• Proceedings of the KIEE Conference
• /
• 2001.07d
• /
• pp.2663-2665
• /
• 2001

역전파 신경망은 반도체 공정 모델링에 효과적으로 응용이 되고 있으며, 최근 선형뉴런을 비선형 함수 대신 출력층에 이용하여 모델의 예측정확도를 향상 시킨 바 있다. 본 연구에서는 그 원인을 규명하기 위한 모델의 평가지표로서의 유클리디언 웨이트 거리(Euclidean Weight Distance)를 제안한다. 이 지표를 이용하여 신경망의 입력층과 은닉층, 그리고 은닉층과 출력층의 웨이트를 감시하였으며, 그 결과 예측정확도의 향상이 이 지표의 감소에 기인하고 있음을 알았다. 모델링에 이용한 실험데이터는 다중 유도결합형 플라즈마 장비로부터 Langmuir Probe 진단 시스템을 이용하여 수집하였다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2018.10a
• /
• pp.380-382
• /
• 2018

사이버 물리 시스템의 발달은 네트워크 기반 제어 시스템을 다양한 형태로 진화시키고 있다. 네트워크에서는 전송 지연과 전송 오류에 의해서 다양한 불규칙한 지연이 발생하고 지연에 강인한 제어기 설계가 요구되고 있다. 마코프 프로세서를 따르는 지연을 갖는 제어 시스템에서 제어기를 설계할 때에 많이 사용되는 모델중에 하나가 확장된 상태 벡터를 기반으로 한 마코프 점프 선형 시스템 모델이다. 본 연구에서는 이 모델을 이용하여 리아푸노프 안전성 조건으로부터 유도되는 선형 행렬 부등식의 안정성 조건으로부터 저복잡도를 가지고 제어기를 유도하는 직관적 방법을 제안한다. 제안된 알고리즘은 모의 실험을 통해서 성능을 확인하였으나, 직관적 방법 구조상에서 제한 조건의 수가 자유 변수의 수보다 많은 이유 때문에 성능이 배우 제한적임을 확인하였고, 이를 개선하기 위한 향후 연구가 요청됨을 확인하였다.

• Proceedings of the Korean Institute of Surface Engineering Conference
• /
• 2017.05a
• /
• pp.124-124
• /
• 2017

Fe-Ni-Co 합금 박막(<$100{\mu}m$)을 황화물계 용액에서 전주공정으로 제조하였다. XRF로 측정한 박판의 평균 조성은 Fe-34 wt.% Ni-3 wt.% Co 이다. AFM으로 측정한 표면 조도는 35.2 nm 이다. 표면의 나노 경도는 평균 5.4 GPa 이었다. Oliver 모델을 적용한 구리 박막의 탄성하강강성도는 약 75 이었다. Alekhin 모델을 적용한 구리 박막의 마찰계수, 피로한계는 각각 0.134, 0.027 이었다. 유한요소법으로 평가한 Berkovich 형 나노압침선단의 하중분포를 이차원 선형 및 비선형 해석하면 1 [mN]의 정적하중을 가한 Fe-Ni-Co 박막은 약 576 [mN]로 예측되었다. 압침선단의 하중집중정도는 표면탐침현미경으로 관찰한 압흔의 변형정도와 유사한 경향을 보였다.

• Journal of the Korea Computer Graphics Society
• /
• v.29 no.3
• /
• pp.117-125
• /
• 2023

In the development of clothing design through virtual simulation, it is essential to minimize the differences between the virtual and the real world as much as possible. The most critical task to enhance the similarity between virtual and real garments is to find simulation parameters that can closely emulate the physical properties of the actual fabric in use. The simulation parameter optimization process requires manual tuning by experts, demanding high expertise and a significant amount of time. Especially, considerable time is consumed in repeatedly running simulations to check the results of applying the tuned simulation parameters. Recently, to tackle this issue, artificial neural network learning models have been proposed that swiftly estimate the results of drape test simulations, which are predominantly used for parameter tuning. In these earlier studies, relatively simple linear stiffness models were used, and instead of estimating the entirety of the drape mesh, they estimated only a portion of the mesh and interpolated the rest. However, there is still a scarcity of research on non-linear stiffness models, which are commonly used in actual garment design. In this paper, we propose a learning model for estimating the results of drape simulations for non-linear stiffness models. Our learning model estimates the full high-resolution mesh model of drape. To validate the performance of the proposed method, experiments were conducted using three different drape test methods, demonstrating high accuracy in estimation.

