• 한국해안해양공학회지
• /
• 제11권1호
• /
• pp.7-19
• /
• 1999

비선형 상호작용에 의한 풍파 성분간 에너지 전달은 스펙트럼의 발달에서 중요한 역할을 한다. 비선형 전달을 표현하는 Boltzmann적분을 계산하는 데에는 방대한 계산시간이 필요하기 때문에 파랑모형에서 비선형 상호작용을 고려하는 것을 불가능하다. 본 연구에서는 산란계수의 특성과 상호작용의 상세균형의 원리를 이용하여 비선형 전달을 효율적으로 계산하는 방법에 대하여 고찰하였다. Webb의 방법(IWm)은 특이점을 퇴화시켜 매우 안정한 계산 결과를 주지만 상세균형의 원리를 적용할 수 없기 때문에 계산의 효율성에는 한계가 있는 것으로 나타났다. 한편, Masuda의 방법(IMM)에서는 특이점을 해석적으로 처리하며 계산시간은 Pentinum 300MHz Processor에서 1.3초가 소요되었다. 따라서 IMM은 1차원 파랑모형에 실용적으로 사용할 수 있으며 취송시간과 취송거리에 의한 풍파 스펙트럼의 성장 과정 연구등에 매우 유용하다.

• 융합정보논문지
• /
• 제9권8호
• /
• pp.248-253
• /
• 2019

노인성 치매의 전 임상단계인 경도인지장애(MCI)를 조기 진단하고, 조기 개입한다면, 치매의 발병률을 줄일 수 있다. 본 연구는 우리나라 지역사회 노인의 MCI 예측 모형을 개발하고 노년기 인지장애의 예방을 위한 기초자료를 제공하였다. 연구대상은 2012년 Korean Longitudinal Survey of Aging(KLoSA)에 참여한 65세 이상 지역사회 노인 3,240명(남성 1,502명, 여성 1,738명)이다. 결과변수는 MCI유병으로 정의하였고, 설명변수는 성, 연령, 혼인상태, 교육수준, 소득수준, 흡연, 음주, 주1회 이상의 정기적인 운동, 월평균 사회활동 참여시간, 주관적 건강, 고혈압, 당뇨병을 포함하였다. 예측모형의 개발은 Restricted Boltzmann Machine(RBM) 인공신경망을 이용하였다. RMB 인공신경망을 이용하여 우리나라 지역사회 노인의 MCI 예측 모형을 구축한 결과, 유의미한 요인은 연령, 성별, 최종학력, 주관적 건강, 혼인상태, 소득수준, 흡연, 규칙적 운동이었다. 이 결과를 기초로 MCI 고위험군의 특성을 고려한 맞춤형 치매 예방 프로그램의 개발이 요구된다.

• 한국정보통신학회논문지
• /
• 제26권2호
• /
• pp.181-186
• /
• 2022

포아즌 볼츠만 방정식 (Poisson-Boltzmann equation, PBE)은 생물물리, 콜로이드 화학 등에서 등장하는 문제들을 모델링하는데 사용되는 방정식이다. 따라서 PBE의 수치해를 효율적으로 예측하는 것은 중요한 이슈이다. 저자들은 기존의 연구에서 PBE를 풀기위한 딥러닝 방법을 제안하였으나, 딥러닝을 훈련하기 위한 샘플을 생성하는 시간이 컸다는 어려움이 있었다. 본 논문에서는 FEM 수치해를 생성하는데 걸리는 시간을 줄이는 두가지 방안을 마련하였다. 첫째로 대수 방정식을 만들 때 bilinar form에 포함되는 penalty 파라메터를 실험적으로 조정하였다. 두 번째로, 대수적멀티그리드 기법을 활용하여 대수 방정식의 컨디션 넘버를 meshsize와 무관하게 만들었다. 따라서 PBE 방정식의 대수 방정식을 풀 때 계산 시간을 효과적으로 줄였다. 이러한 대수적 멀티그리드를 사용한 방법은 다양한 분야에서 딥러닝의 샘플을 생성하는데 효과적으로 활용될 수 있을 것으로 기대된다.

