• 한국멀티미디어학회논문지
• /
• 제13권9호
• /
• pp.1299-1313
• /
• 2010

Top-k 질의는 데이터베이스에서 사용자가 가장 원하는 k개의 객체를 구하는 질의이다. Top-k 질의를 효율적으로 처리하는 대표적인 연구로 Partitioned-Layer Index (간단히, PL-index) 방법이 있다. PL-index는 데이터베이스를 여러 개의 더 작은 데이터베이스로 분할하고 각 분할된 데이터베이스에 대해 sublayer들의 list (간단히, sublayer list)를 구성한다. 이때, 분할된 데이터베이스에 대해서 top-i 결과가 될 수 있는 객체들을 그 분할된 데이터베이스에 대한 i번째 sublayer로 구성한다. 그리고 주어진 질의에 맞춰 그 sublayer list들을 병합함으로써 질의 결과를 구한다. PL-index는 질의 처리 시 데이터베이스로부터 읽어 들이는 객체의 개수가 매우 작다는 장점을 가지지만, sublayer list들을 병합할 때에 임의 접근(random access)이 많이 발생하기 때문에 디스크 기반의 데이터베이스 환경에서 질의 처리 성능이 저하된다. 이에 본 논문에서는 임의 접근 횟수를 줄임으로써 디스크 기반의 데이터베이스 환경에서 PL-index의 질의 처리 성능을 크게 향상시키는 요약된(Abstracted) Partitioned-Layer Index (간단히, APL一index)를 제안한다. 먼저, PL-index의 각 sublayer를 가상의 (점) 객체로 요약함으로써 sublayer list들을 이러한 점 객체들의 list들(즉, APL-index)로 변형한다. 그리고 APL-index에 대해 질의 처리를 가상으로 수행하여 실제 질의 처리 시 접근할 sublayer를 예측한다, 그리고 예측된 sublayer들을 sublayer list별로 한꺼번에 읽어 들임으로 PL-index에서 발생하는 임의 접근 횟수를 줄인다. 합성 데이터와 실제 데이터에 대한 실험을 통하여 제안한 APL-index가 PL-index의 임의 접근 횟수를 크게 줄일 수 있음을 보인다.

• Journal of Mechanical Science and Technology
• /
• 제18권10호
• /
• pp.1782-1798
• /
• 2004

A numerical study of a natural convection in a rectangular cavity with the low-Reynolds-number differential stress and flux model is presented. The primary emphasis of the study is placed on the investigation of the accuracy and numerical stability of the low-Reynolds-number differential stress and flux model for a natural convection problem. The turbulence model considered in the study is that developed by Peeters and Henkes (1992) and further refined by Dol and Hanjalic (2001), and this model is applied to the prediction of a natural convection in a rectangular cavity together with the two-layer model, the shear stress transport model and the time-scale bound ν$^2$- f model, all with an algebraic heat flux model. The computed results are compared with the experimental data commonly used for the validation of the turbulence models. It is shown that the low-Reynolds-number differential stress and flux model predicts well the mean velocity and temperature, the vertical velocity fluctuation, the Reynolds shear stress, the horizontal turbulent heat flux, the local Nusselt number and the wall shear stress, but slightly under-predicts the vertical turbulent heat flux. The performance of the ν$^2$- f model is comparable to that of the low-Reynolds-number differential stress and flux model except for the over-prediction of the horizontal turbulent heat flux. The two-layer model predicts poorly the mean vertical velocity component and under-predicts the wall shear stress and the local Nusselt number. The shear stress transport model predicts well the mean velocity, but the general performance of the shear stress transport model is nearly the same as that of the two-layer model, under-predicting the local Nusselt number and the turbulent quantities.

• 한국콘크리트학회:학술대회논문집
• /
• 한국콘크리트학회 2005년도 추계 학술발표회 제17권2호
• /
• pp.177-180
• /
• 2005

The specified tensile strength provided by the manufacturer is determined on the basis of the reliable lower limit ($X-3{\sigma}$ : X=average tensile strength, $\sigma$=standard deviation) obtained from the material test results. Most of these data, however, are based on the test results of 1 layer of FRP sheet. Also, the partial strength reduction factor for strengthening RC members with FRP is based on the small-scale model tests. But, the failure mechanisms of small-scale model tests are reported to be different from those of the full-scale tests. This paper present the test results of full-scale RC beams strengthened with multi-layer GFRP sheets up to 3 layer as well as the material tests. From the material tests, it was observed that the average tensile strengths of GFRP sheets are decreased as the number of layer are increased. Also the premature debonding failure of RC beams strengthened with multi-layer GFRP sheets are observed in inverse proportion to the number of GFRP sheets

• 천문학논총
• /
• 제9권1호
• /
• pp.21-38
• /
• 1994

We have developed a cylindrical mixing layer model of a stellar jet including cooling effect in order to understand an optical emission mechanism along collimated high velocity stellar jets associated with young stellar objects. The cylindrical results have been calculated to be the same as the 2D ones presented by Canto & Raga(1991) because the entrainment efficiency in our cylindrical model has been obtained to be the same value as the 2D model has given. We have discussed the morphological and physical characteristics of the mixing layers by the cooling effect. As the jet Mach number increases, the initial temperature of the mixing layer goes high because the kinetic energy of the jet partly converts to the thermal energy of the mixing layer. The initial cooling of the mixing layer is very severe, changing its outer boundary radius. A subsequent change becomes adiabatic. The number of the Mach disks in the stellar jet and the total radiative luminosity of the mixing layer, based on our cylindrical calculation, have quite agreed with the observations.

