• 한국전산유체공학회지
• /
• 제15권4호
• /
• pp.25-31
• /
• 2010

Interface tracking of two phase is significant to analyze multi-phase phenomena. The VOF(Volume of Fluid) and level set are well known interface tracking method. However, they have limitations to solve compressible flow and incompressible flow at the same time. CIP(Cubic Interpolate Propagation) method is appropriate for considering compressible and incompressible flow at once by solving the governing equation which is divided up into advection and non-advection term. In this article, we analyze the droplet impingement according to various We number using improved CIP method which treats nonlinear term once more comparison with original CIP method. Furthermore, we compare spread radius after droplet impingement on the wall with the experimental data and original CIP method. The result using improved CIP method shows the better result of the experiments, comparison with result of original CIP method, and it reduces the mass conservation error which is generated in the numerical analysis comparison with original CIP method.

• 한국공작기계학회논문집
• /
• 제18권2호
• /
• pp.170-177
• /
• 2009

The objective of this study is to evaluate an acoustic emission (AE) source characterization and fracture behavior of the SM45C steel by using back-propagation neural network (BPN). In previous research Ref. [8] about k-nearest neighbor classifier (k-NNC) continuity, we used K-means clustering method as an unsupervised learning method for obtaining multi-variate AE main data sets, such as AE counts, energy, amplitude, risetime, duration and counts to peak. Similarly, we applied k-NNC and BPN as a supervised learning method for obtaining multi-variate AE working data sets. According to the error of convergence for determinant criterion Wilk's ${\lambda}$, heuristic criteria D&B(Rij) and Tou values are discussed. As a result, in k-NNC before fracture signal is detected or when fracture signal is detected, showed that produce some empty classes in BPN. And we confirmed that could save trouble in AE signal processing if suitable error of convergence or acceptable encoding error give to BPN.

• 한국자동차공학회논문집
• /
• 제16권6호
• /
• pp.39-47
• /
• 2008

In this research, a web-based design support system is constructed for the design process of automotive steel pulley to gather engineering knowledge from pulley design data. In the design search module, a clustering tool for design data is proposed using K-means clustering algorithm. To obtain correlational patterns between design and FEA (Finite Element Analysis) data, a Multi-layer Back Propagation Network (MBPN) is applied. With the analyzed patterns from a number of simulation data, an estimation of minimum von mises can be provided for given design parameters of pulleys. The case study revealed fast estimation of minimum stress in the pulley within 12% error.

• International Journal of Quality Innovation
• /
• 제3권2호
• /
• pp.113-131
• /
• 2002

The relationships among multi-dimensional data (such as medical examination data) with ambiguity and variation are difficult to explore. The traditional approach to building a data classification system requires the formulation of rules by which the input data can be analyzed. The formulation of such rules is very difficult with large sets of input data. This paper first describes two classification approaches using back-propagation (BP) neural network and Mahalanobis distance (MD) classifier, and then proposes two classification approaches for multi-dimensional feature selection. The first one proposed is a feature selection procedure from the trained back-propagation (BP) neural network. The basic idea of this procedure is to compare the multiplication weights between input and hidden layer and hidden and output layer. In order to simplify the structure, only the multiplication weights of large absolute values are used. The second approach is Mahalanobis-Taguchi system (MTS) originally suggested by Dr. Taguchi. The MTS performs Taguchi's fractional factorial design based on the Mahalanobis distance as a performance metric. We combine the automatic thresholding with MD: it can deal with a reduced model, which is the focus of this paper In this work, two case studies will be used as examples to compare and discuss the complete and reduced models employing BP neural network and MD classifier. The implementation results show that proposed approaches are effective and powerful for the classification.

• 한국추진공학회:학술대회논문집
• /
• 한국추진공학회 2012년도 제38회 춘계학술대회논문집
• /
• pp.304-308
• /
• 2012

본 연구는 에틸렌-공기 혼합물로 채워져 있는 변형 가능한 구리 관에서의 초음속 화염 전파를 수치적으로 살펴보았다. 탄화수소의 화염 전파를 해석하기 위하여 지배방정식으로 Navier-Stokes 방정식과 Arrhenius 형태의 1단계 화학 반응식을 활용하였으며 변형 가능한 관을 해석하기 위하여 Inviscid Euler 방정식을 활용하였다. 또한, 두 물질 간 경계면 추적을 위하여 Level-set 기법을, 경계값 결정을 위하여 ghost fluid 기법을 사용하였다. 이러한 수치적 기법을 바탕으로 관의 변화에 따른 초음속 화염 내 팽창파의 전파 및 그에 따른 간섭 현상을 밀도 및 속도 변화를 통해 확인하였으며 초음속 화염 전파에도 안전성이 확보되는 최소 관 두께를 예측할 수 있는 수치적 기반을 마련하였다.

• Structural Engineering and Mechanics
• /
• 제63권3호
• /
• pp.329-345
• /
• 2017

This paper devoted to study dynamic interaction between crack and inclusion or void by developing the eXtended Finite Element Methods (XFEM). A novel XFEM approximation is presented for these structures containing multi discontinuities (void, inclusion, and crack). The level set methods are used so that elements that include a crack segment, the boundary of a void, or the boundary of an inclusion are not required to conform to discontinuous edges. The investigation covers the effects of a single circular or elliptical void / stiff inclusion, and multi stiff inclusions on the crack propagation path under dynamic loads. Both the void and the inclusion have a significant effect on the dynamic crack propagation path. The crack initially curves towards into the void, then, the crack moves round the void and propagates away the void. If a large void lies in front of crack tip, the crack may propagate into the void. If an enough small void lies in front of crack tip, the void may have a slight or no influence on the crack propagation path. For a stiff inclusion, the crack initially propagates away the inclusion, then, after the crack moves round the inclusion, it starts to propagate along its original path. As ${\delta}$ (the ratio of the elastic modulus of the inclusion to that of the matrix) increases, a larger curvature of the crack path deflection can be observed. However, as ${\delta}$ increases from 2 to 10, the curvature has an evident increase. By comparison, the curvature has a slight increase, as ${\delta}$ increases from 10 to 1000.