• 대한기계학회:학술대회논문집
• /
• 대한기계학회 2007년도 춘계학술대회B
• /
• pp.2797-2802
• /
• 2007

It has been confirmed that implementation of the no-slip boundary conditions for the lattice-Boltzmann method play an important role in the overall accuracy of the numerical solutions as well as the stability of the solution procedure. We in this paper propose a new algorithm, i.e. the method of the dynamic boundary condition for no-slip boundary condition. The distribution functions on the wall along each of the links across the physical boundary are assumed to be composed of equilibrium and nonequilibrium parts which inherit the idea of Guo's extrapolation method. In the proposed algorithm, we apply a dynamic equation to reflect the computational slip velocity error occurred on the actual wall boundary to the correction; the calculated slip velocity error dynamically corrects the fictitious velocity on the wall nodes which are subsequently employed to the computation of equilibrium distribution functions on the wall nodes. Along with the dynamic selfcorrecting process, the calculation efficiently approaches the steady state. Numerical results show that the dynamic boundary method is featured with high accuracy and simplicity.

• 한국전산유체공학회지
• /
• 제16권2호
• /
• pp.24-29
• /
• 2011

Immersed boundary lattice Boltzmann method has been applied to analyze the characteristics of the self-propelled fish motion swimming robot. The airfoil NACA0012 with caudal fin stroke model was considered to examine the characteristics. The foil in steady forward motion and a combination of steady-state harmonic deformation produces thrust through the formation of a flow downstream from the trailing edge. The harmonic motion of the foil causes unsteady shedding of vorticity from the trailing edge, while forming the vortices at the leading edge as well. The resultant thrust is developed by the pressure difference formed on the upper and lower surface of the airfoil. and the time averaged thrust coefficient increases as Re increase in the region of $Re{\leqq}700$. The suggested numerical method is suitable to develop the fish-motion model to control the swimming robot, however It would need to extend in 3D analysis to examine the higher Re and to determine the more detail mechanism of thrust production.

• Journal of Advanced Marine Engineering and Technology
• /
• 제34권1호
• /
• pp.53-61
• /
• 2010

In the present study, a laminar natural convection flow of $H_2O$-Cu nanofluid in a two dimensional enclosure has been investigated using a thermal lattice Boltzmann approach with the Bhatnagar-Gross-Krook (BGK) model. The effect of suspended nanoparticles on the fluid flow and heat transfer process have been studied for different controlling parameters such as particle volume fraction ($\Phi$), Rayleigh number (Ra). For this investigation the Rayleigh number changes from 104 to 106 and volume fraction varied from 0 to 10% with three different particle diameters (dp), say 10 nm, 20 nm and 40 nm. It is shown that increasing the Rayleigh number (Ra) and the volume fraction of nanofluid causes an increase of the effective heat transfer rate in terms of average Nusselt number (Nu) as well as the thermal conductivity of nanofluid. On the other hand, increasing the particle diameter causes the decrease of the heat transfer rate and thermal conductivity. The result of the analysis are compared with experimental and numerical data both for pure and nanofluids and it is seen a relatively good agreement.

• 한국전산유체공학회:학술대회논문집
• /
• 한국전산유체공학회 2009년 추계학술대회논문집
• /
• pp.97-102
• /
• 2009

Translocation of biopolymers such as DNA and RNA through a nano-pore is an important process in biotechnology applications. The translocation process of a biopolymer through an artificial nano-pore in the presence of a fluid solvent is simulated. The polymer motion is simulated by Langevin molecular dynamics (MD) techniques while the solvent dynamics are taken into account by lattice-Boltzmann method (LBM). The hydrodynamic interactions are considered explicitly by coupling the polymer and solvent through the frictional and the random forces. From simulation results we found that the hydrodynamic interactions between polymer and solvent speed-up the translocation process. The translocation time ${\tao}_T$ scales with the chain length N as ${{\tau}_T}^{\propto}N^{\alpha}$. The value of scaling exponents($\alpha$) obtained from our simulations are $1.29{\pm}0.03$ and $1.41{\pm}0.03$, with and without hydrodynamic interactions, respectively. Our simulation results are in good agreement with the experimentally observed value of $\alpha$, which is equal to $1.27{\pm}0.03$, particularly when hydrodynamic interaction effects are taken into account.

