http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Back‐propagation based nonbinary optical code for OCDMA with fiber gratings
Min, Seong‐,Sik Wiley Subscription Services, Inc., A Wiley Company 2007 MICROWAVE AND OPTICAL TECHNOLOGY LETTERS - Vol.49 No.3
<P><B>Abstract</B></P><P>A new optical code is proposed for OCDMA networks with optical gratings and balanced detection, in which the back‐propagation algorithm of a neural network is applied to nonbinary optical code generation. The proposed codes have high security level and a code ratio of at least 1.0. © 2007 Wiley Periodicals, Inc. Microwave Opt Technol Lett 49: 632–636, 2007; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/mop.22213</P>
리튬 원료에 따른 리튬이온전지 Ni-rich 단결정 양극재 건식합성법 연구
김민성(Minseong Kim),고민성(Minseong Ko) 한국전지학회 2022 한국전지학회지 Vol.2 No.1
리튬이온전지 NCM계 단결정 양극재는 높은 에너지 밀도와 우수한 용량 유지율을 바탕으로 활발한 연구가 이루어지고 있다. 그러나 일반적인 단결정은 공침법을 통해 합성되며 다량의 폐용액 발생, 복잡한 공정 등을 수반할 수 있다. 여기서 우리는 공침법을 활용하지 않고 고상합성법을 통해 NCM계 단결정을 합성하였다. 각각의 원료로서 전이금속 산화물만을 이용하였으며, 리튬원료에 따른 단결정을 비교 분석하였다. Single-crystal cathode materials (SCM) of Ni-rich layered oxides (LiNi<SUB>x</SUB>Co<SUB>y</SUB>Mn<SUB>z</SUB>O₂ (NCM), x+y+z=1, x≥0.8) is considered as high energy density and outstanding capacity retention for lithium-ion batteries. However, typically synthesis of SCM by coprecipitation have some problems such as, waste fluid, complicated process, and micro-crack in polycrystal morphology. Herein, we introduced pelletizing process for SCM synthesis without coprecipitated precursor (NCM(OH)₂), using transition metal oxides (NiO, Co₃O₄, and MnO₂) and lithium source (LiOH or Li₂CO₃). Pelleted precursor enables to lower sintering temperature by increasing contact area of particles. These SCM represented well-arranged layered structure. In particular, SCM using lithium carbonate exhibited a high discharge capacity of 191.1 mAh/g (0.1C, 3.0-4.3 V cut-off) and excellent capacity retention of 93.6% after 50 cycles.
탄소나노튜브/흑연펠트 전극의 산소작용기를 활용한 바나듐 레독스 흐름 전지의 수소발생 억제 효과
김민성(Minseong Kim),고민성(Minseong Ko) 한국표면공학회 2021 한국표면공학회지 Vol.54 No.4
Vanadium redox flow batteries (VRFB) have emerged as large-scale energy storage systems (ESS) due to their advantages such as low cross-contamination, long life, and flexible design. However, Hydrogen evolution reaction (HER) in the negative half-cell causes a harmful influence on the performance of the VRFB by consuming current. Moreover, HER hinders V<SUP>2+</SUP>/V<SUP>3+</SUP> redox reaction between electrode and electrolyte by forming a bubble. To address the HER problem, carbon nanotube/graphite felt electrode (CNT/GF) with oxygen functional groups was synthesized through the hydrothermal method in the H₂SO₄ + HNO₃ (3:1) mixed acid solution. These oxygen functional groups on the CNT/GF succeed in suppressing the HER and improving charge transfer for V<SUP>2+</SUP>/V<SUP>3+</SUP> redox reaction. As a result, the oxygen functional group applied electrode exhibited a low overpotential of 0.395 V for V<SUP>2+</SUP>/V<SUP>3+</SUP> redox reaction. Hence, this work could offer a new strategy to design and synthesize effective electrodes for HER suppression and improving the energy density of VRFB.
