• Journal of the Korean Institute of Intelligent Systems
• /
• v.21 no.5
• /
• pp.624-629
• /
• 2011

In this paper, a new kernelized approach for use in a recommender system (RS) is proposed. Using a machine learning technique, the proposed method predicts the user's preferences for unknown items and recommends items which are likely to be preferred by the user. Since the ratings of the users are generally inconsistent and noisy, a robust binary classifier called a dual margin Lagrangian support vector machine (DMLSVM) is employed to suppress the noise. The proposed method is applied to MovieLens databases, and its effectiveness is demonstrated via simulations.

• Proceedings of the KIEE Conference
• /
• 1996.11a
• /
• pp.133-136
• /
• 1996

In the paper, an artificial neural network (ANN) approach based on Lagrange multiplier method (Lagrangian ANN) is used to solve an economic load dispatch (ELD) problem. Traditionally ELD problem has one convex cost function as its objective function and nonlinear constraints such as power balance and maximum-minimum limits of real power. In this study, modification is given to the Lagrangian ANN proposed by Gong et all[5] to guarantee the convergence to the optimal solution. Simulation results demonstrate the effectiveness of the proposed method applied to the ELD problem.

• Journal of the Korean Geotechnical Society
• /
• v.17 no.6
• /
• pp.85-98
• /
• 2001

Cone penetration was analyzed by arbitrary Lagangian-Eulerian(ALE) method. In order to simulate full penetration, steady state analyses were performed using ABAQUS/Explicit, which models upward flow of soil layers. In the analysis of homogeneous layer it was found that the paths and the strain of soil particles were consistent with the result of the strain path method and that the ultimate resistance were reasonably evaluated. The cone penetration through different soil layers was also analyzed and that showed the transfer of cone resistance. The steady state ALE analysis could perform full penetration through the layered soils.

• 한국전산유체공학회:학술대회논문집
• /
• 2002.10a
• /
• pp.41-46
• /
• 2002

A numerical simulation was made to determine the motion of particles in the fluid. The simulation is based on the Eulerian-Lagrangian method. The fluid motion was solved using a PISO-based finite-element method and a $\kappa-\epsilon$ model of turbulence. In the Lagrangian method for the solid phase, the trajectories of particles are calculated by integrating the equations of motion of a single Particle, and the collision between particles are taken into account. The influence of particles on the fluid phase is taken into account by introducing source terms in the Eulerian equations govering the fluid flow. It is known as the particle-source-in-cell (PSIC) method. Also, the turbulent effect in the particles and fluid notion is considered. The numerical results were compared with the experiment for a two-phase flow in a vertical pipe.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.10 no.3
• /
• pp.29-38
• /
• 1990

A totally, new approach of Lagrangian formulation named 'Pseudo Lagrangian Formulation(PLF)' for large deformation analysis of continue and structures by the finite of element method has been presented, and the efficiency and accuracy of nonlinear analysis beam element formulated by PLF has been discussed by solving several numerical examples. In PLF, the deformation of a body is maeasured by assigning a nonphysical 'Pseudo' configuration as reference. The Lagrangian deformation and the finite element mapping of the traditonal Lagrangian approaches are then carried out directly at the same time, The result of numerical tests shows superior performance of PLF to the traditional Lagrangian methods, Applications of PLF to small and finite deformation problems indicate that PLF not only serves as an alternative but has certain implementational advantages over total or updated Lagrangian formulations.

• Applied Science and Convergence Technology
• /
• v.27 no.1
• /
• pp.14-18
• /
• 2018

In order to understand the mechanism of the synthesis of particles using a plasma torch, it is necessary to understand the reaction mechanisms using a computer simulation. In this study, we have developed a simulation method to combine the Lagrangian scheme to follow microparticles and a nodal method to treat nanoparticles categorized with different particle sizes. The Lagrangian scheme includes the Coulomb force which affects the dynamics of larger particles. In contrast, the nodal method is adequate for the nanoparticles because the charge effect is negligible for nanoparticles but the number of nanoparticles is much larger than that of microparticles. This method is helpful to understand the dynamics and growth mechanism of micro- and nano-powder mixture observed in the experiment.

