• The Journal of the Acoustical Society of Korea
• /
• v.13 no.1E
• /
• pp.20-24
• /
• 1994

A new speech recognition model, DLSMM(Dynamically Localized Self-organizing Map Model) and its effective training algorithm are proposed in this paper. In DLSMM, temporal and spatial distortions of speech are efficiently normalized by dynamic programming technique and localized self-organizing maps, respectively. Experiments on Korean digits recognition have been carried out. DLSMM has smaller Experiments on Korean digits recognition have been carried out. DLSMM has smaller connections than predictive neural network models, but it has scored a little high recognition rate.

• Journal of Korean Society of societal Security
• /
• v.3 no.1
• /
• pp.33-42
• /
• 2010

This study predicts the stress history for truss bridge with local damages by using global-local model combination method. For this end, the global structure is modeled by 3D frame elements and the selected local details are modeled by shell elements. Then superposition principle enable the global-local model to be combined interactively. Because the frame model cannot consider the rigidity of gusset plate and the interation of structural members due to the complexity of stress distribution in truss connection. The section modification factors are proposed to calibrate the stiffness of global frame element. The global-local model combination is verified by comparing the numerical results with experimental data obtained from the proof loading test to the operating truss bridge. Furthermore, stress histrories of the truss bridge are generated in the consideration of the rigidity of truss connection with local damage by using the combination method.

• Journal of Korean Society of Steel Construction
• /
• v.17 no.2 s.75
• /
• pp.139-149
• /
• 2005

Post-local buckling behavior is a very important consideration in plastic and seismic design of steel structures. It describes the structural behavior up to the final collapse state. In order to assess the actual reliability of structures under severe repeated loading, such as strong earthquakes, it is necessary to evaluate the progressive cyclic deterioration of stiffness as well as the strength and energy dissipation capacity of the structures after local buckling happens. In this study, a simple analytical model developed for predicting post-local buckling behavior for cyclic and non-proportional loading histories, has been proposed. This analytical model uses the stress resultant model based on the two surface model. Analytical moment-curvature relationship using this model compare well with the experimental results in constant amplitude cycling, and linearized energy deterioration which is very important in seismic design can be predicted from the proposed model.

• Journal of the Korean Society for Railway
• /
• v.11 no.5
• /
• pp.465-470
• /
• 2008

Sound transmission characteristics are investigated on the honeycomb panels used for railway vehicles. Equivalent orthotropic plate model and equivalent mass law are applied to predict the sound transmission loss (STL) of the honeycomb panels. The predicted values of the STL are compared with the measured values. The reliability and the limitation of the prediction models are investigated. Coincidence effect and local resonance effect on STL are considered. The result of the study shows that the equivalent orthotropic plate model can be used as a good prediction model, if the local resonance frequency is properly applied. finally, ways to improve the severe STL drop by local resonance are proposed and the effect on the sound insulation performance is analysed.

• Journal of the Computational Structural Engineering Institute of Korea
• /
• v.27 no.2
• /
• pp.87-94
• /
• 2014

In solid mechanics, the peridynamics theory has provided a suitable framework for material failure and damage propagation simulation. Peridynamics is computationally expensive since it is required to solve enormous nonlocal interactions based upon integro-differential equations. Thus, multiscale coupling methods with other local models are of interest for efficient and accurate implementations of peridynamics. In this study, peridynamic models are restricted to regions where discontinuities or stress concentrations are present. In the domains characterized by smooth displacements, classical local models can be employed. We introduce a recently developed blending scheme to concurrently couple bond-based peridynamic models and the Navier equation of classical elasticity. We demonstrate numerically that the proposed blended model is suitable for point loads and static fracture, suggesting an alternative framework for cases where peridynamic models are too expensive, while classical local models are not accurate enough.