Education Competition and Redistribution
Minseong Kim,Bong Je Choi 한국경제연구학회 2010 Korea and the World Economy Vol.11 No.2
The paper develops a general equilibrium model of education competition where entry into the higher education sector is a costly choice due to limited accommodation. We show that wealth distribution plays a crucial role in determining the extent of education competition. Redistribution via a simple tax-and-subsidy program may increase or decrease the extent of education competition depending on the initial configuration of the equilibrium. We also construct a measure of college premium and show that the redistribution program reduces the relative college premium for rich individuals.
Minseong Kim,Byung Gi Kim,김진영,Woongsik Jang,Dong Hwan Wang 한국공업화학회 2020 Journal of Industrial and Engineering Chemistry Vol.88 No.-
A split-ligand mediated re-precipitation (Split-LMRP) technique for colloidal perovskite quantum dots(QDs) was designed by separately dissolving rich oleic acid (OA) and amine ligands in the synthesisprocess. OA was used to control the polarity of the nucleation environment and was simultaneouslyemployed as a stabilizer. The Split-LMRP technique allowed purification via the precipitation of QDs froma colloidal solution. The fabrication process is performed under ambient conditions, and the resultingCH3NH3PbX3 (X = Br, I) QDs exhibited strong photoluminescence (PL) emission with a maximum PLquantum yield of 91.5%. The size of the resulting perovskite QDs is tuned in the range of 2–5 nm byvarying the ligand concentration and type of halide. We also investigated the charge-transport propertiesof the synthesized QDs using space-charge-limited current analysis and confirmed stable charge carriermobility even when the QDs solution was spin-coated on a hydrophilic poly (3,4-ethylenedioxythio-phene) polystyrene sulfonatefilm. Furthermore, the enhanced stability of CH3NH3PbX3 (X = Br, I) QDsimproved the power conversion efficiency by the uniform surface passivation of the perovskite activelayer which induces efficient exciton generation and charge transport.
Optimizing TensorFlow Performance by Reconstructing the Convolution Routine
Minseong Kim,Kyu Hyun Choi,Yoonah Paik,Seon Wook Kim 대한전자공학회 2021 IEIE Transactions on Smart Processing & Computing Vol.10 No.2
Using deep learning, we can currently build computational models composed of multiple processing layers to learn representations of data. Convolutional neural networks (CNNs) have been widely adopted to achieve significant performance in image recognition and classification. TensorFlow, an open-source deep learning framework from Google, uses profiling to select one convolution algorithm, from among several available, as the core of a CNN to deliver the best performance in terms of execution time and memory usage. However, the overhead from profiling is considerably significant, because TensorFlow executes and profiles all the available algorithms for the best selection whenever an application is launched. We observe that memory usage overshoots during profiling, which limits data parallelism, and thus, fails to deliver maximum performance. In this paper, we present a novel profiling method to reduce overhead by storing the profile result from the first run and reusing it from the second run on. Using Inception-V3, we achieved up to 1.12 times and 1.11 times higher throughput, compared to the vanilla TensorFlow and TensorFlow with XLA JIT compilation, respectively, without losing accuracy.
Pretreatment of Silicone Coated PET Release Film for the Recycling of PET
( Minseong Kang ),( Yunsu Kim ),( Do Hyun Kim ) 한국폐기물자원순환학회(구 한국폐기물학회) 2019 ISSE 초록집 Vol.2019 No.-
Polyethylene terephthalate (PET) release film is a functional film that acts as a material barrier to protect the surface of commercial products such as window, semiconductor wafer, mobile phone screen and label tape. Usually, PET release film is composed of multiple layers to enhance the physical properties and grant various functionality. Especially, PET film coated with polydimethylsiloxane (PDMS) is widely used among PET released film. However, recycling of silicone coated PET film is difficult without removing silicone layer from the film. In this work, we propose pretreatment process to remove silicone layer from the PET release film by swelling and physical detachment. To swell PDMS layer on a PET film, acetone was used. Then, glass beads were added to physically break PDMS surface. After the agitation of mixture at room temperature, swollen PDMS waste and glass beads were filtered from the mixture. Treated PET film was then put in acetone to further rinse the PDMS waste left on the surface. From energy dispersive X-ray spectroscopy (EDS) analysis, the amount of silicon element on the surface of PET film decreased by 98.7%, suggesting that PDMS layer was successfully removed from PET film.