• Proceedings of the Korean Society For Composite Materials Conference
• /
• 2003.04a
• /
• pp.79-82
• /
• 2003

With the wide application of fiber-reinforced composite material in aero-structures and mechanical parts, the design of composite joint have become a very important research area because they are often the weakest areas in composite structures. In this paper, the failure area index method to predict the failure load of the mechanically fastened composite joint was used and the prediction accuracies of the linear finite element analysis were compared with those of nonlinear finite element analysis.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.33 no.8
• /
• pp.773-778
• /
• 2009

Chevron rubber springs are used in primary suspensions for rail vehicle. Chevron rubber spring have function which reduce vibration and noise, support load carried in operation of rail vehicle. Prediction and evaluation of characteristics are very important in design procedure to assure the safety and reliability of the rubber spring. The computer simulation using the nonlinear finite element analysis program executed to predict and evaluate the load capacity and stiffness for the chevron spring. The non-linear properties of rubber which are described as strain energy functions are important parameters. These are determined by material tests which are uniaxial tension, equi-biaxial tension and shear test. The appropriate shape and material properties are proposed to adjust the required characteristics of rubber springs in the three modes of flexibility.

• The Journal of the Acoustical Society of Korea
• /
• v.13 no.1E
• /
• pp.64-70
• /
• 1994

Predictive neural network models are powerful speech recognition models based on a nonlinear pattern prediction. But those models suffer from poor discrimination between acoustically similar words. In this paper we propose an discriminative training algorithm for predictive neural network models. This algorithm is derived from GPD (Generalized Probabilistic Descent) algorithm coupled with MCEF(Minimum Classification Error Formulation). It allows direct minimization of a recognition error rate. Evaluation of our training algoritym on ten Korean digits shows its effectiveness by 30% reduction of recognition error.

• Proceedings of the KIPE Conference
• /
• 2002.07a
• /
• pp.470-475
• /
• 2002

Instantaneous following PWM control technique is pulsed nonlinear dynamic control method. This new control technique using analog integrator is proposed to control the duty ratio D of do-dc converter. In this control method, the duty ratio of a switch is exactly equal In or proportional to the control reference in the steady state or in a transient. Proposed control method compensates power source perturbation in one switching cycle, and the average value of the dynamic reference in one switching cycle. There is no steady state error nor dynamic error between the control reference and the average value of the switched variable. Experiments with buck converter have demonstrated the robustness of the control method and verified theoretical prediction. The control method is very general and applicable to all type PWM

• The Transactions of the Korean Institute of Electrical Engineers D
• /
• v.49 no.8
• /
• pp.415-422
• /
• 2000

This paper presents a new algorithm for hierarchical optimal control of nonlinear systems. The proposed method is simple because the solutions are obtained by only exchanging informations of coefficient vector based on interaction prediction principle and FWT(fast Walsh transform) in upper and lower level. Since we solve two point boundary problem with Picard's iterative method and the backward integral operational matrix of Walsh function to obtain the optimal vector of each independent subsystem, the algorithm is simple and its operation is fast without inverse matrix and kronecker product operation. In simulation, the proposed algorithm's usefulness is proved by comparison with the global optimal control methods.

• Proceedings of the KIEE Conference
• /
• 2002.11c
• /
• pp.378-381
• /
• 2002

Instantaneous following PWM control technique is pulsed nonlinear dynamic control method. This new control technique using analog integrator is proposed to control the duty ratio D of DC-DC converter. In this control method, the duty ratio of a switch is exactly equal to or proportional to the control reference in the steady state or in a transient. Proposed control method compensates power source perturbation in one switching cycle, and the average value of the dynamic reference in one switching cycle. There is no steady state error nor dynamic error between the control reference and the average value of the switched variable. Experiments with buck converter have demonstrated the robustness of the control method and verified theoretical prediction. The control method is very general and applicable to all type PWM.

• Journal of Korean Society for Quality Management
• /
• v.27 no.2
• /
• pp.263-276
• /
• 1999

In this paper, we consider the change request data collected from the system test of a large-scale telecommunication software and analyze the types and causes of failures. And we develop statistical models that incorporate a functional relation between the faults and some software metrics. To this end, we consider three possible regression models including a stepwise regression model and two nonlinear models. Three developed models are evaluated with respect to the predictive quality. We also discuss the advantage of proposed models and the application of our model to a new project.

• Journal of the Korean Statistical Society
• /
• v.31 no.1
• /
• pp.63-75
• /
• 2002

Most attempts at Bayesian analysis of neural networks involve hierarchical modeling. We believe that similar results can be obtained with simpler models that require less computational effort, as long as appropriate restrictions are placed on parameters in order to ensure propriety of posterior distributions. In particular, we adopt a model first introduced by Lee (1999) that utilizes an improper prior for all parameters. Straightforward Gibbs sampling is possible, with the exception of the bias parameters, which are embedded in nonlinear sigmoidal functions. In addition to the problems posed by nonlinearity, direct sampling from the posterior distributions of the bias parameters is compounded due to the duplication of hidden nodes, which is a source of multimodality. In this regard, we focus on sampling from the marginal posterior distribution of the bias parameters with Markov chain Monte Carlo methods that combine traditional Metropolis sampling with a slice sampler described by Neal (1997, 2001). The methods are illustrated with data examples that are largely confined to the analysis of nonparametric regression models.

• Transactions of Materials Processing
• /
• v.18 no.6
• /
• pp.494-499
• /
• 2009

The time-dependent constitutive law was utilized based on viscoelastic-plasticity to predict the time-dependent spring-back behavior of aluminum alloy 6022-T4 sheets. Besides nonlinear viscoelasticity, non-quadratic anisotropic yield function, Yld2000-2d, was used to account for the anisotropic yield behavior, while the combined isotropic-kinematic hardening law was used to represent the Bauschinger effect and transient hardening. For verification purposes, finite element simulations were performed for the draw-bending and the results were compared with experimental results.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.35 no.2
• /
• pp.81-90
• /
• 2017

This study aimed to construct an integrated database using the same spatio-temporal unit by employing various public-data types with different real-time information provision cycles and spatial units. Towards this end, three temporal interpolation methods (piecewise constant interpolation, linear interpolation, nonlinear interpolation) and a spatial matching method by district boundaries was proposed. The case study revealed that the linear interpolation is an excellent method, and the spatial matching method also showed good results. It is hoped that various prediction models and data analysis methods will be developed in the future using different types of data in the analysis database.