http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
손창현(C. H. Sohn) 한국전산유체공학회 1997 한국전산유체공학회지 Vol.2 No.2
Two versions of anisotropic k-ε turbulence model are incorporated in the modi fied k-? model of Sohn et a1. to avoid the need for the experimental normal stress value in the model and applied to convergent and divergent flows wi th strong and adverse pressure gradients in the plane of symmetry of a body of revolution. The models are the nonlinear k-ε model of Speziale and the anisotropic model of Nisizima & Yoshizawa. All of the models yield satisfactory results for relatively complex flow on a plane-of-symmetry boundary layer. The results of the models are compared wi th those results of experimental normal stress value.
손창현(C. H. Sohn) 한국전산유체공학회 1997 한국전산유체공학회지 Vol.2 No.2
Two versions of anisotropic k-ε turbulence model are incorporated in the modi fied k-? model of Sohn et a1. to avoid the need for the experimental normal stress value in the model and applied to convergent and divergent flows wi th strong and adverse pressure gradients in the plane of symmetry of a body of revolution. The models are the nonlinear k-ε model of Speziale and the anisotropic model of Nisizima & Yoshizawa. All of the models yield satisfactory results for relatively complex flow on a plane-of-symmetry boundary layer. The results of the models are compared wi th those results of experimental normal stress value.
실내 화재 검출 정확도 개선을 위한 데이터 증강 기반 퓨샷 러닝 최적화 방법
이준목(Jun-Mock, Lee),강대성(Dae-Seong, Kang) 한국정보기술학회 2019 Proceedings of KIIT Conference Vol.2019 No.11
딥러닝을 활용하여 객체를 검출하는 기술은 다양한 방면으로 가파른 발전을 거듭하고 있다. 딥러닝 모델의 정확도는 양질의 학습데이터와 연관이 있는 것은 이미 널리 알려진 사실이다. 양질의 학습데이터 확보가 어려운 범죄, 실내 화재 등을 검출할 딥러닝 모델을 개발해야 하는 경우 모델의 정확도를 향상하는 데에는 한계점이 존재한다. 본 논문에서는 학습데이터의 수집이 제한된 상황에서 데이터 증강을 통해 학습데이터를 스스로 생성하여 검출 모델의 정확도를 향상하는 방법을 제안한다. 제안한 방법을 통해 대량의 데이터로 학습된 모델과 유사한 검출 정확도를 보여주며, 적은 데이터로 학습하기 때문에 학습 소요 시간 또한 단축되었음을 실험을 통해 보여준다.
부력의 영향을 포함한 점탄성 유체의 열전달에 관한 수치해석
손창현(C. H. Sohn),안성태(S. T. Ahn),장재환(J. H. Jang) 한국전산유체공학회 1998 한국전산유체공학회 학술대회논문집 Vol.1998 No.-
The present numerical study investigates flow characters and heat transfer enhancement by the viscoelastic-driven secondary flow and buoyancy effect in a 2:1 rectangular duct. Three versions of thermal boundary conditions involving difference combination of heated walls and adiabatic walls are analyzed in this study. The Reiner-Rivlin model is adopted as a viscoelastic fluid model to simulate the secondary flow and temperature-dependent viscosity model is used. Calculated Nusselt numbers are very good agreement with experimental results for reported viscoelastic fluids, It is found that the heat transfer enhancement is mainly caused by the viscoelastic-driven secondary flow and buoyancy-induced secondary flow playa role of promoting this effect.
손동현(D.H.Sohn),손창현(C.H. Sohn),Gowda 한국자동차공학회 2007 한국자동차공학회 지부 학술대회 논문집 Vol.- No.-
In the present study the reverse flow phenomenon in a square duct is investigated using PIV. The flow features are much different in this case compared to the two-dimensional channel as we are dealing with the three dimensional square duct flow (side:b) and the obstruction now is a square plate. The flow at the rear end of the duct where reverse flow enters, reveals features of three dimensional vortical flow and shear layer interactions. Figure 1 shows the flow field at the entrance and rear ends of the duct for a typical case. The detailed flow field including animation and PIV measurements will be presented in the final version of the paper.
손동현,손창현,하재훈,Son, D.H.,Sohn, C.H.,Ha, J.H. 한국군사과학기술학회 2010 한국군사과학기술학회지 Vol.13 No.6
This paper presents a numerical evaluation of the launch dynamics and thermo-fluid phenomena for gas generator launch eject system. The existing gas dynamic model for launching eject body used ideal gas and adiabatic assumption with empirical energy loss model. In present study, a turbulent Navier-Stokes solver with CHIMERA mesh is employed to predict the detail unsteady thermo-fluid dynamics for the launched body. The calculation results show that proper grid number is necessary for good agreement with experimental data. The important effects for accurate prediction are a gap distance and thermal boundary condition on the wall. The computational results show good agreement with experiment data.