3.3{\mu}m}$ polycyclic aromatic hydrocarbon (PAH) feature and its connection to active galactic nucleus (AGN) properties for a sample of 54 hard X-ray selected bright AGN, including both Seyfert 1 and Seyfert 2 type objects. The sample is selected from the 9-month Swift/BAT survey in the 14-195 keV band and all of the sources have known neutral hydrogen column densities ($N_H$). The ${3.3{\mu}m}$ PAH luminosity ($L_{3.3{\mu}m}$) is used as a proxy for star-formation (SF) activity and hard X-ray luminosity ($L_{14-195keV}$) as an indicator of the AGN power. We explore for possible difference of SF activity between type 1 (un-absorbed) and type 2 (absorbed) AGN. We use several statistical analyses taking the upper-limits of the PAH lines into account utilizing survival analysis methods. The results of our log($L_{14-195keV}$) versus log($L_{3.3{\mu}m}$) regression shows a positive correlation and the slope for the type 1/unobscured AGN is steeper than that of type 2/obscured AGN at a $3{\sigma}$ level. Also our analysis shows that the circum-nuclear SF is more enhanced in type 2/absorbed AGN than type 1/un-absorbed AGN for low $L_{14-195keV}$ luminosity/low Eddington ratio AGN, while there is no significant dependence of SF activity on the AGN type in the high $L_{14-195keV}$ luminosities/Eddington ratios.

• 보존과학회지
• /
• 제34권5호
• /
• pp.359-368
• /
• 2018

본 연구에서는 경기도 안산시 단원구 대부도에서 발굴된 대부도2호선의 목부재와 초본류의 수종분석과 방사성탄소연대분석을 통해 사용된 목재의 재질과 고선박이 난파된 시기를 추정하고자 하였다. 수종분석을 통해 선체부재로 사용된 목재는 모두 소나무속의 경송류로 식별되었으며, 기타 선체부재와 나무못은 경송류, 밤나무속, 참나무속 상수리나무아속 상수리나무류, 느티나무, 오리나무속의 사용을 확인하였다. 동반 출토된 목제빗과 참빗은 단풍나무속과 대나무아과로 확인되었고, 밧줄은 칡으로 만들어졌음이 밝혀졌다. 선수재와 받침목 그리고 씨앗류(감씨)의 방사성탄소연대측정과 위글매치 분석을 통해 대부도2호선은 AD 1151~1224년 사이에 난파된 것으로 밝혀졌다. 이 고선박은 12~13세기 초중반에 난파되었음을 시사한다. 이는 출토된 도자기의 편년과 일치하는 결과이다.

• 한국언론정보학보
• /
• 제16권
• /
• pp.207-254
• /
• 2001

This study is focused on forecasting the future of tole-democracy. Many Scholars expect that internet provides technological space needing realizing the tole-democracy or deliberate democracy. Especially in Korea, this expectation is higher than other nations because of political corruption and inefficiency. Therefore internet is intended to considering as new technologies reforming political process. In 16th the general election period, many candidates established home page and used as election campaign tool. And a little of home pages is payed attention to among voters. In spite of using internet as political medium, many political communication researchers have a doubt that internet will realize ideal direct democracy. It's reason is that internet is open and anonymous space. At anonymous space, communication participators is tended to be irresponsible and non-serious. Therefore it is hard that cyber-politics will be ideal type of democracy. In this context, this paper analyzed how pauicipators communicate with others at cyber bulletin board establishing candidate's home pages. Main research questions is how do discussions at cyberspace fulfill the conditions of deliberate democracy. Therefore, concrete questions include; who are participators at candidate's cyber bulletin board; which pattern do they communicate; what is the theme of communication; which effects do the anonymous character of cyberspace influence. To that purpose, this study conducted content analysis on 4,210 written matters at 82 cyber bulletin boards of candidate's home page establishing during 16th the general election period. It can be found that cyberspace in Korea still is not deliberate democracy space and not will be. Firstly, discussion patterns at cyber bulletin board is "candidate with supporters communication space. To be exact, it is said that candidate's home page is "the space of self-convincing among supporters." Secondly, the main themes of discussion are simple emotional expressions; "I support you!" "fighting" "Be vigor" etc. By contrast, real political contents'-central or local political matters - is relatively few. In the mode of expression, real political messages are more positive, logical than simple expressions, candidates private matters. Especially this characteristic will make cyberspace as "mutual slander space" consolidating anonymous characteristic of cyberspace. finally, Cyberspace in Korea still is not real "public sphere" realizing deliberate process. Therefore to be real public sphere, it is needed to participant's ethical maturity and political citizenship. In conclusion, it is difficult that cyberspace will reconstruct the Athene's Agora. On the contrary, Cyberagora will like to be irrespectable area fulfilling the sweeping. Making the deliberate space, technological possibility and ethical condition will have to be balanced together.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2003년도 ICCAS
• /
• pp.1295-1299
• /
• 2003

The Mass Flow Controller(MFC) has become crucial in semiconductor manufacturing equipments. It is an important element because the quality and the yield of a semiconductor process are decided by the accurate flow control of gas. Therefore, the demand for the high speed and the highly accurate control of MFCs has been requested. It is hard to find an article of the control algorithm applied to MFCs. But, it is known that commercially available MFCs have PID control algorithms. Particularly, when the system detects the flow by way of heat transfer, MFC control problem contains the time delay and the nonlinearity. In this presentation, MFC control algorithm with the superior performance to the conventional PID algorithm is discussed and the superiority is demonstrated through the experiment. Fuzzy controller was utilized in order to compensate the nonlinearity and the time delay, and the performance is compared with that of a product currently available in the market. The control system, in this presentation, consists of a personal computer, the data acquisition board and the control algorithm carried out by LabWindows/CVI program on the PC. In addition, the method of estimating an actual flow from sensor output containing the time delay and the nonlinearity is presented. In conclusion, according to the result of the experiment, the proposed algorithm shows better accuracy and is faster than the conventional controller.