• 한국전산유체공학회:학술대회논문집
• /
• 한국전산유체공학회 1999년도 추계 학술대회논문집
• /
• pp.17-22
• /
• 1999

Convergence characteristics and efficiency of three implicit approximate factorization schemes(ADI, DDADI and MAF) are examined using 2-Dimensional compressible upwind Navier-Stokes code. Second-order CSCM(Conservative Supra Characteristic Method) upwind flux difference splitting method with Fromm scheme is used for the right-hand side residual evaluation, while generally first-order upwind differencing is used for the implicit operator on the left-hand side. Convergence studies are performed using an example of the flow past a NACA0012 airfoil at steady transonic flow condition, i. e. Mach number 0.8 at $1.25^{\circ}$ angle of attack. The results were compared with other computational results in order to validate the current numerical analysis. The results from the implicit AF algorithms were compared well in low surface with the other computational results; however, not well in upper surface. It might be due to lack of the grid around the shock position. Because the algorithm minimizes the errors of the approximate decomposition, the improved convergence rate with MAF were observed.

• 한국전산유체공학회지
• /
• 제3권2호
• /
• pp.17-26
• /
• 1998

The three-dimensional incompressible Navier-Stokes equations have been solved by a node-centered finite volume method with unstructured hybrid grids. The pressure-velocity coupling is handled by the artificial compressibility algorithm and convective fluxes are obtained by Roe's flux difference splitting scheme with linear reconstruction of the solutions. Euler implicit method with Jacobi matrix solver is used for the time-integration. The viscous terms are discretised in a manner to handle any kind of grids such as tetragedra, prisms, pyramids, hexahedra, or mixed-element grid. Inviscid bump flow is solved to check the accuracy of high order convective flux discretisation. And viscous flows around a circular cylinder and a sphere are studied to show the efficiency and accuracy of the solver.

• 한국음향학회:학술대회논문집
• /
• 한국음향학회 2001년도 추계학술발표대회 논문집 제20권 2호
• /
• pp.143-146
• /
• 2001

본 논문에서는 랜덤초기화 방법을 사용하여 초기 코드북을 생성하고, 이를 이용하여 초기 반복학습 시 수렴영역을 벗어난 2 이상의 가중치에 의한 K-means 알고리즘을 제안한다. 기존의 K-means 알고리즘이 국부적으로 최적화되고 초기 반복학습 시에 가중치의 영향이 크다는 점을 이용하여, 제안된 방법에서는 초기 반복학습 시의 가중치를 수렴영역에서 벗어난 큰 값으로 주고 이후 반복학습시의 가증치는 수렴영역 안에 있는 값으로 고정하여 코드북을 설계한다. 또한 초기 코드북을 얻기 위해 Splitting 방법과 같은 추가적인 과정 없이 랜덤한 방법에 의한 초기 코드북을 적용함으로써 제안된 알고리즘이 단순한 구조를 가지며, 구해진 코드북의 성능도 우수함을 확인할 수 있었다.

• 대한기계학회:학술대회논문집
• /
• 대한기계학회 2000년도 춘계학술대회논문집B
• /
• pp.51-56
• /
• 2000

The Finite Element Solutions Is reported on solid-liquid phase change in porous media with natural convection including freezing. The model is based on volume averaged transport equations, while phase change is assumed to occur over a small temperature range. The FEM (Finite Element Method) algorithm used in this study is 3-step time-splitting method which requires much less execution time and computer storage the velocity-pressure integrated method and the penalty method. And the explicit Lax-Wendroff scheme is applied to nonlinear convective term in the energy equation. For natural convection including melting and solidification the numerical results show reasonable agreement with FDM (Finite Difference Method) results.

• 한국정보과학회:학술대회논문집
• /
• 한국정보과학회 2000년도 봄 학술발표논문집 Vol.27 No.1 (B)
• /
• pp.307-309
• /
• 2000

진화 알고리즘은 현재까지 다양한 최적화 문제들에 사용되어 왔고, 또한 이러한 최적화 문제들은 효율적으로 해결하기 위하여 많은 진화 알고리즘이 개발되어 왔다. 그러나 이러한 진화 알고리즘들의 공통적인 문제점은 탐색공간의 확대에 대하여 전반적으로 탐색시간이 오래 걸린다는 것이다. 실제로 최적화 해야 할 변수의 증가에 따라 탐색 차원이 증가하므로 탐색 시간도 기하급수적으로 늘어난다. 따라서 최근의 진화 알고리즘에 대한 연구는 탐색공간의 축소나, 진화 속도의 향상에 초점이 맞추어져 있었고, 이러한 경향에 따라 많은 연구성과가 있었다. Potter와 Dejong의 협력 공진화와, Weicker의 적응적 공진화가 바로 그것이다. 그러나 이 방법들도 최적화 해야 할 변수들이 서로 강한 의존성을 가지고 있는 경우나, 대부분의 변수가 서로 의존성을 가지고 있는 경우에는 그다지 좋은 결과를 보이지 못하는 문제점을 가지고 있다. 본 논문에서는 이러한 연구들을 기반으로 하여 각 방법의 단점들을 보완함으로써 효율을 향상시킨 새로운 진화 알고리즘을 제안한다.

