• 대한기계학회:학술대회논문집
• /
• 대한기계학회 2007년도 춘계학술대회B
• /
• pp.2666-2671
• /
• 2007

This work presents microchannel heat sink shape optimization procedure using Kriging method. Design variables relating to microchannel width, depth and fin width are selected, and thermal resistance has been taken as objective function. Design points are selected through a three-level fractional factorial design of sampling method. Navier-Stokes and energy equations for laminar flow and conjugate heat transfer are solved at these design points using a finite volume solver. Solutions are carefully validated with experimental results. Using the numerically evaluated objective function, a surrogate model (Kriging) is constructed and optimum point is searched by sequential quadratic programming. The process of shape optimization greatly improves the thermal performance of microchannel heat sink under constant pumping power.

• 대한기계학회논문집B
• /
• 제31권9호
• /
• pp.733-740
• /
• 2007

Microchannel heat sink shape optimization is performed using Kriging method. Design variables relating to microchannel width, depth and fin width are selected, and thermal resistance has been taken as objective function. Design points are selected through a three-level fractional factorial design of sampling method. Navier-Stokes and energy equations for laminar flow and conjugate heat transfer are solved at these design points using a finite volume solver. Solutions are carefully validated with experimental results. Using the numerically evaluated objective function, a surrogate model (Kriging) is constructed and optimum point is searched by sequential quadratic programming. The process of shape optimization greatly improves the thermal performance of microchannel heat sink under constant pumping power.

• 한국소성가공학회:학술대회논문집
• /
• 한국소성가공학회 1994년도 추계학술대회 논문집
• /
• pp.133-140
• /
• 1994

In this paper, we have proposed a new method to determine the initial billet for the forged products using a function approximation in neural networks. the architecture of neural network is a three-layer neural network and the back propagation algorithm is employed to train the network. By utilizing the ability of function approximation of neural network, an optimal billet is determined by applying nonlinear mathematical relationship between shape ratio in the initial billet and the final products. A volume of incomplete filling in the die is measured by the rigid-plastic finite element method. The neural network is trained with the initial billet shape ratio and that of the un-filled volume. After learning, the system is able to predict the filling region which are exactly the same or slightly different to results of finite element method. It is found that the prediction of the filling shape ratio region can be made successfully and the finite element method results are represented better by the neural network.

• 한국유체기계학회 논문집
• /
• 제14권2호
• /
• pp.47-51
• /
• 2011

This paper describes the shape optimization of a blower impeller used for a refuse collection system. Two design variables, which are used to define the blade angles of an impeller, are introduced to increase the blower performance. A blower efficiency is selected as an object function, and the shape optimization of the blade angles is performed by a response surface method (RSM). Three-dimensional Navier-Stokes equations are introduced to analyze the internal flow of the blower and to find the value of object function for the training data. Relatively good agreement between experimental measurements and numerical simulation is obtained in the present study. Throughout the shape optimization, blower efficiency for the optimal blade angles is successfully increased up to 3.6% compared with that of reference at the design flow rate. Detailed flow field inside the turbo blower is also analyzed and discussed.

• 소음진동
• /
• 제8권3호
• /
• pp.494-503
• /
• 1998

In this paper, in order to analyze dynamic characteristics of an automobile steering system consisting of many components, natural frequencies and transfer functions of each component and the total system are found on a FFT analyzer by experiments. Then, the data are transmitted to a commercial package program, CADA-PC. By analyzing the data, the mode shape of each natural frequency and damping values are obtained. Also, the function of a rubber coupling in column and telescoping effects on system are considered. C.A.E commercial programs are used to compare with the results of experiments. For the finite element modeling, I-DEAS is used. Data processing and post processing are operated on NASTRAN and XL, respectively. The ball-bearing and the linkage of shaft with column are modeled by spring element. Stiffness is modified from the results of experiments. The results of those show close agreement. In the mode shape of total system, wheel mode is dominant at lower frequency, while the column mode is main mode at higher. The role of rubber coupling in vibration isolation is clear on mode shape. Telescoping function makes natural frequency of column changed.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• 제8권4호
• /
• pp.306-311
• /
• 2008

Fuzzy regression, a nonparametric method, can be quite useful in estimating the relationships among variables where the available data are very limited and imprecise. It can also serve as a sound methodology that can be applied to a variety of management and engineering problems where variables are interacting in an uncertain, qualitative, and fuzzy way. A close examination of the fuzzy regression algorithm reveals that the resulting possibility distribution of fuzzy parameters, which makes this technique attractive in a fuzzy environment, is dependent upon an h parameter value. The h value, which is between 0 and 1, is referred to as the degree of fit of the estimated fuzzy linear model to the given data, and is subjectively selected by a decision maker (DM) as an input to the model. The selection of a proper value of h is important in fuzzy regression, because it determines the range of the posibility ditributions of the fuzzy parameters. In this paper, we discuss the interdependent relationship among the h value, membership function shape, and the spreads of fuzzy parameters in fuzzy linear regression with fuzzy input-output using shape-preserving operations.

