• 제어로봇시스템학회:학술대회논문집
• /
• 2001.10a
• /
• pp.169.4-169
• /
• 2001

Conventional methods to solve the nonlinear programming problem range from augmented Lagrangian methods to sequential quadratic programming (SQP) methods. NPSOL, which is a SQP code, has been widely used to solve various optimization problems but is still subject to many numerical problems such as convergence to local optima, difficulties in initialization and in handling non-smooth cost functions. Recently, many evolutionary methods have been developed for constrained optimization. Among them, CEALM (Co-Evolutionary Augmented Lagrangian Method) shows excellent performance in the following aspects: global optimization capability, low sensitivity to the initial parameter guessing, and excellent constraint handling capability due to the benefit of the augmented Lagrangian function. This algorithm is ...

• Journal of applied mathematics & informatics
• /
• v.33 no.1_2
• /
• pp.193-201
• /
• 2015

In this paper, we present preconditioned generalized accelerated overrelaxation (GAOR) methods for solving weighted linear least square problems. We compare the spectral radii of the iteration matrices of the preconditioned and the original methods. The comparison results show that the preconditioned GAOR methods converge faster than the GAOR method whenever the GAOR method is convergent. Finally, we give a numerical example to confirm our theoretical results.

• Proceedings of the KSME Conference
• /
• 2002.08a
• /
• pp.35-38
• /
• 2002

The convergence acceleration methods for the compressible Wavier-Stokes equations are studied ,which are multigrid method and implicit preconditioned multistage time stepping method. In this paper, the performance of implicit preconditioning methods are studied for the full-coarsening multigrid methods on the high Reynolds number compressible flow computations. The effect of numerical flux on the convergence are investigated for the inviscid and viscous calculations.

• Journal of applied mathematics & informatics
• /
• v.28 no.3_4
• /
• pp.663-677
• /
• 2010

In this paper, the generalized SOR-like methods are presented for solving the saddle point problems. Based on the SOR-like methods, we introduce the uncertain parameters and the preconditioned matrixes in the splitting form of the coefficient matrix. The necessary and sufficient conditions for guaranteeing its convergence are derived by giving the restrictions imposed on the parameters. Finally, numerical experiments show that this methods are more effective by choosing the proper values of parameters.

• Journal of applied mathematics & informatics
• /
• v.28 no.3_4
• /
• pp.773-781
• /
• 2010

Recently, Yun [8] proposed a new three-step iterative method with the fourth-order convergence for solving nonlinear equations. By using his ideas, we develop a general form of multi-step iterative methods with higher order convergence for solving nonlinear equations, and then we study convergence analysis of the multi-step iterative methods. Lastly, some numerical experiments are given to illustrate the performance of the multi-step iterative methods.

• Journal of the Chungcheong Mathematical Society
• /
• v.24 no.3
• /
• pp.437-447
• /
• 2011

In this paper we present a new variant of the Euler-Chebyshev method for solving nonlinear equations. Analysis of convergence is given to show that the presented methods are at least fifth-order convergent. Several numerical examples are given to illustrate that newly presented methods can be competitive to other known fifth-order methods and the Newton method in the efficiency and performance.

• Communications for Statistical Applications and Methods
• /
• v.28 no.2
• /
• pp.135-150
• /
• 2021

We propose principal differential analysis based classification methods. Computations of squared multiple correlation function (RSQ) and principal differential analysis (PDA) scores are reviewed; in addition, we combine principal differential analysis results with the logistic regression for binary classification. In the numerical study, we compare the principal differential analysis based classification methods with functional principal component analysis based classification. Various scenarios are considered in a simulation study, and principal differential analysis based classification methods classify the functional data well. Gene expression data is considered for real data analysis. We observe that the PDA score based method also performs well.

• Journal of Mechanical Science and Technology
• /
• v.17 no.7
• /
• pp.1073-1082
• /
• 2003

Many researchers have investigated the blood flow characteristics through bileaflet mechanical heart valves using computational fluid dynamics (CFD) models. Their numerical approach methods can be classified into three types; steady flow analysis, pulsatile flow analysis with fixed leaflets, and pulsatile flow analysis with moving leaflets. The first and second methods have been generally employed for two-dimensional and three-dimensional calculations. The pulsatile flow analysis interacted with moving leaflets has been recently introduced and tried only in two-dimensional analysis because this approach method has difficulty in considering simultaneously two physics of blood flow and leaflet behavior interacted with blood flow. In this publication, numerical calculation for pulsatile flow with moving leaflets using a fluid-structure interaction method has been performed in a three-dimensional geometry. Also, pulsatile flow with fixed leaflets has been analyzed for comparison with the case with moving leaflets. The calculated results using the fluid-structure interaction model have shown good agreements with results visualized by previous experiments. In peak systole. calculations with the two approach methods have predicted similar flow fields. However, the model with fixed leaflets has not been able to predict the flow fields during opening and closing phases. Therefore, the model with moving leaflets is rigorously required for advanced analysis of flow fields.

• Journal of the Korean Society of Safety
• /
• v.33 no.4
• /
• pp.84-89
• /
• 2018

The number of landslides has increased since the 2000s due to the increased frequency of heavy rainfall caused by abnormal weather. A variety of debris flow prevention facilities have been installed as a countermeasure against this problem. However, it is not easy to evaluate the efficiency of debris flow prevention structures except for the structures with constant volume such as the erosion dam, because the other structures are limited to be reproduced in simulation program for debris flow. Therefore, the methods by which the debris flow prevention structures were modeled were proposed, and the efficiency of four prevention structures installed in Baekyang Mt. in Busan was evaluated with UDS, which accuracy had been verified, using these methods. The initial amount of debris flow was determined based on landslide which occurred in 2014, and specifications of the complex retaining walls around the settlements were measured and applied modeling for terrain. The numerical results showed that the efficiency of debris flow prevention structures could be quantitatively presented. Among the debris flow prevention structures installed in Baekyang Mt., prevention structure of barrier type for debris flow was the most efficiency and debris flow prevention device was the lowest efficiency when the only depth of debris was evaluated. It seems that this study is meaningful to propose the methods which were used to model the debris flow prevention structures that could not be reproduced in most 2D debris flow numerical analysis programs. If precise verification of the presented methods is carried out, it will be possible to provide clear criteria for the efficiency evaluation method of disaster prevention structures.

• International Journal of Highway Engineering
• /
• v.9 no.4
• /
• pp.215-226
• /
• 2007

To verify the performance of roadside barriers, occupant risk indices are calculated from acceleration and angular velocity data of vehicle crash tests. The occupant risk indices to be computed include THIV(Theoretical Head Impact Velocity), PHD(Post-impact Head Deceleration), ASI(Acceleration Severity Index), OIV(Occupant Impact Velocity) and ORA(Occupant Ridedown Acceleration). There is a confusion due to different values of occupant risk indices produced for the same test data because various computational procedures and data processing methods can be applied to compute them. To slove this problem the effects of various numerical procedures and data processing methods on occupant risk indices were investigated. If the sampling rate specified in the guidelines is used for full-scale vehicle crash tests, an interpolation of impact time and numerical integration methods do not result in an appreciable change of THIV and OIV. The way to determine 10msec moving average for PHD and zero offset of data processing should be specified in the guidelines because 10msec moving average and zero offset methods have a significant influence on occupant risk indices.