• Proceedings of the Korean Information Science Society Conference
• /
• 2003.10a
• /
• pp.22-24
• /
• 2003

작은 데이터에서 베이지안망 분류기(Bayesian network classifier, BNC)를 학습할 때, 과대적합(overfitting)으로 인한 일반화 성능의 저하가 초래된다 이런 경우, 베이지안 모델 평균화(Bayesian model averaging, BMA)는 모델 자체에 대한 불확실성을 분석 과정에서 고려함으로써, 성능 저하를 피할 수 있는 수단을 제공한다. 본 논문에서는 BNC의 BMA의 작은 데이터에 대한 성능을 평가 및 분석한다. 특히, 노드의 순서에 대한 평균화의 효과가 연구된다. 인공데이터에 대한 실험 결과, 노드의 순서가 BNC의 BMA의 분류 성능에 미치는 영향은 지대하며, 이는 데이터의 크기가 극히 작은 경우의 성능 저하에 직접적인 원인이 된다.

• Journal of Power Electronics
• /
• v.13 no.6
• /
• pp.1008-1015
• /
• 2013

By using fractional calculus and the circuit-averaging technique, the modeling and analysis of a Buck converter with fractional order inductor and fractional order capacitor in discontinuous conduction mode (DCM) operations is investigated in this study. The equivalent averaged circuit model of the fractional order Buck converter in DCM operations is established. DC analysis is conducted by using the derived DC equivalent circuit model. The transfer functions from the input voltage to the output voltage, the duty cycle to the output voltage, the input impedance, and the output impedance of the fractional order Buck converter in DCM operations are derived from the corresponding AC-equivalent circuit model. Results show that the DC equilibrium point, voltage ratio, and all derived transfer functions of the fractional order Buck converter in DCM operations are affected by the inductor order and/or capacitor order. The fractional order inductor and fractional order capacitor are designed, and PSIM simulations are performed to confirm the correctness of the derivations and theoretical analysis.

• Journal of Korean Society for Atmospheric Environment
• /
• v.24 no.E2
• /
• pp.63-73
• /
• 2008

In this study, neural network models (NNMs) were examined as alternatives to dispersion models in predicting the short-term $SO_2$ concentrations in a coastal area because the performances of dispersion models in coastal areas have been found to be unsatisfactory. The NNMs were constructed for various combinations of averaging time and prediction time in advance by using the historical data of meteorological parameters and $SO_2$ concentrations in 2002 in the coastal area of Boryeung, Korea. The NNMs were able to make much more accurate predictions of 1 hr $SO_2$ concentrations at ground level in the morning in coastal area than the atmospheric dispersion models such as fumigation models, ADMS3 and ISCST3 for identical conditions of atmospheric stability, area, and weather. Even when predictions of 24-h $SO_2$ concentrations were made 24 hours in advance, the predictions and measurements were in good accordance(correlation coefficient=0.65 for n=216). This accordance level could be improved by appropriate expansion of training parameters. Thus it may be concluded that the NNMs can be successfully used to predict short-term ground level concentrations averaged over time less than 24 hours even in complex terrain. The prediction performance of ANN models tends to improve as the prediction lagging time approaches the concentration averaging time, but to become worse as the lagging time departs from the averaging time.

• Journal of the Korean earth science society
• /
• v.32 no.4
• /
• pp.360-372
• /
• 2011

In order to clarify the advection and dispersion characteristics of volcanic tephra to be emitted from the Mt. Baekdu, several numerical experiments were carried out using three-dimensional atmospheric dynamic model, Weather and Research Forecast (WRF) and Laglangian particles dispersion model FLEXPART. Four different temporally averaged meteorological values including wind speed and direction were used, and their averaged intervals of meteorological values are 1 month, 10 days, and 3days, respectively. Real time simulation without temporal averaging is also established in this study. As averaging time of meteorological elements is longer, wind along the principle direction is stronger. On the other hands, the tangential direction wind tends to be clearer when the time become shorten. Similar tendency was shown in the distribution of volcanic tephra because the dispersion of particles floating in the atmosphere is strongly associated with wind pattern. Wind transporting the volcanic tephra is divided clearly into upper and lower region and almost ash arriving the Korean Peninsula is released under 2 km high above the ground. Since setting up the temporal averaging of meteorological values is one of the critical factors to determine the density of tephra in the air and their surface deposition, reasonable time for averaging meteorological values should be established before the numerical dispersion assessment of volcanic tephra.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.61 no.12
• /
• pp.1920-1925
• /
• 2012