• 산업경영시스템학회지
• /
• 제37권4호
• /
• pp.187-192
• /
• 2014

Lot-streaming is the process of splitting a job (lot) into sublots to allow the overlapping of operations between successive machines in a multi-stage production system. A new genetic algorithm (NGA) is proposed for an n-job, m-machine, lot-streaming flow shop scheduling problem with equal-size sublots in which the objective is to minimize the total stretch. The stretch of a job is the ratio of the amount of time the job spent before its completion to its processing time. NGA replaces the selection and mating operators of genetic algorithms (GAs) by marriage and pregnancy operators and incorporates the idea of inter-chromosomal dominance and individuals' similarities. Extensive computational experiments for medium to large-scale lot-streaming flow-shop scheduling problems have been conducted to compare the performance of NGA with that of GA.

• 한국철도학회:학술대회논문집
• /
• 한국철도학회 2004년도 추계학술대회 논문집
• /
• pp.503-509
• /
• 2004

With modem computational fluid dynamics method (CFD), air-charging models of the air brake pipe system and auxiliary reservoir are built. Compared with one-dimension model, no empirical formula is introduced to solve branch pipe fields for two-dimension model. A modified operator-splitting method is presented to solve the coupled equations of pressure and velocity, and numerical simulation shows that it is very stable. Compare the numerical results with empirical data of heavy haul trains in home and abroad so as to prove the correctness of the theory and algorithm presented. This paper gives theoretic reference to the experiments of braking effects of heavy haul trains, and forms a basis for development of complete freight train air brake system simulation.

• 산업공학
• /
• 제9권3호
• /
• pp.64-71
• /
• 1996

In the survivability and simplicity aspect, SONET Self-healing Ring(SHR) is one of the most important schemes for the high-speed telecommunication networks. Since the ring capacity requirement is defined by the largest STS-1 cross-section in the ring, load balancing is the key issue in the design of SONET SHR. Recently, most of the research on load balancing problem have been concentrated on the SONET single-ring case. However, in certain applications, multiple-ring configuration is necessary because of the geographical limitations or the need for extra bandwidth. In this paper, the load balancing problem for SONET dual-ring is considered by assuming symmetric inter-ring demands. We present a linear programming based formulation of the problem. Initial solution and improvement procedures are presented, which solves the routing and interconnection between the two rings for each demand. Computational experiments are performed on various size of networks with randomly generated demand sets. Results show that the proposed algorithm is excellent in both the solution quality and the computational time requirement. The average error bound of the solutions obtained is 0.26% of the optimum.

• 한국전산유체공학회:학술대회논문집
• /
• 한국전산유체공학회 1999년도 춘계 학술대회논문집
• /
• pp.137-143
• /
• 1999

The unsteady transonic viscous flow has been analyzed over an oscillatory airfoil. The CSCM(Conservative Supra Characteristic Method) upwind flux difference splitting method and the iterative time marching scheme having first order accuracy in time and second to third order accuracy in space was applied on dynamic meshes. A steady flow field of Mach number 0.7 has been calculated for the verification of unsteady algorithm. The time-accurate unsteady calculations have been done for NACA 0012 airfoil oscillating around quarter chord about freestream Mach number 0.6 on dynamic meshes. The results have been compared with the AGARD Case 3 experimental data. The periodic characteristics have been compared with the experimental results.

• 대한기계학회논문집
• /
• 제19권11호
• /
• pp.3031-3040
• /
• 1995

Two-dimensional incompressible Navier-Stokes equations have been solved by the node-centered finite volume method with the unstructured triangular meshes. The pressure-velocity coupling is handled by the artificial compressibility algorithm due to its computational efficiency associated with the hyperbolic nature of the resulting equations. The convective fluxes are obtained by the Roe's flux difference splitting scheme using edge-based connectivities and higher-order differences are achieved by a reconstruction procedure. The time integration is based on an explicit four-stage Runge-Kutta scheme. Numerical procedures with local time stepping and implicit residual smoothing have been implemented to accelerate the convergence for the steady-state solutions. Comparisons with experimental data and other numerical results have proven accuracy and efficiency of the present unstructured approach.