• 한국음향학회지
• /
• 제20권2호
• /
• pp.3-9
• /
• 2001

본 논문에서는 음성엔코더 (speech-waveform coder) 시뮬레이션에 사용할 수 있는 난수발생기를 설계하였다. 설계된 난수발생기는 규정된 확률밀도함수와 규정된 스펙트럼을 동시에 만족해야 하는 환경에서 필요로 하는 것이다. 선형필터와 메모리가 없는 비선형 회로로 구성되는 Sondhi 알고리즘 [1]을 기반으로 하여 난수발생기를 구현하였고, 균일 분포, 이진 분포, 감마 분포와 같은 연속적인 확률분포함수들을 얻기 위해 필요한 선형필터와 비선형 회로를 구현하는 방법을 연구하였다. 또한 분석적인 방법을 통하여 구현해야 하는 비선형 회로가 비대칭 형태로 표현되지 않는 경우 (카이-제곱 분포, lognormal 분포)와 수학식으로 표현할 수 없는 경우 (Student-t 분포, F 분포)에는 Sondhi 알고리즘이 동작하지 않음을 보였다.

• Fibers and Polymers
• /
• 제7권3호
• /
• pp.276-285
• /
• 2006

Development and numerical implementation for an elastoplastic constitutive model for anisotropic and asymmetric materials are presented in this paper. The Coulomb-Mohr yield criterion was modified to consider both the anisotropic and asymmetric properties. The modified yield criterion is an isotropic function of the principal values of a symmetric matrix which is linearly transformed from the Cauchy stress space. In addition to the constitutive equation, the numerical treatment for the singularity in the vertex region of yield surface and stress integration algorithm based on elastoplasticity were presented. In order to assess the accuracy of numerical algorithm, isoerror maps were considered. Also, extension of a strip with a circular hole was simulated and results compared with those obtained using the (smooth) Mises yield criterion to validate stress output for a complex stress state.

• Structural Engineering and Mechanics
• /
• 제66권3호
• /
• pp.329-341
• /
• 2018

This paper considers the smooth receding contact problem between a homogeneous half-plane and a composite laminate composed of an inhomogeneously coated elastic layer. The inhomogeneity of the elastic modulus of the coating is approximated by an exponential function along the thickness dimension. The three-component structure is pressed together by either a concentrated force or uniform pressures applied at the top surface of the composite laminate. Both semianalytical and finite element analysis are performed to solve for the extent of contact and the contact pressure. In the semianalytical formulation, Fourier integral transformation of governing equations and boundary conditions leads to a singular integral equation of Cauchy-type, which can be numerically integrated by Gauss-Chebyshev quadrature to a desired degree of accuracy. In the finite element modeling, the functionally graded coating is divided into homogeneous sublayers and the shear modulus of each sublayer is assigned at its lower boundary following the predefined exponential variation. In postprocessing, the stresses of any node belonging to sublayer interfaces are averaged over its surrounding elements. The results obtained from the semianalytical analysis are successfully validated against literature results and those of the finite element modeling. Extensive parametric studies suggest the practicability of optimizing the receding contact peak stress and the extent of contact in multilayered structures by the introduction of functionally graded coatings.

• Nuclear Engineering and Technology
• /
• 제50권7호
• /
• pp.1120-1130
• /
• 2018

An analysis of a pyroprocessing safeguards methodology employing the Pu-to-$^{244}Cm$ ratio is presented. The analysis includes characterization of representative used nuclear fuel assemblies with respect to computed nuclide composition. The nuclide composition data computationally generated is appropriately reformatted to correspond with the material conditions after each step in the head-end stage of pyroprocessing. Uncertainty in the Pu-to-$^{244}Cm$ ratio is evaluated using the Geary-Hinkley transformation method. This is because the Pu-to-$^{244}Cm$ ratio is a Cauchy distribution since it is the ratio of two normally distributed random variables. The calculated uncertainty of the Pu-to-$^{244}Cm$ ratio is propagated through the mass flow stream in the pyroprocessing steps. Finally, the probability of Type-I error for the plutonium Material Unaccounted For (MUF) is evaluated by the hypothesis testing method as a function of the sizes of powder particles and granules, which are dominant parameters to determine the sample size. The results show the probability of Type-I error is occasionally greater than 5%. However, increasing granule sample sizes could surmount the weakness of material accounting because of the non-uniformity of nuclide composition.

• Applied Microscopy
• /
• 제47권3호
• /
• pp.131-147
• /
• 2017

In this work further optical and electrical investigations of pure and Zr doped $Mn_3O_4$ (from 0 up to 20 at.%) thin films as a function of frequency. First, the refractive index, the extinction coefficient and the dielectric constants in terms of Zr content are reached from transmittance and reflectance data. The dispersion of the refractive index is discussed by means of Cauchy model and Wemple and DiDomenico single oscillator models. By exploiting these results, it was possible to estimate the plasma pulse ${\omega}_p$, the relaxation time ${\tau}$ and the dielectric constant ${\varepsilon}_{\infty}$. Second, we have performed original ac and dc conductivity studies inspired from Jonscher model and Arrhenius law. These studies helped establishing significant correlation between temperature, activation energy and Zr content. From the spectroscopy impedance analysis, we investigated the frequency relaxation phenomenon and hopping mechanisms of such thin films. Moreover, a special emphasis has been putted on the effect of the oxidation in air of hausmannite thin films to form $Mn_2O_3$ ones at $350^{\circ}C$. This intrigue phenomenon which occurred at such temperature is discussed along with this electrical study. Finally, all results have been discussed in terms of the thermal activation energies which were determined with two methods for both undoped and Zr doped $Mn_3O_4$ thin films in two temperature ranges.

