• Proceedings of the KIEE Conference
• /
• 2006.10d
• /
• pp.215-217
• /
• 2006

본 논문에서는 잉크젯 프린팅 기술을 이용한 산업용 FPCB의 패턴형성을 위한 잉크드롭의 최적형성을 연구목적으로 하였다. 이를 위하여 압전헤드에 대한 구동특성을 해석하였으며 드롭형성을 위한 전압파형의 크기, 펄스폭, 기울기(감속)조정을 통하여 잉크드롭을 최적으로 형성하도록 제어기(FPGA)를 구성하였으며 실험결과를 통하여 잉크의 최적형성 과정을 확인할 수 있었으며 잉크 드롭시 1[mm]에 ${\pm}10$[um]의 드롭오차를 ${\pm}2$[um]까지 감소시켜 50[um]패턴을 프린트하였다.

• Proceedings of the KIEE Conference
• /
• 2009.07a
• /
• pp.1205_1206
• /
• 2009

고속철도 전차선로에 설치된 드롭퍼의 와이어가 파단되어 원인을 분석하기 위해 드롭퍼의 구간별/설치위치별/지역별 단선현황 분석, 파단면 SEM(Scanning Electron Microscope) 분석, 미세조직 분석 및 드롭퍼에 작용하는 정적하중을 계산하였다. 또한 고속철도 전차선로에서 드롭퍼의 정적 및 동적하중 측정과 분석을 통하여 드롭퍼의 피로하중 특성을 확인하였고, 지지점의 압상량을 측정 및 분석하였다.

• Transactions of the Korean Society of Mechanical Engineers
• /
• v.14 no.4
• /
• pp.830-838
• /
• 1990

본 연구에서는 에지 드롭 현상이 가장 심하게 나타나고 있는 M-class 저급 무 방향성 규소 강판을 대상으로 MKW압연기의 롤 탄성변형 해석을 컴퓨터 시뮬레이션을 통하여 수행함으로써 에지 드롭을 최소화할 수 있는 개선책을 마련하였고 실제 현장 압연실험을 통하여 그 결과를 확인하였다.

• Proceedings of the Optical Society of Korea Conference
• /
• 2002.07a
• /
• pp.166-167
• /
• 2002

광자 크리스탈은 빛의 진행을 제어할 수 있는 잠재적인 능력과 Wavelength Division Multiplexing 통신 시스템에서 광소자의 구현 가능성 때문에 활발히 연구가 진행 중이다. (1-3) 본 논문에서는 광자 크리스탈의 채널 드롭 터널링 현상을 이용하여 1x4 Demultiplexer의 구현 가능성을 고찰하였다. 광자 크리스탈을 이용해서 공진 시스템을 구성하였고, 중앙에 위치한 2개의 작은 디펙트의 크기를 조절하여 채널 드롭 현상을 관찰하였다. (중략)

• Proceedings of the KIEE Conference
• /
• 2008.07a
• /
• pp.1071-1072
• /
• 2008

본 논문에서는 고속철도 전차선로 드롭퍼 와이어의 파단현상에 대해 분석을 통하여 철도운행의 안정성 향상을 위한 기초연구를 수행하였다. 300km/h로 주행하는 KTX의 판토그래프와 전차선이 직접 접촉하면서 발생되는 기계적 피로도의 누적으로 인한 증가로 경부고속선 전차선로 시스템의 장애사고 예방을 목적으로 하고 있다. 추후, 고속철도 전차선로 드롭퍼 수명한계 측정으로 유지보수관리 체계 효율화와 분석된 파단원인을 바탕으로 드롭퍼 구조개선 방안을 마련하여 합리적 개량방안 수립을 하고자 한다.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.54 no.4
• /
• pp.3-9
• /
• 2017

Due to difficulties of continuous electric power provision to underwater communication nodes, the efficient power usage is highly required in underwater network protocol. In this paper, we studied the energy efficient MAC(Medium Access Control) protocol for underwater network supporting mobile nodes such as UUV(Unmanned Underwater Vehicle) and AUV(Autonomous Underwater Vehicle). The mobile nodes could waste the electric power in vain when the receiver moves out of the radio propagation coverage during the data exchange and thus the transmitted data fails in reaching the receiver. Expecially, such a failure is much more obvious in underwater acoustic channels since the propagation delay is about $10^5$ times slower than in terrestrial radio channels. This proposed mobility-MAC controls the data dropping stochastically in the Dropping Zone by considering the receiver's location and moving velocity. In conclusion, this selective dropping method not only improves latency and throughput by reducing invalid droppings but also boosts power efficiency by valid droppings.

