• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.44 no.2 s.314
• /
• pp.93-101
• /
• 2007

To jointly optimize the spatial registration and the exposure compensation, an iterative registration algorithm, the Lucas-Kanade algorithm, is combined with an exposure compensation algorithm, which is based on the histogram transformation function. Based on a simple regression model, a nonparametric estimator, the empirical conditional mean, and its polynomial fitting are used as histogram transformation functions for the exposure compensation. Since the proposed algorithm is composed of separable optimization phases, the proposed algorithm is more advantageous than the joint approaches of Mann and Candocia in the aspect of implementation flexibility. The proposed algorithm performs a better registration for real images than the case of registration that does not consider the exposure difference.

• Proceedings of the Korean Institute of Electrical and Electronic Material Engineers Conference
• /
• 2010.06a
• /
• pp.206-206
• /
• 2010

본 연구에서는 신경망을 이용하여 SiN 박막의 특성을 예측하는 모델을 개발하였다. 신경망으로는 일반화된 회귀 신경망 (generalized regression neural network-GRNN)을 이용하였고, GRNN 모델의 예측수행은 유전자 알고리즘 (genetic algorithm-GA)을 이용하여 최적화 하였다. 개발된 모델을 이용하여 증착률과 굴절률 및 균일도를 공정변수의 함수로 예측하였다.

• Water for future
• /
• v.23 no.4
• /
• pp.421-434
• /
• 1990

The problems of hydrologic similarity among river basins was analyzed by a geomorphologic response model using Hortons*s ordering scheme. The Nash model was used for deriving the geomorphologic response function, and for the optimization of the responsefunction, imcomplete gamma function andRosso*s regression equation were used. The application of this method was tested on some observed flood data of Pyungchang river basin and Wi Stream basin and Bocheong stream, and predictions of hydrologic response were compared with that of the Moment method. The results show that the proposed model and dimensionless instantaneous unit hydrograph can be used for the runoff analysis of an ungauged basin and the analysis of hydrologic similarity.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2004.04a
• /
• pp.442-445
• /
• 2004

본 논문은 유전자 기반 퍼지다항식 뉴럴네트워크(Genetic based fuzzy polynomial neural networks： gFPNN)를 제안한다. gFPNN 구조는 퍼지집합을 기반으로 설계되며, 유전자 알고리즘에 의해 구조 및 파라미터를 최적화한 구조이다. 퍼지집합을 기반으로 설계되어진 퍼지뉴럴네트워크는 간략추론 구조와 선형추론 구조로 설계된다. 본 논문에서는 간략추론 및 선형추론 구조를 통합 및 확장한 퍼지다항식 뉴럴네트워크를 설계한다. 이 구조는 연결가중치를 이용하여 회귀다항식을 네트워크 구조로 표현하며, 간략추론(Type 0), 선형추론(Type 1), 회귀다항식추론(Type 2)을 모두 포함한다. 또한 퍼지규칙 후반부의 다항식 차수를 각 규칙에 대해 다르게 선택할 수 있으며, 일률적인 형식의 구조를 벗어나 주어진 시스템의 특성에 따라 유연한 구조를 설계할 수 있도록 한다. 여기에 더하여, 네트워크 구조와 파라미터 동조에 유전자 알고리즘을 적용하며, 구조와 파라미터 동정에 대한 효율적인 방법을 논의한다. 제안된 모델의 평가를 위해 수치예제를 이용한다.

• Annual Conference of KIPS
• /
• 2001.10a
• /
• pp.585-588
• /
• 2001

현실 세계의 많은 조합 최적화 문제들은 변수들이 강하게 상호 작용함에 따라 만족해야하는 목표 함수가 매우 복잡하게 주어진다. 복잡한 목표 함수에는 많은 지역 최적해들이 존재하기 때문에 전역 최적해를 얻는 것은 엄청난 시간을 필요로 한다. 이러한 문제에 대해 해를 찾는 방법의 하나로 평균장 어닐링 알고리즘(MFA)이 있다. 본 논문에서는 기존의 이진 상태 공간에만 적용할 수 있었던 평균장 어닐링 알고리즘을 연속 상태 공간을 갖는 조합 최적화 문제에 적용할 수 있도록 알고리즘을 수정, 제안한다. 그리고 제안된 알고리즘을 제한된 연속 상태 공간을 가지는 단순 회귀 모델의 D-최적 설계에 적용하였다. 실험결과 제안된 알고리즘이 시뮬레이티드 어닐링 알고리즘(SSA)과 비교하여 비슷한 수준의 결과를 내면서도 계산 속도면에서는 수 배 정도의 빠른 좋은 결과를 얻었다.

