• 한국게임학회 논문지
• /
• 제15권5호
• /
• pp.109-118
• /
• 2015

고대 중국에서 기원된 바둑은 인공지능 분야에서 가장 어려운 도전 중의 하나로 간주된다. 지난 수년에 걸쳐 MCTS를 기반으로 하는 정상급 컴퓨터바둑 프로그램이 놀랍게도 접바둑에서 프로기사를 물리쳤다. MCTS는 게임이 끝날 때까지 일련의 무작위 유효착수를 시뮬레이션 하는 접근법이며, 기존의 지식기반 접근법을 대체했다. 저자는 MCTS의 변형인 UCT 알고리즘을 삼목 게임에 적용하여 최선의 첫 수를 찾고자 했으며, 순수 MCTS의 결과와 비교를 했다. 아울러 UCB 이해를 위한 다중슬롯머신 문제를 풀기 위해 엡실론-탐욕 알고리즘과 UCB 알고리즘을 소개 및 성능을 비교하였다.

• 한국산학기술학회논문지
• /
• 제14권12호
• /
• pp.6447-6454
• /
• 2013

기존 해수취수시스템의 경우 해수오염, 부유물질, 취수의 불안정 및 유지관리의 어려움 등으로 인해 해수취수의 어려움이 발생하곤 하였다. 이와 같은 단점을 극복하고자 매설식 해수취수시스템을 개발하여 격포항에 설치하였다. 본 연구에서는 새로운 시스템의 거동을 검토하고자, 수질관측 및 3차원 수치모의 실험을 수행하였다. 이 취수시스템에 대하여 총 5회에 걸쳐 COD, 총질소, 총인, PH 그리고 부유물질에 대하여 수질분석을 수행한 결과, COD, 총질소, 총인, PH의 경우 취수 전 후 저감 효과는 미미하였다. 그러나 부유물질의 경우 수산용수 1급 5 mg/L 이하로 정화되는 효과가 나타났다. 수치모형은 유한체적법 기반의 CFX 모형과 RNG $k-{\epsilon}$ 알고리즘을 선정하였으며, 여과사, 외부관 지름 및 두께를 재현하기 위해서 다공성 알고리즘을 적용하였다. 수치모의 실험을 수행한 결과, 2중관 구조가 내외부관 사이의 공간에 의해 압력의 분포를 균등하게 하여, 흐름 상태나 안정적인 취수측면에서 유리한 것으로 분석되었으며, 취수관을 다열 배치하였을 때 중앙 취수관과 비교하여 양쪽 취수관에서 유량이 0.98배로 감소하였지만 유동의 간섭현상은 발생하지 않았다.

• 한국항해항만학회지
• /
• 제47권4호
• /
• pp.231-238
• /
• 2023

등부표는 조류, 바람 등 외력에 영향을 받아 위치가 항상 유동적이고 위치는 항로표지용 AIS 또는 RTU를 통해 확인할 수 있다. 위치 확인이 가능한 등부표의 최근 5년간(2017~2021년) 위치 데이터 분석 결과 위치 오류 데이터는 평균 15.4%로 나타났으며 항해 안전사고예방 및 관리를 위해서는 위치 오류 데이터를 검출하고 정제된 위치 데이터 획득이 필요하다. 본 연구에서는 항로표지용 AIS 또는 RTU를 통해 획득한 위치 데이터를 DBSCAN Clustering하여 위치 오류 데이터를 검출하고 정제된 위치 데이터를 획득하고자 한다. 이를 위하여 위치 오류가 가장 많은 서해 해역 중 RTU가 설치된 군산항 1호 등부표의 21년도 위치 데이터를 Python library를 사용하여 DBSCAN Clustering 하였다. DBSCAN Clustering에 필요한 minPts는 2차원 데이터에 일반적으로 사용하는 값을 적용하였고 epsilon은 k-NN(최근접 이웃)알고리즘을 사용하여 값을 산출 및 적용하였다. DBSCAN Clustering 결과 minPts와 epsilon을 만족하지 못하는 위치 오류 데이터를 검출하였고 정제된 위치 데이터를 획득할 수 있었다. 본 연구는 항로표지용 AIS 또는 RTU가 설치된 등부표의 신뢰성 있는 위치 데이터를 획득할 수 있는 기초 자료로 활용할 수 있으며 항해 안전사고 예방에도 큰 도움이 될 것으로 판단된다.

• Journal of Information Processing Systems
• /
• 제16권6호
• /
• pp.1324-1342
• /
• 2020

Various modified algorithms of rapidly-exploring random tree (RRT) have been previously proposed. However, compared to the RRT algorithm for collision avoidance with global and static obstacles, it is not easy to find a collision avoidance and local path re-planning algorithm for dynamic obstacles based on the RRT algorithm. In this study, we propose boundary-RRT*, a novel-algorithm that can be applied to aerial vehicles for collision avoidance and path re-planning in a three-dimensional environment. The algorithm not only bounds the configuration space, but it also includes an implicit bias for the bounded configuration space. Therefore, it can create a path with a natural curvature without defining a bias function. Furthermore, the exploring space is reduced to a half-torus by combining it with simple right-of-way rules. When defining the distance as a cost, the proposed algorithm through numerical analysis shows that the standard deviation (σ) approaches 0 as the number of samples per unit time increases and the length of epsilon ε (maximum length of an edge in the tree) decreases. This means that a stable waypoint list can be generated using the proposed algorithm. Therefore, by increasing real-time performance through simple calculation and the boundary of the configuration space, the algorithm proved to be suitable for collision avoidance of aerial vehicles and replanning of local paths.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1996년도 한국자동제어학술회의논문집(국내학술편); 포항공과대학교, 포항; 24-26 Oct. 1996
• /
• pp.245-248
• /
• 1996

This paper suggests a new concept for missile guidance control, called linearized common curvature guidance law that enhances the probability to kill a target. The proposed guidance system is composed of two switching modes; one for the midcourse guidance and the other for the terminal guidance, which is switched by a specified critical value (.epsilon.). And the system and the commands are formulated and its simulations are provided in comparison with the conventional commanded line of sight guidance algorithm. Miss distance and angle of attack are denoted as performance of parameters. This new concept, common curvature guidance algorithm, revises the navigation guidance and accompanies, various considerations.

