• Proceedings of The KACE
• /
• 2017.08a
• /
• pp.209-211
• /
• 2017

기계학습은 인공지능(AI, Artificial Intelligence)의 일종으로 다른 인공지능 알고리즘이 정해진 규칙을 기반으로 주어진 임무(Task)를 해결하는 것과는 달리, 기계학습은 수집된 Data를 기반으로 최적의 솔루션을 학습한 후 미래의 값들을 예측하거나 해석하는 방법을 사용하고 있다. 더욱이 인터넷을 통한 연결성의 확대와 컴퓨터의 연산능력 발전으로 가능하게 된 Big-Data를 기반으로 하고 있어 이전의 인공지능 알고리즘에 비해 월등한 성능을 보여주고 있다. 그러나 기계학습 알고리즘이 Data를 학습할 때 학습 결과를 사람이 해석하기에 너무 복잡하여 사람이 그 내부 구조를 이해하는 것은 사실상 불가능하고, 이에 따라 학습된 기계학습 모델의 단점 또는 한계 등을 알지 못하는 문제가 있다. 본 연구에서는 이러한 블랙박스화된 기계학습 알고리즘의 특성을 이해하기 위해, 기계학습 알고리즘이 특정 입력에 대한 결과를 예측할 때 어떤 입력들로 부터 영향을 많이 받는지 그리고 어떤 입력으로부터 영향을 적게 받는지를 알아보는 방법을 소개하고 기존 연구의 단점을 개선하기 위한 방법을 제시한다.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.11 no.6
• /
• pp.487-490
• /
• 2001

In this paper, we realize a system that moves a cursor with a finger using the circular pattern vector algorithm that in one of the image analysis algorithms. To apply this algorithm, we use central point of the biggest circle among the various circles that recognize the image of the hand , and find out the pointing finger by looking for the distance of the outline of the hand from the central point. The horizontal direction of the cursor on the display is controlled by converting the direction of the pointing finger to the analysis of the plane corrdinate. Because of setting up only one camera of the upper, the middle and the lower discretely. On account of the discrete movement of the cursor of the vertical direction, we move th cursor to the objective, which the user wants. by expanding the local are to the whole area.

• Journal of the Institute of Electronics Engineers of Korea SC
• /
• v.41 no.3
• /
• pp.25-32
• /
• 2004

An advanced algorithm for measurement of phase angle between two sinusoidal signals is proposed in this paper. This algorithm uses discrete sample data of two input signals for calculation of phase angle and amplitude. And the key parameters of the measurement algorithm are described by analytical express, so the calculation of phase angle is simplified. In this paper it is proved that harmonic distortion of the input sinusoidal signals does not affect the measurement value of phase angle by using the proposed algorithm when a whole cycle is sampled. And measurement error by the white Gaussian noise is very small compared by other algorithms.

• The Journal of the Acoustical Society of Korea
• /
• v.19 no.8
• /
• pp.12-21
• /
• 2000

In this paper, we propose a novel hearing aid algorithm for sensorinural hearing loss restoration with multi-channel(band) dynamic range compression and psychoacoustics. In this way, we can present a normal perception condition to the impaired listener. The proposed algorithm make loudness scaling function achieve proper loudness level, and analysis masking property for the signal will be perceived to impaired listener, and then, restore normal spectral contrast using SMR(signal-to-masking ratio) defined by distance between the level of each frequency and masking threshold.

• Journal of the Korean Society for Aeronautical & Space Sciences
• /
• v.41 no.11
• /
• pp.865-873
• /
• 2013

In this paper, numerical methods for wrinkle prediction of thin membrane were studied by finite element analysis. Techniques using membrane and shell elements were applied for triangular membrane. In case of membrane element method, the wrinkling was accounted for by the wrinkle algorithm of property modification, which was implemented to ABAQUS as a user subroutine. In case of shell method, geometrically nonlinear post-buckling analysis was performed to obtain the wrinkle deformation explicitly. The wrinkling deformation was induced by seeding the mesh with a random geometric imperfection. The results were investigated focusing on the mesh convergence and the solution accuracy.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2022.05a
• /
• pp.247-247
• /
• 2022