• Nuclear Engineering and Technology
• /
• 제52권8호
• /
• pp.1626-1637
• /
• 2020

This work covers an important point of the benchmark released by the expert group on Uncertainty Analysis in Modeling of Light Water Reactors. This ambitious benchmark aims to determine the uncertainty in light water reactors systems and processes in all stages of calculation, with emphasis on multi-physics (coupled) and multi-scale simulations. The Gesellschaft für Anlagen und Reaktorsicherheit methodology is used to propagate the thermal-hydraulic uncertainty of macroscopic parameters through TRACE5.0p3/PARCSv3.0 coupled code. The main innovative points achieved in this work are i) a new thermal-hydraulic model is developed with a highly-accurate 3D core discretization plus an iterative process is presented to adjust the 3D bypass flow, ii) a control rod insertion occurrence -which data is obtained from a real PWR test- is used as a transient simulation, iii) two approaches are used for the propagation process: maximum response where the uncertainty and sensitivity analysis is performed for the maximum absolute response and index dependent where the uncertainty and sensitivity analysis is performed at each time step, and iv) RESTING MATLAB code is developed to automate the model generation process and, then, propagate the thermal-hydraulic uncertainty. The input uncertainty information is found in related literature or, if not found, defined based on expert judgment. This paper, first, presents the Gesellschaft für Anlagen und Reaktorsicherheit methodology to propagate the uncertainty in thermal-hydraulic macroscopic parameters and, then, shows the results when the methodology is applied to a PWR reactor.

• 지구물리와물리탐사
• /
• 제21권4호
• /
• pp.213-219
• /
• 2018

본 연구에서는 제온 파이 x200 프로세서를 이용하여 3차원 파동 전파 모델링을 수행하고 기존의 제온 CPU를 사용한 경우와 병렬 연산 성능을 비교하였다. 제온 파이 1세대 프로세서인 제온 파이 나이츠 코너 보조프로세서와 달리 제온 파이 2세대 프로세서인 x200 프로세서는 직접 운영체제 실행이 가능하므로 내장 메모리와 주메모리 사이의 추가적인 통신이 필요 없다. 또한 제온 파이 x200 프로세서는 대용량 주메모리와 고대역폭 메모리를 이용하여 대규모 컴퓨팅을 독립적으로 실행할 수 있다. 병렬 연산 성능 비교를 위해 MPI (Message Passing Interface)와 OpenMP (Open Multi-Processing)를 이용해 모델링을 수행하였다. SEG/EAGE 암염돔 모델을 이용한 수치 실험 결과 제온 파이에서 다량의 연산 코어와 고대역폭 메모리를 이용해 12 코어 CPU 대비 2.69 ~ 3.24배 우수한 모델링 성능을 얻을 수 있었다.

• 대한토목학회논문집
• /
• 제26권4A호
• /
• pp.639-645
• /
• 2006

FRP(Fiber Reinforced Polymer)는 부식의 저항성, 고강도, 피로저항 능력 및 성형성 등에서 우수한 건설 신소재이다. 광섬유 브래그 격자(Fiber Bragg Grating; FBG) 센서는 전자기 저항, 작은 소재의 크기, 그리고 높은 내구성 등의 이점으로 smart sensor로서 현재 많이 사용되고 있다. 하지만 FBG 센서의 변위 측정 기술 능력의 부족으로 현재까지는 변형률, 온도 등의 물리량 측정센서로서 활용되고 있는 실정이다. 본 연구에서는 FRP와 FBG센서의 기능 복합화(Hybrid)를 통하여 smart FRP Rod를 개발 한 후 인장시험을 실시하였다. 또한, FBG센서에 의해 측정된 변형률 데이터를 신경망(Neural Network) 기법을 이용하여 변위 추정 모형을 개발함으로서 FBG 센서 단점인 변형률 계측만을 위한 센싱 역할을 극복하고자 한다. 인공신경망 모형은 MLP(Multi-layer Perceptron)로, 오차범위 0.001에 수렴 될 수 있도록 학습(training)을 실시하였다. 학습에는 비선형 목적함수와 역전파 학습(Back-propagation) 알고리즘을 적용하였으며 모형의 검증은 UTM에서 측정된 변위 값과 수치해석에 의한 결과 값을 비교함으로서 실시하였다.

• Journal of Electrical Engineering and Technology
• /
• 제13권6호
• /
• pp.2511-2520
• /
• 2018

In this study, Multi-Radial Basis Function Support Vector Machine (Multi-RBF SVM) classifier is introduced based on a composite kernel function. In the proposed multi-RBF support vector machine classifier, the input space is divided into several local subsets considered for extremely nonlinear classification tasks. Each local subset is expressed as nonlinear classification subspace and mapped into feature space by using kernel function. The composite kernel function employs the dual RBF structure. By capturing the nonlinear distribution knowledge of local subsets, the training data is mapped into higher feature space, then Multi-SVM classifier is realized by using the composite kernel function through optimization procedure similar to conventional SVM classifier. The original training data set is partitioned by using some unsupervised learning methods such as clustering methods. In this study, three types of clustering method are considered such as Affinity propagation (AP), Hard C-Mean (HCM) and Iterative Self-Organizing Data Analysis Technique Algorithm (ISODATA). Experimental results on benchmark machine learning datasets show that the proposed method improves the classification performance efficiently.