• Proceedings of the Computational Structural Engineering Institute Conference
• /
• 2011.04a
• /
• pp.583-586
• /
• 2011

가새골조의 정확한 내진성능평가를 위해서는 실험과 유사한 이력거동과 파단 시점을 잘 예측할 수 있는 해석적 모델이 필요하다. 가새의 이력거동을 모사히기 위해 본 연구에서는 이전 연구자들에 의해 제안된 물리적 이론 모델을 사용하였다. 또한, 가새부재의 국부좌굴에 의한 파단 예측을 위해 피로변수의 보정계수를 도입한 손상 지표를 개선하였다. 결과적으로 실험과 해석 결과를 비교하여 콘크리트충전 각형 강관가새부재의 국부좌굴 효과를 반영한 피로 보정계수를 판폭두께비에 따라 회귀분석을 통해 결정하고, 본 연구에서 제안한 해석 모델의 결과와 이전 연구의 결과를 비교하여 검증하였다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2010.11a
• /
• pp.471-473
• /
• 2010

본 논문은 선형회귀(LR: Linear Regression)와 국부적인 방사기저함수 네트워크(RBFN: Radial Basis Function Networks)를 결합한 점진적인 모델(incremental model)의 설계와 관련되어진다. 전형적인 RBFN에 의한 모델링과는 달리, 제안된 방법의 근본적인 원리는 두 단계에 의해 고려되어진다. 첫째, 전체 모델의 설계과정에서 전역적인 모델로써 선형회귀에 의해 데이터의 선형부분을 구축한다. 다음으로, 모델링 오차는 오차가 존재하는 국부적인 공간에서 RBFN에 의해 보상되어진다. 여기서, 오차의 분포로부터 RBFN을 설계하기 위해 컨텍스트 기반 퍼지 클러스터링(CFC: Context-based Fuzzy Clustering)를 통해 정보입자의 형태로 구축되어진다. 실험은 자동차 mpg 연료소비량 예측과 부동산 가격예측문제를 통해 제안된 방법의 우수성을 증명한다.

• Journal of Korean Society of Steel Construction
• /
• v.20 no.5
• /
• pp.617-624
• /
• 2008

Well-designed steel moment connections will undergo local buckling before they exhaust their available rotation capacity, and inelastic post-buckling deformation plays a major role in defining the connection rotation capacity. An approximate analytical method to model strength degradation and failure of beam plastic hinges due to local buckling and estimation of the seismic rotation capacity of fully restrained beam-column connections in special steel moment-resisting frames under both monotonic and cyclic loading conditions is proposed in this study. This method is based on the plastic mechanism and a yield line plastic hinge (YLPH) model whose geometry is determined using the shapes of the buckled plastic hinges observed in experiments. The proposed YLPH model was developed for the improved WUF-W and RBS connections and validated in comparison with experimental data. The effects of the beam section geometric parameters on the rotation capacity were discussed in the companion paper (parametric studies).

• Journal of the Korean Society of Propulsion Engineers
• /
• v.16 no.1
• /
• pp.9-15
• /
• 2012

An unsteady internal ballistic performance model was proposed to take account for the variance of local regression rate along the grain port of a hybrid rocket combustor. The characteristic parameters of hybrid rocket motor was investigated. The performance model of concern in the study was fairly comparable with the test result. The combustion coefficients and local burning characteristics of a hybrid rocket motor were evaluated. The local variation of the oxidizer mass flow rate results in the changes of local regression rate, pressure, temperature, and gas velocity to flow direction, which was analyzed quantitatively.

• Proceedings of the IEEK Conference
• /
• 2008.06a
• /
• pp.1001-1002
• /
• 2008

본 논문에서는 조명 또는 장면의 갑작스러운 변화에 효과적으로 배경모델링을 하기 위해 국부이진패턴을 이용한 다중 배경모델링 방법을 제안한다. 제안하는 방법은 각 장면에서 독립적인 배경모델을 이용하여 모델 업데이트를 실시한다. 이후 검출된 전경 영역의 비율이 일정 임계치를 넘게 되면 기존의 모델 중 적합한 모델을 찾거나 새로운 모델을 생성하여 현재 배경모델을 대체한다. 이는 배경모델의 성능을 유지하면서 효율적으로 장면의 변화에 바로 대응할 수 있는 장점이 있다. 실험결과에서는 실내조명이 갑작스럽게 변하는 영상과 Pan Tilt Zoom 카메라를 이용한 다중 영상에서 제안한 방법이 효과적으로 동작함을 확인할 수 있었다.