제트류는 복잡한 흐름 중 하나로 다양한 크기의 에디가 다양한 운동량을 가지고 있다. 이러한 제트류를 구현하기 위해서는 난류 운동 에너지 등 제트류의 특성을 잘 반영하여야 한다. 제트를 구현하기 위해서는 수리학적 모델, 현장 실험 등 많은 방법이 있으며, 본 연구에서는 상대적으로 공간, 시간적 비용이 적게 드는 수치해석 방법을 사용하여 연구를 진행하였다. 대표적인 수치해석방법에는 DNS(Direct Numerical Simulation), LES(Large Eddy Simulation), RANS(Reynolds Averaged Navier Stokes) 등이 있다. RANS는 시간 평균 흐름 특성만 산출하며 제트의 복잡성을 재현하는 데 한계가 있어, 본 연구는 DNS와 LES 모델을 이용하여 제트류를 구현하는 것에 초점을 맞추었다. DNS는 해당 격자에서 발생하는 모든 에디를 직접 해석 때문에 난류 모델링이 필요하지 않지만, 많은 수의 그리드가 필요하여 수치해석 시 소요시간이 긴 편이다. LES는 대규모 에디는 직접 해석하지만 일정 크기 이하의 소용돌이를 해석하기 위해서 모델이 필요하다. 따라서 서브 그리드 모델에 따라 약간 다른 결과를 보인다. 이러한 문제점을 해결하기 위해 본 연구에서는 LES의 기존 서브 그리드 모델을 사용하지 않고 신경망 모델로 학습한 DNS 결과를 활용하는 방법을 제안한다. 우선 DNS와 LES 모델을 사용하여 에너지 스펙트럼을 비교하여 서브 그리드 모델이 시작하는 파수를 찾는다. 이후 특정 파수 아래의 작은 에디를 모사할 적절한 신경망 모델을 결정하여 DNS의 작은 에디를 신경망 알고리즘이 모사할 수 있도록 학습시킨다. 이후 기존 서브 그리드 모델을 사용하지 않고 학습된 신경망 알고리즘을 사용한 LES 모델이 모사한 제트류와 실제 DNS 모델을 사용한 제트류를 비교 및 평가한다.

• Journal of Korean Society of Steel Construction
• /
• v.16 no.2 s.69
• /
• pp.201-213
• /
• 2004

A GA-based optimum design algorithm and a program for plane steel frame structures with semi-rigid connections are presented. The algorithm is incorporated with the refined plastic hinge analysis method wherein geometric nonlinearity is considered by using the stability functions of beam-column members, and material nonlinearity, by using the gradual stiffness degradation model that includes the effects of residual stresses, moment redistribution through the occurrence of plastic hinges, semi-rigid connections, and geometric imperfection of members. In the genetic algorithm, the tournament selection method and micro-GAs are employed. The fitness function for the genetic algorithm is expressed as an unconstrained function composed of objective and penalty functions. The objective and penalty functions are expressed as the weight of steel frames and the constraint functions, respectively. In particular, the constraint functions fulfill the requirements of load-carrying capacity, serviceability, ductility, and construction workability. To verify the appropriateness of the present method, the optimal design results of two plane steel frames with rigid and semi-rigid connections are compared.

• Journal of the Korean Society of Propulsion Engineers
• /
• v.9 no.4
• /
• pp.81-88
• /
• 2005

In order to estimate the gas turbine engine performance precisely, the component maps containing their own performance characteristics should be needed. In this study a component map generation method which may identify compressor map conversely from a performance deck provided by engine manufacturer using genetic algorithms was newly proposed. As a demonstration example for this study, the PW 206C turbo shaft engine for the tilt rotor type Smart UAV(Unmanned Aerial Vehicle). In order to verify the proposed method, steady-state performance analysis results using the newly generated compressor map was compared with them performed by EEPP(Estimated Engine Performance Program) deck provided by engine manufacturer. And also the performance results using the identified maps were compared with them using the traditional scaling method. When the performance analysis is performed at far away operation conditions from the design point, in case of use of e component map by the traditional scaling method, the error of the performance analysis results is greatly increasing. In the other hand, if in case of use of the compressor map generated by the proposed GAs scheme, the performance analysis results are closely met with those by the performance deck, EEPP.

• Journal of Korean Tunnelling and Underground Space Association
• /
• v.15 no.3
• /
• pp.357-373
• /
• 2013

In this paper, a back analysis program for analyzing the behavior of tunnel-ground system and evaluating the material properties and tunnel design parameters was developed. This program was designed to be able to implement the back analysis of underground structure by combination of using FLAC and optimized algorithm as direct method. In particular, Rosenbrock method which is able to do direct search without obtaining differential coefficient was adopted for the back analysis algorithm among optimization methods. This back analysis program was applied to the site to evaluate the design parameters. The back analysis was carried out using field measurement results from 5 sites. In the course of back analysis, nonlinear regression analysis was carried out to identify the optimum function of the measured ground displacement. Exponential function and fractional function were used for the regression analysis and total displacement calculated by optimum function was used as the back analysis input data. As a result, displacement recalculated through the back analysis using measured displacement of the structure showed 4.5% of error factor comparing to the measured data. Hence, the program developed in this study proved to be effectively applicable to tunnel analysis.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.30 no.3C
• /
• pp.83-91
• /
• 2005

We analyze the performance of an adaptive hybrid search code acquisition algorithm for direct-sequence code division multiple access (DS-CDMA) systems under slowly-moving mobile environments. The code acquisition algorithm is designed to provide the desired feature of constant false alarm rate (CFAR) to cope with nonstationarity of the interference in CDMA forward links. An analytical expression for the mean acquisition time is first derived and the probabilities of detection, miss, and false alarm are then obtained for frequency-selective Rayleigh fading environments. The fading envelope of a received signal is assumed to be constant over the duration of post-detection integration (PDI), considering slow fading environments. Finally, the performance of the designed code acquisition algorithm shall be evaluated numerically to examine the effect of some design parameters such as the sub-window size, size of the PDI, decision threshold, and so on, considering cdma2000 environments.