• Proceedings of the Computational Structural Engineering Institute Conference
• /
• 2003.10a
• /
• pp.375-382
• /
• 2003

Porous media consist of physically and chemically different materials and have an extremely complicated behavior due to the different material properties of each of its constituents. In addition, the internal structure of porous media has generally a complex geometry that makes the description of its mechanical behavior quite complex. Thus, in order to describe and clarify the deformation behavior of porous media, constitutive models for deformation of porous media coupling several effects such as flow of fluids of thermodynamical change need to be developed in frame of Arbitrary Lagrangian Eulerian (ALE) description. The aim of ALE formulations is to maximize the advantages of Lagrangian and Eulerian methods, and to minimize the disadvantages. Therefore, this method is appropriate for the analysis of porous media that are considered for the behavior of solids and fluids. First of all, governing equations for saturated porous media based on ALE description are derived. Then, weak forms of these equations are obtained in order to implement numerical method using finite element method. Finally, Petrov-Galerkin method Is applied to develop finite element formulation.

• Proceedings of the Korean Society for Technology of Plasticity Conference
• /
• 2006.05a
• /
• pp.99-104
• /
• 2006

A semi-Lagrangian finite element scheme with objective time stepping algorithm for solving viscoelastic flow problem is presented. The convection terms in the momentum and constitutive equations are treated using a quasi-monotone semi-Lagrangian scheme, in which characteristic feet on a regular grid are traced backwards over a single time-step. Concerned with the generalized midpoint rule type of algorithms formulated to exactly preserve objectivity, we use the geometric transformation such as pull-back, push-forward operation. The method is applied to the 4:1 planar contraction problem for an Oldroyd B fluid for both creeping and inertial flow conditions.

• Proceedings of the Computational Structural Engineering Institute Conference
• /
• 2003.04a
• /
• pp.235-242
• /
• 2003

The solids and the fluids in porous media have a relative velocity to each other. Due to physically and chemically different material properties and their relative velocity, the behavior of saturated porous media is extremely complicated. Thus, in order to describe and clarify the deformation behavior of saturated porous media, constitutive models for deformation of porous media coupling several effects such as flow of the fluids or thermodynanical change need to be developed in frame of Arbitrary Lagrangian Eulerian (ALE) description. The aim of ALE formulations is to maximize the advantages of Lagrangian and Eulerian elements, and to minimize the disadvantages. Therefore, this method is appropriate for the analysis of porous media that are considered for the behavior of the solids and the fluids. In this work, governing equations of porous media based on ALE description are obtained from governing equations in frame of updated Lagrangian description. Then, weak forms of these equations are derived using arbitrary weighting functions.

• Magazine of the Korean Society of Agricultural Engineers
• /
• v.36 no.1
• /
• pp.128-140
• /
• 1994

Coastal areas, especially embayments are apt to be polluted easily and many embayments in Korea are already suffering from pollution problems. To manage such pollution, it is strongly needed to develop technique to trace movements of pollution. Such technique cove- ring the embayment affected by the tidal influence, should take account both of the convection and the diffusion motions which cause lots of problems in numerical calculation. In this study, a Eulerian-Lagrangian Analysis(ELA) model was applied to Young Il bay and tested for its applicablity, which was developed by using the Eulerian-Lagrangian Method that reduce the numerical disperison and oscillation by way of solving convection and diffusion terrns separately. Concentration of Chemical Oxygen Demand(COD) and Suspend Solid(SS) of the embay- ment area were estimated by the model and compared with the observed values and the sound results were obtained. At the same time the diffsion coefficient and decay coefficient for Chemical Oxygen Demand in the Young II Bay were found as Dx = Dy = 20m$^2$/sec, kd=2.5 ${\times}$ 10-5/sec respectively, and for Suspend Solid, Dx =Dy = 30m$^2$/sec, kd=5.0${\times}$ 10-5/sec