• 한국전산유체공학회:학술대회논문집
• /
• 한국전산유체공학회 2000년도 추계 학술대회논문집
• /
• pp.135-140
• /
• 2000

Numerical calculations are carried out in order to evaluate the performance of low-Re Reynolds stress model based on SSG model for a swirling turbulent flow in a pipe. The results are compared with those of $\kappa-\epsilon$ model and GL model, and the experimental data. The finite volume method is used for the discretization, and the power-law scheme is employed as a numerical scheme. The SIMPLE algorithm is used for velocity-Pressure correction in the governing equations.

• 대한기계학회논문집B
• /
• 제21권7호
• /
• pp.911-918
• /
• 1997

The turbulent viscous wake flows behind a single airfoil, two-dimensional stationary blade row and three-dimensional rotating blade row were calculated, and the numerical results were compared with experimental ones. The numerical technique was based on the SIMPLE algorithm using three turbulent closure models, standard k-.epsilon. model(WFM), low Reynolds number k-.epsilon. model(LRN) and Reynolds stress model (RSM). In the case of a single airfoil, WFM, LRN and RSM presented fairly good velocity distributions in the wake compared with experimental data. In the case of the stationary blade row, LRN and RSM presented better results than WFM for wake velocity distribution, and especially LRN showed best results among these three turbulent models. In the case of the rotating blade row, WFM and LRN showed fairly good agreement with experimental data of the three-dimensional velocity component distributions in the range from hub to mid span region. LRN was also superior to WFM in accuracy of prediction for the wake velocity distribution as same with the cases of a airfoil and the stationary blade row.

• 한국수자원학회논문집
• /
• 제44권6호
• /
• pp.429-438
• /
• 2011

In order to simulate a free surface flow in a trench channel, a three-dimensional incompressible unsteady Reynolds-averaged Navier-Stokes (RANS) equations are closed with the ${\kappa}-{\epsilon}$ model. The artificial compressibility (AC) method is used. Because the pressure fields can be coupled directly with the velocity fields, the incompressible Navier-Stokes (INS) equations can be solved for the unknown variables such as velocity components and pressure. The governing equations are discretized in a conservation form using a second order accurate finite volume method on non-staggered grids. In order to prevent the oscillatory behavior of computed solutions known as odd-even decoupling, an artificial dissipation using the flux-difference splitting upwind scheme is applied. To enhance the efficiency and robustness of the numerical algorithm, the implicit method of the Beam and Warming method is employed. The treatment of the free surface, so-called interface-tracking method, is proposed using the free surface evolution equation and the kinematic free surface boundary conditions at the free surface instead of the dynamic free surface boundary condition. AC method in this paper can be applied only to the hydrodynamic pressure using the decomposition into hydrostatic pressure and hydrodynamic pressure components. In this study, the boundary-fitted grids are used and advanced each time the free surface moved. The accuracy of our RANS solver is compared with the laboratory experimental and numerical data for a fully turbulent shallow-water trench flow. The algorithm yields practically identical velocity profiles that are in good overall agreement with the laboratory experimental measurement for the turbulent flow.

• Journal of applied mathematics & informatics
• /
• 제26권1_2호
• /
• pp.213-222
• /
• 2008

Iterative methods are often suitable for solving least squares problems min$||Ax-b||_2$, where A $\epsilon\;\mathbb{R}^{m{\times}n}$ is large and sparse. The well known LSQR algorithm is among the iterative methods for solving these problems. A good preconditioner is often needed to speedup the LSQR convergence. In this paper we present the numerical experiments of applying a well known preconditioner for the LSQR algorithm. The preconditioner is based on the $A^T$ A-orthogonalization process which furnishes an incomplete upper-lower factorization of the inverse of the normal matrix $A^T$ A. The main advantage of this preconditioner is that we apply only one of the factors as a right preconditioner for the LSQR algorithm applied to the least squares problem min$||Ax-b||_2$. The preconditioner needs only the sparse matrix-vector product operations and significantly reduces the solution time compared to the unpreconditioned iteration. Finally, some numerical experiments on test matrices from Harwell-Boeing collection are presented to show the robustness and efficiency of this preconditioner.

• 대한기계학회:학술대회논문집
• /
• 대한기계학회 2001년도 춘계학술대회논문집E
• /
• pp.499-505
• /
• 2001

The objective of the paper was to measure the pressure drop and to investigate the recirculation region of the conical orifices used in Kwang-yang Iron & Steel Company. The flow field with water used as a working fluid was the turbulent flow for Reynolds number of $2{\times}10^4$. The effective parameters for the pressure drop and the recirculation region were the conical orifice's inclined angle (${\theta}$) against the wall, the interval(L) between orifices, the relative angle of rotation(${\alpha}$) of the orifices, the shape of the orifice's hole(circle, rectangle, triangle) having the same area. It was found that the shape of the orifice's hole affected the pressure drop and the flow field a lot, But the other parameters did not make much differences to the pressure drop. The PISO algorithm with FLUENT code was employed.