• 한국음향학회지
• /
• 제27권6호
• /
• pp.271-278
• /
• 2008

압전판의 길이나 폭에 따라 두께가 변화하는 압전 세라믹을 사용하여 광대역 특성을 구현하는 초음파 트랜스듀서에 대하여 압전진동자의 측면에서 본 형태에 따른 전기-음향적 특성해석을 이론적으로 수행하였다. 압전진동자의 길이방향에 따른 두께 변화를 지수함수의 조합으로 표현하고 이 함수를 이용하여 압전진동자의 전기단자에서 본 자유 어드미턴스 및 파워전달함수에 대한 식을 도출하였다. 대표적인 PZT압전 세라믹을 예로 들어 압전 진동자의 측면 형태변화에 따른 비 대역폭을 고찰해본 결과 넓은 대역폭를 얻기 위한 최적의 형태가 존재함을 알 수 있었으며, 압전진동자의 최대두께에 대한 최소두께 비가 자아짐에 따라 대역폭은 100%이상까지도 넓어질 수 있으나 파워전달함수는 반대로 감소하는 경향을 확인할 수 있었다. 또 압전진동자의 길이가 길어질수록 전달함수의 크기는 증가하나 광대역 특성을 갖는 압전진동자의 형태는 매우 한정적이 됨을 확인할 수 있었으며 이는 고효율의 광대역 초음파 트렌스듀서 제작에 있어서는 정밀한 가공이 요구됨을 확인할 수 있었다.

• Journal of applied mathematics & informatics
• /
• 제23권1_2호
• /
• pp.397-410
• /
• 2007

Visualization of 2D and 3D data, which arises from some scientific phenomena, physical model or mathematical formula, in the form of curve or surface view is one of the important topics in Computer Graphics. The problem gets critically important when data possesses some inherent shape feature. For example, it may have positive feature in one instance and monotone in the other. This paper is concerned with the solution of similar problems when data has convex shape and its visualization is required to have similar inherent features to that of data. A rational cubic function [5] has been used for the review of visualization of 2D data. After that it has been generalized for the visualization of 3D data. Moreover, simple sufficient constraints are made on the free parameters in the description of rational bicubic functions to visualize the 3D convex data in the view of convex surfaces.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제11권8호
• /
• pp.3950-3964
• /
• 2017

With the increasing popularity of 3D technology such as 3D printing, 3D modeling, etc., there is a growing need to search for similar models on the internet. Matching non-rigid shapes has become an active research field in computer graphics. In this paper, we present an efficient and effective non-rigid model retrieval method based on topological structure and local volume. The integral geodesic distances are first calculated for each vertex on a mesh to construct the topological structure. Next, each node on the topological structure is assigned a local volume that is calculated using the shape diameter function (SDF). Finally, we utilize the Hungarian algorithm to measure similarity between two non-rigid models. Experimental results on the latest benchmark (SHREC' 15 Non-rigid 3D Shape Retrieval) demonstrate that our method works well compared to the state-of-the-art.

• 한국CDE학회논문집
• /
• 제2권3호
• /
• pp.175-185
• /
• 1997

Given the widespread use of the Finite Element Method in strength analysis, automatic mesh generation is an important component in the computer-aided design of parts and assemblies. For a given resolution of geometric accuracy, the purpose of mesh generators is to discretize the continuous model of a part within this error limit. Sticking to this condition often produces many small elements around small features in spite that these regions are usually of little interest and computer resources are thus wasted. Therefore, it is desirable to selectively suppress small features from the model before discretization. This can be achieved by low-pass filtering a CAD model. A spatial function of one dimension higher than the model of interest is represented using the Fourier basis functions and the region where the function yields a value greater than a prescribed value is considered as the extent of a shape. Subsequently, the spatial function is low-pass filtered, yielding a shape without the small features. As an undesirable effect to this operation, all sharp corners are rounded. Preservation of sharp corners is important since stress concentrations might occur there. This is why the LPF (low-pass filtered) model can not be directly used. Instead, the distances of the boundary elements of the original shape from the LPF model are calculated and those that are far from the LPF model are identified and removed. It is shown that the number of mesh elements generated on the simplified model is much less than that of the original model.