• Journal of Advanced Marine Engineering and Technology
• /
• 제33권1호
• /
• pp.62-70
• /
• 2009

In this paper, the fluid flow behavior past a pair of rectangular cylinders placed in a two dimensional horizontal channel has been investigated using Lattice-Boltzmann Method(LBM). The LBM has built up on the D2Q9 model and the single relaxation time method called the Lattice-BGK(Bhatnagar-Gross-Krook)model. Streamlines, velocity, vorticity and pressure contours are provided to analyze the important characteristics of the flow field for a wide range of non dimensional parameters that present in our simulation. Special attention is paid to the effect of spacing(d) between two cylinders and the blockage ratio A(=h/H), where H is the channel height and h is the rectangular cylinder height. for different Reynolds numbers. The first cylinder is called upstream cylinder and the second one as downstream cylinder. The downstream fluid flow fields have been more influenced by its blockage ratios(A) and Reynolds numbers(Re) whereas the upstream flow patterns(in front of downstream cylinder) by the gap length(d) between two cylinders. Moreover, it is observed that after a certain gap, both upstream and downstream flow patterns are almost similar size and shape. The simulation result has been compared with analytical solution and it is found to be in excellent agreement.

• 한국지능시스템학회논문지
• /
• 제11권6호
• /
• pp.505-509
• /
• 2001

본 연구에서는 학습기근을 갖는 결정론적 볼츠만 머신의 은닉충 뉴런에 비단조 활성화 함수를 적요한 경위의 학습성능을 XOR 문제와 ADD 학습에 대하여 수지 시뮬레이션을 통하여분석한다. 단조 활성화함수를 사용한 경우와 비교하여 학습 수렴률, 학습안정도, 및 학습 속도에 있어서 성능이 크게 향상됨을 확인하였다. 또한 네트워크의 막전위 분포를 조사함으로서 end-cut-off 타입의 비단조 함수를 이용한 경우에 나타는 다음 층의 뉴런에 영향을 주지 않는 뉴런의 출현, 즉, 신경회로망에 있어서 은닉층 뉴런늬 수을 자율적으로 조정하는것을 확인하였따. 이것은 학습문제에 대하여 네트워크 은닉층 뉴런의 수를 명확하게 결정할수 없는 현재의 상황에 있어서는 새로운 돌파구가 될것으로 기대된다.

• 전기학회논문지P
• /
• 제64권4호
• /
• pp.241-245
• /
• 2015

Energy Distribution Function in pure $CH_4$, $CF_4$ and mixtures of $CF_4$ and Ar, have been analyzed over a range of the reduced electric field strength between 0.1 and 350[Td] by the two-term approximation of the Boltzmann equation (BEq.) method and the Monte Carlo simulation (MCS). The calculations of electron swarm parameters require the knowledge of several collision cross-sections of electron beam. Thus, published momentum transfer, ionization, vibration, attachment, electronic excitation, and dissociation cross-sections of electrons for $CH_4$, $CF_4$ and Ar, were used. The differences of the transport coefficients of electrons in $CH_4$, mixtures of $CH_4$ and Ar, have been explained by the deduced energy distribution functions for electrons and the complete collision cross-sections for electrons. The results of the Boltzmann equation and the Monte Carlo simulation have been compared with the data presented by several workers. The deduced transport coefficients for electrons agree reasonably well with the experimental and simulation data obtained by Nakamura and Hayashi. The energy distribution function of electrons in $CF_4$-Ar mixtures shows the Maxwellian distribution for energy. That is, $f({\varepsilon})$ has the symmetrical shape whose axis of symmetry is a most probably energy.