• Applied Chemistry for Engineering
• /
• v.26 no.6
• /
• pp.730-736
• /
• 2015

Cyclotrimethylene trinitramine (RDX) is a high explosive commonly used for military applications. Submicronization of RDX particles has been a critical issue in order to alleviate the unintended and accidental stimuli toward safer and more powerful performances. The purpose of this study is to optimize experimental variables for drowning-out crystallization applied to produce submicron RDX particles. Effects of RDX concentration, anti-solvent temperature and anti-solvent mass were analyzed by the central composite rotatable design. The adjusted determination coefficient of regression model was calculated to be 0.9984 having the p-value less than 0.01. Response surface plots based on the central composite rotatable design determined the optimum conditions such as RDX concentration of 3 wt%, anti-solvent temperature of $0.2^{\circ}C$ and anti-solvent mass of 266 g. The optimum and experimental diameters of RDX particles were measured to be $0.53{\mu}m$ and $0.53{\mu}m$, respectively. The regression model satisfactorily predicts the average diameter of RDX particles prepared by drowning-out crystallization. Structure of RDX crystals was found to be ${\alpha}$-form by X-ray diffraction analysis and FT-IR spectroscopy.

• The Journal of the Acoustical Society of Korea
• /
• v.29 no.2
• /
• pp.118-124
• /
• 2010

In this paper, design parameters of dynamic vibration absorber, which is used to reduce bending vibrations of a vehicle drive line, is optimized. For obtaining the correct dynamic response characteristics, a flexible-body drive line is made by applying the flexibility data extracted from vibration analysis of propeller shafts to the drive line dynamic model. Inner tube mass, rubber stiffness and rubber damping coefficient of the dynamic vibration absorber are taken as design parameters for optimization. To minimize the vertical acceleration of the drive line, a second-order regression equation of the objective function is generated by performing the central composite experimental design with 3 factors, 2 levels and 15 test runs. And the design parameters of the dynamic vibration absorber are determined by using optimization program. The vehicle model with optimized dynamic vibration absorber reduces the vertical acceleration peak of the drive line by 17.1 % in compared with the initial model.

• Journal of the Korean GEO-environmental Society
• /
• v.25 no.11
• /
• pp.5-12
• /
• 2024

In this study, the liquefaction potential index was assessed using actual borehole data and seismic waves, and a predictive model was developed based on machine learning methods. A total of 10 features were selected including factors reflecting the characteristics of the seismic waves. To identify candidate methods, a preliminary test was conducted using commonly used machine learning methods for regression, followed by Bayesian optimization to optimize the hyperparameters for these candidate methods. Among artificial neural networks, Gaussian process regression, and random forest, it was found that the random forest effectively predicted the liquefaction potential index, as indicated by a low root mean square error, a high coefficient of determination, and considerations regarding overfitting. However, it was noted that the model tends to underestimate the liquefaction potential index when the index was 5 or higher.

• Journal of the Korea Concrete Institute
• /
• v.16 no.2 s.80
• /
• pp.195-204
• /
• 2004

Poisson's ratio due to multiaxial creep of concrete reported by existing experimental works was controversial. Poisson's ratio calculated from measured strain is very sensitive to small experimental error. This sensitivity make it difficult to find out whether the Poisson's ratio varies with time or remain constant, and whether the Poisson's ratio has different value with stress states or not. A new approach method is needed to resolve the discrepancy and obtain reliable results. This paper presents analytical study on multiaxial creep test results. Microplane model as a new approach method is applied to optimally fitting the test data extracted from experimental studies on multiaxial creep of concrete. Double-power law is used as a model to present volumetric and deviatoric creep evolutions on a microplane. Six parameters representing the volumetric and deviatoric compliance functions are determined from regression analysis and the optimum fits accurately describe the test data. Poisson's ratio is calculated from the optimum fits and its value varies with time. Regression analysis is also performed assuming that Poisson's ratio remains constant with time. Four parameters are determined for this condition, and the error between the optimum fits and the test data is slightly larger than that for six parameter regression results. The constant Poisson's ratio with time is obtained from four parameter analysis results and the constant value can be used in practice without serious error.

• Journal of the Korean GEO-environmental Society
• /
• v.13 no.6
• /
• pp.51-57
• /
• 2012

This study investigated on the photolytic degradation of Triclosan by E-beam process. The optimization of process was investigated during a series of batch experiments by design of experiments(DOEs). The DOE was one of the statistical application that was used for designed the response surface to determine the effects of each parameters. The responses were applied as removal rate of Triclosan(%, $Y_1$) and TOC removal rate(%, $Y_2$). Two independent variables were concentration of Triclosan and irradiation intensity that were designed as "$x_1$" and irradiation intensity was designed as "$x_2$". The regression equation in coded parameter between the Triclosan removal efficiencies(%) and TOC removal efficiencies(%) was $Y_1=63-12.4335x_1+15.1835x_2+5.8125x{_1}^2-5.6875x{_2}^2-0.75x_1x_2(R^2=95.1%,\;R^2(Adj)=91.7%)$ and $Y_2=46-8.8462x_1+11.7175x_2-0.75x{_1}^2-6.25x{_2}^2(R^2=98.7%,\;R^2(Adj)=97.7%)$, respectively. The model predictions agreed well with the experimentally observed results $R^2$ and $R^2(Adj)$ over 90% within both of $Y_1$ and $Y_2$. This result shows that the regression model express well about the effects of parameters on E-beam process and the statistical method was successfully applied.