• 한국결정성장학회지
• /
• 제16권4호
• /
• pp.133-140
• /
• 2006

The growth characteristics and properties of large size $SrAlF_5$ single crystals are described and compared with those of $BaMgF_4$. Transmission spectra in the vacuum ultraviolet wavelength region indicate a high transparency of $SrAlF_5$ (about 90% without considering surface reflection loses) down to 150 nm, on contrast to the optical loses observed for $BaMgF_4$. The ferroelectric character of $SrAlF_5$ is evidenced by the reversal of the spontaneous polarization in a hysteresis loop. The higher potential of $SrAlF_5$ in comparison with $BaMgF_4$ for the realization of all-solid-state lasers in the ultraviolet wavelength region by the quasi-phase matching (QPM) technique is pointed out. $SrAlF_5$, besides a higher grade of transparency, shows a nonlinear effective coefficient similar to that of quartz and uniaxial nature, on contrast to the one order smaller nonlinear coefficient and biaxial character of $BaMgF_4$. The refractive index of $SrAlF_5$ from the ultraviolet to the near-infrared wavelength region is measured by the minimum deviation method. The Sellmeier and Cauchy coefficients are obtained from the fits to the curves of the ordinary and extraordinary refractive indices, and the grating period for the first order QPM is estimated as a function of the wavelength. The poling periodicity for 193 nm SHG from 386 nm is $4{\mu}m$.

• 대한조선학회논문집
• /
• 제29권4호
• /
• pp.114-131
• /
• 1992

본 논문은 2차원 자유표면파문제에서 시간영역해법을 이용하여 2차원 운동문제에 적용할 수 있는 수치해석을 하였다. 경계조건으로는 엄밀한 물체표면 경계조건과 비선형 자유표면경계 조건을 부과했다. 수치해를 구하는데는 코시이론을 이용하여 제2종 프레드흘름 경계적분방정식을 도출하고 이를 이산화시켜 처리하였다. 수치계산을 위해 전영역을 유한한 영역으로 제한하여야 한다. 제한된 영역에서 방사해의 부과를 위해 전영역을 수치해석영역과 외부영역으로 나누고, 외부영역의 해는 그린 제2정리를 이용하고, 선형자유표면조건을 만족하는 과도그린함수를 사용한다. 위의 그린 제2정리를 이용한 식으로 부터 초기조건, 선형 자유표면조건, 무한원방조건을 이용하여 단순화시킨 다음 포텐셜과 유동함수의 관계식으로 치환하면 비선형해와 정합할 수 있는 정합행렬을 구할 수 있다. 본 논문에서 개발한 정합방법을 이용하여 적용할 문제로서 첫째는 주상체가 상하동요, 수평동요하는 경우 계산이고 두번째로는 수면하에서 타원형실린더가 일정속도로 항진할 때 계산을 수행한 결과를 고차스펙트럴방법과 비교하였다.

• 응용통계연구
• /
• 제32권5호
• /
• pp.693-702
• /
• 2019

딥러닝은 대용량의 데이터의 분류 및 예측하는 방법으로 각광받고 있다. 데이터의 양이 많아지면서 신경망의 구조는 더 깊어 지고 있다. 이때 초기값이 지나치게 클 경우 층이 깊어 질수록 활성화 함수의 기울기가 매우 작아지는 포화(Saturation)현상이 발생한다. 이러한 포화현상은 가중치의 학습능력을 저하시키는 현상을 발생시키기 때문에 초기값의 중요성이 커지고 있다.이런 포화현상 문제를 해결하기 위해 Glorot과 Bengio (2010)과 He 등 (2015) 층과 층 사이에 데이터가 다양하게 흘러야 효율적인 신경망학습이 가능하고 주장했다. 데이터가 다양하게 흐르기 위해서는 각 층의 출력에 대한 분산과 입력에 대한 분산이 동일해야 한다고 제안했다. Glorot과 Bengio (2010)과 He 등 (2015)는 각 층별 활성화 값의 분산이 같다고 가정해 초기값을 설정하였다. 본 논문에서는 절단된 코쉬 분포와 절단된 정규분포를 활용하여 초기값을 설정하는 방안을 제안한다. 출력에 대한 분산과 입력에 대한 분산의 값을 동일하게 맞춰주고 그 값이 절단된 확률분포의 분산과 같게 적용함으로써 큰 초기값이 나오는 걸 제한하고 0에 가까운 값이 나오도록 분포를 조정하였다. 제안된 방법은 MNIST 데이터와 CIFAR-10 데이터를 DNN과 CNN 모델에 각각 적용하여 실험함으로써 기존의 초기값 설정방법보다 모델의 성능을 좋게 한다는 것을 보였다.