• KIPS Transactions on Software and Data Engineering
• /
• v.6 no.12
• /
• pp.573-580
• /
• 2017

Deep Neural Network (DNN) is a large layered neural network which is consisted of a number of layers of non-linear units. Deep Learning which represented as DNN has been applied very successfully in various applications. However, many issues in DNN have been identified through past researches. Among these issues, generalization is the most well-known problem. A Recent study, Dropout, successfully addressed this problem. Also, Dropout plays a role as noise, and so it helps to learn robust feature during learning in DNN such as Denoising AutoEncoder. However, because of a large computations required in Dropout, training takes a lot of time. Since Dropout keeps changing an inter-layer representation during the training session, the learning rates should be small, which makes training time longer. In this paper, using mutation operation, we reduce computation and improve generalization performance compared with Dropout. Also, we experimented proposed method to compare with Dropout method and showed that our method is superior to the Dropout one.

• The Korean Journal of Applied Statistics
• /
• v.32 no.6
• /
• pp.899-908
• /
• 2019

Massive in-depth neural networks with numerous parameters are powerful machine learning methods, but they have overfitting problems due to the excessive flexibility of the models. Dropout is one methods to overcome the problem of oversized neural networks. It is also an effective method that randomly drops input and hidden nodes from the neural network during training. Every sample is fed to a thinned network from an exponential number of different networks. In this study, instead of feeding one sample for each thinned network, two or more samples are used in fitting for one thinned network known as a Hybrid Dropout. Simulation results using real data show that the new method improves the stability of estimates and reduces the minimum error for the verification data.

• 한국체육학회지인문사회과학편
• /
• v.51 no.4
• /
• pp.419-425
• /
• 2012

The purpose of the study was to investigate the effects of knee brace on anterior cruciate ligament injuries risk factors during one-legged landing of female gymnasts. Eleven female gymnasts were recruited and performed randomly one-legged drop landing in height of her's knee with (3times) and without (3times) knee brace. Kinematics and ground reaction data were collected to estimate the anterior cruciate ligament injuries risk factors. Data were analyzed with paired samples t-test. Female gymnasts with knee brace showed more reduced the distance from ankle joint center to knee joint center in sagittal plane and knee maximum joint torque than without knee brace. In conclusion, Female gymnasts with knee brace reduced anterior cruciate ligament stress.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.17 no.8
• /
• pp.714-721
• /
• 2016

This paper proposes a simple process to fabricate tin and brass metal discs with a large surface area from molten droplets for the wet-refining process of nonferrous metals by assuming they have precious metal elements. To optimize the droplet condition in a graphite crucible, the appropriate nozzle size was determined using a simulation program (STAR-CCM+) by varying the diameters (0.5, 1.0, and 2.0 mm). The simulation results showed that both tin and brass do not fall out with a 0.5 mm diameter nozzle but they do fall out in continuous ribbon mode with a 2.0 mm nozzle. Only the 1.0mm nozzle was expected to fabricate droplets. Finally, solidified metal discs were fabricated successfully with the 1.0 mm nozzle within 10 minutes by impacting the droplets with a cooling water flowing over a Ti plate placed at the $40^{\circ}$ falling direction. The weight, average thickness, and surface area of the tin discs were 0.15 g, $107.8{\mu}m$, and $3.71cm^2$, respectively. The brass discs were 1.16 g, $129.15{\mu}m$, and $23.98cm^2$, respectively. The surface area of the tin and brass disc were 8.2 and 17.6 times the size of the tin and brass droplets, respectively. This process for precious metal extraction is expected to save cost and time.