• The Journal of Engineering Geology
• /
• v.19 no.1
• /
• pp.71-79
• /
• 2009

Step-drawdown tests were conducted at four pumping Wells, two in porous media and two in fractured rocks, respectively. In general, P = 2.0 suggested by Jacob (1947) is applied to porous media and fractured rocks in terms of drawdowns of step-drawdown test. In an attempt to review problems of linear model (Jacob's graphic method) in interpreting the step-draw down test, the outcomes of well parameters (aquifer loss coefficient (B), well loss coefficient (C) and well loss exponent (P)) calculated from linear and nonlinear model (Labadie and Helweg's least-squares method) were compared and analyzed. The values of C and P calculated from linear and nonlinear models differed according to permeability of aquifer and the conditions of pumping well. The value C obtained from nonlinear models in porous media and fractured rocks is about $10^0{\sim}10^{-2}$ and $10^{-3}{\sim}10^{-6}$ times lower than in their linear models, respectively. The value P of porous media obtained from nonlinear model ranged from 2.123 to 2.775, while it ranged from 3.459 to 5.635 for fractured rocks. In case of nonlinear model, well loss highly depends on the value P. At this time, well efficiencies calculated from linear and nonlinear models were $1.56{\sim}14.89%$ for porous media and $8.73{\sim}24.71%$ for fractured rocks, showing a significant error according to chosen models. In nonlinear model, it was found that the regression analysis using the least squares method was very useful to interpret step-drawdown test in all aquifer.

• International Journal of Highway Engineering
• /
• v.10 no.1
• /
• pp.1-9
• /
• 2008

Five bases for of highways are sight distance, vehicle stability, driver comfort, drainage and aesthetics. Among these bases, the sight distance is the most critical one in the highway alignment design. It is certain that highways and drivers' observation actions are 3-dimensional in nature. But, the highway alignment design and the sight distance analysis have been performed on 2-dimensional planes because of various constraints and complexity of calculating 3-dimensional sight distance. The purpose of this study is to derive a model for determining exact available sight distance in 3-D combined horizontal and vertical alignments. Application of the model is illustrated using numerical examples. The proposed model should be useful in quantifying the criteria for 3-D highway alignments.

• Journal of Korean Society of Steel Construction
• /
• v.14 no.6
• /
• pp.729-734
• /
• 2002

The effect of several simplified methods of seismic analysis is estimated. The pounding/contacting of superstructures were considered in the multispan simply supported bridge and the multispan continuous bridge. Although nonlinear time history analysis is generally used for seismic analysis of bridges, many codes including AASHTO propose several simplified analysis methods. AASHTO, however, does not mention pounding. Therefore, the simplified methods may produce results that are different from those of nonlinear time history analysis. This study developed nonlinear analytical models of the two types of bridges mentioned. The models were then modified to the simplified linear models for simplified analysis. The results of the simplified methods were compared with those of nonlinear time history analysis. It was found that including of the pounding/contacting element in the simplified methods generated responses similar to those of the nonlinear time history analysis.

• Solar Energy
• /
• v.16 no.2
• /
• pp.79-86
• /
• 1996

In order to reduce damages due to cavitation effects and to improve performance of fluid machinery, supercavitation around the cascade and the hydraulic characteristics of supercavitating cascade must be analyzed accurately. And the study on the effects of cavitation on fluid machinery and analysis on the performances of supercavitating hydrofoil through various elements governing flow field are critically important. In this study comparison of experiment results with the computed results of linear theory using singularity method was obtainable. Specially singularity points like sources and vortexes on hydrofoil and freestreamline were distributed to analyze two dimensional flow field of supercavitating cascade, and governing equations of flow field were derived and hydraulic characteristics of cascade were calculated by numerical analysis of the governing equations.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.27 no.1
• /
• pp.29-34
• /
• 2017

Recently, studying chaotic dynamic have been concerned by many researchers in areas of physics, chemistry, mathematics, engineering and social science. Especially, model of addiction, happiness, family and love become major research subjects in the social science. Among these models, love is one of the four emotions that human being have. There are many definitions for love, however, each definitions of love does not coincide with each other. Recently, one of the most important efforts for research is love and it is represented by derivative equation. Then they try to find nonlinear or chaotic behavior from this derivative equation. This paper propose Gaussian fuzzy membership function in order to make external force that are close to action and awareness of human based on love model of Romeo and Juliet with external force. This paper also confirms the existence of nonlinear characteristics through time series and phase portrait.

• Phonetics and Speech Sciences
• /
• v.13 no.3
• /
• pp.71-78
• /
• 2021

Most neural-network-based speech synthesis models utilize neural vocoders to convert mel-scaled spectrograms into high-quality, human-like voices. However, neural vocoders combined with mel-scaled spectrogram prediction models demand considerable computer memory and time during the training phase and are subject to slow inference speeds in an environment where GPU is not used. This problem does not arise in linear spectrogram prediction models, as they do not use neural vocoders, but these models suffer from low voice quality. As a solution, this paper proposes a Tacotron 2 and Transformer-based linear spectrogram prediction model that produces high-quality speech and does not use neural vocoders. Experiments suggest that this model can serve as the foundation of a high-quality text-to-speech model with fast inference speed.