• 한국전산구조공학회:학술대회논문집
• /
• 한국전산구조공학회 1998년도 가을 학술발표회 논문집
• /
• pp.283-289
• /
• 1998

In the latticed domes which is a set of space frame, buckling is derived if the external force reaches a limitation by the lightness of the material and the minimization of the member section area. these are concerned with a geometric shape, network pattern, the number of layer, and so on. Most of all, the number of layer of the lattice dome is a important factor from the viewpoint of initial and structure design. Therefore this study compared buckling characteristics of single-layer with double-layer latticed domes and investigated the relativity of buckling-stress-ratio and member-density-ratio according to rise ratio to improve that designers could extend the range of .design selection

• 한국정보처리학회논문지
• /
• 제6권11호
• /
• pp.3086-3096
• /
• 1999

To declare conformance of multi-protocol Implementation Under Test(IUT), every layer of the multi-protocol IUT should be tested. According to ISO9646, single-layer test method is applied to testing the highest layer of multi-protocol IUT and single-layer embedded test method is used for the layers by layer all the protocols in a multi-protocol IUT. This paper proposes a new method for testing a multi-protocol IUT. The proposed test method assumes that a multi-protocol IUT is under test and that the interfaces between the layers cannot be controlled or observed by the tester. We apply the proposed test method to TCP/IP and compare the application results with those of the existing test turns out that the proposed test method significantly reduces the number of test cases as well as the number of test events while providing the same test coverage. In addition, the proposed test method shows the capability to locate the layer that is source of failure in testing multi-protocol IUTs.

• 한국가시화정보학회지
• /
• 제4권2호
• /
• pp.51-58
• /
• 2006

A boundary layer visualization was carried out in order to investigate the influence of Reynolds number on an oscillating airfoil. An NACA 0012 airfoil is sinusoidally pitched at the quarter chord point with oscillation amplitude of ${\pm}6^{\circ}$. A smoke-wire technique was employed to visualize the boundary layer and the near-wake. The freestream velocities are 1.98, 2.83 and 4.03m/s and corresponding chord Reynolds numbers are $2.3{\times}10^4,\;3.3{\times}10^4$, and $4.8{\times}10^4$, respectively. As the reduced frequency of K=0.1 is fixed, the corresponding frequency of an airfoil was adjusted in each case. The results reveal that the point at which the shear stress in an unsteady boundary layer separation disappears does not correspond with the position of the breakdown of the boundary layer, and that the breakdown of the boundary layer occurs further downstream.

• 한국환경과학회지
• /
• 제11권3호
• /
• pp.201-207
• /
• 2002

This study was carried out to evaluate the artificial neural network algorithm for water quality forecasting in Chungju lake, north Chungcheong province. Multi-layer perceptron(MLP) was used to train artificial neural networks. MLP was composed of one input layer, two hidden layers and one output layer. Transfer functions of the hidden layer were sigmoid and linear function. The number of node in the hidden layer was decided by trial and error method. It showed that appropriate node number in the hidden layer is 10 for pH training, 15 for DO and BOD, respectively. Reliability index was used to verify for the forecasting power. Considering some outlying data, artificial neural network fitted well between actual water quality data and computed data by artificial neural networks.

• 전기의세계
• /
• 제28권4호
• /
• pp.47-52
• /
• 1979

Total number of resonant modes in a microwave oven cavity may be maximized for a given frequency bandwidth to obtain more uniform power distribution by choosing proper size of the cavity. The total number of modes is calculated for a dielectrically loaded rectangular cavity and its size is suggested here for which the change in the number of modes is less sensitive to the change of dielectric layer thickness and its total number of modes is maximized in a given range of cavity sizes. A prove coupled rectangular cavity is constructed and the total existing modes are measured to see the change of modes depending on the dielectric layer thickness and the cavity size. Surface wave mode existing in the dielectric layer is confirmed by measuring Q and the input impedance of the cavity for this mode, which closely compares with the calculation.

• 한국정보시스템학회지:정보시스템연구
• /
• 제9권1호
• /
• pp.27-44
• /
• 2000

The structure of neural networks is represented by a weighted directed graph with nodes representing units and links representing connections. Each link is assigned a numerical value representing the weight of the connection. In learning process, the values of weights are adjusted by errors. Following experiment results, the interval of adjusting weights, that is, epoch size influenced neural networks＇ performance. As epoch size is larger than a certain size, neural networks＇performance decreased drastically. And the number of hidden layer＇s node also influenced neural networks＇performance. The networks＇performance decreased as hidden layers have more nodes and then increased at some number of hidden layer＇s node. So, in implementing of neural networks the epoch size and the number of hidden layer＇s node should be decided by systematic methods, not empirical or heuristic methods.