In this paper, we propose a novel method to improve the performance of the improved minima controlled recursive averaging (IMCRA) in non-stationary noisy environment. The conventional IMCRA algorithm efficiently estimate the noise power by averaging past spectral power values based on a smoothing parameter that is adjusted by the signal presence probability in frequency subbands. Since the minimum of smoothing parameter is defined as 0.85, it is difficult to obtain the robust estimates of the noise power in non-stationary noisy environments that is rapidly changed the spectral characteristics such as babble noise. For this reason, we proposed the modified IMCRA, which adaptively estimate and updata the noise power according to the noise type classified by the Gaussian mixture model (GMM). The performances of the proposed method are evaluated by perceptual evaluation of speech quality (PESQ) and composite measure under various environments and better results compared with the conventional method are obtained.

• Smart Structures and Systems
• /
• v.5 no.1
• /
• pp.69-79
• /
• 2009

A non-clipped semi-active stochastic optimal control strategy for nonlinear structural systems with MR dampers is developed based on the stochastic averaging method and stochastic dynamical programming principle. A nonlinear stochastic control structure is first modeled as a semi-actively controlled, stochastically excited and dissipated Hamiltonian system. The control force of an MR damper is separated into passive and semi-active parts. The passive control force components, coupled in structural mode space, are incorporated in the drift coefficients by directly using the stochastic averaging method. Then the stochastic dynamical programming principle is applied to establish a dynamical programming equation, from which the semi-active optimal control law is determined and implementable by MR dampers without clipping in terms of the Bingham model. Under the condition on the control performance function given in section 3, the expressions of nonlinear and linear non-clipped semi-active optimal control force components are obtained as well as the non-clipped semi-active LQG control force, and thus the value function and semi-active nonlinear optimal control force are actually existent according to the developed strategy. An example of the controlled stochastic hysteretic column is given to illustrate the application and effectiveness of the developed semi-active optimal control strategy.

• Journal of Korean Society for Atmospheric Environment
• /
• v.5 no.2
• /
• pp.84-96
• /
• 1989

To evaluate the short term air pollution dispersion model, the diffusion experiment was conducted on the flat terrain near Chuncheon. Sulfur hexafluoride $(SF_6)$ gas was used to determine the horizontal spread of plume $(\sigmay)$ for calculated by CRSTER model. Results show that CRSTER model underestimates $\sigma$y because averaging time adjustment is not applied to calculate the $\sigma$y. The scheme that can estimate the atmospheric stability more accurate than Turner method, was presented.

• Journal of Power Electronics
• /
• v.18 no.5
• /
• pp.1434-1447
• /
• 2018

The control design of inductive power transfer (IPT) systems has attracted a lot of attention in the field of wireless power transmission. Due to the high-order resonant networks and multiple loads in IPT systems, a simplified model of an IPT system is preferred for analysis and control design, and a controller with strong robustness is required. Hence, an active disturbance rejection control (ADRC) for IPT systems is proposed in this paper. To realize the employment of ADRC, firstly a small-signal model of an LC series-compensative IPT system is derived based on generalized state-space averaging (GSSA), then the ADRC is implemented in the designed IPT system. The ADRC not only provides superior robustness to unknown internal and external disturbances, but also requires few knowledge of the IPT system. Due to the convenient realization of ADRC, the designed IPT system retains its simple structure without any additional circuits. Finally, a frequency domain analysis and experimental results have validated the effectiveness of the employed ADRC, especially its robustness in the presence of frequency drifts and other common disturbances.

• Journal of Electrical Engineering and Technology
• /
• v.13 no.4
• /
• pp.1566-1577
• /
• 2018

This paper presents a stability analysis of AC-DC power system feeding a speed controlled DC motor in which this load behaves as a constant power load (CPL). A CPL can significantly degrade power system stability margin. Hence, the stability analysis is very important. The DQ and generalized state-space averaging methods are used to derive the mathematical model suitable for stability issues. The paper analyzes the stability of power systems for both speed control natural frequency and DC-link parameter variations and takes into account controlled speed motor dynamics. However, accurate DC-link filter and DC motor parameters are very important for the stability study of practical systems. According to the measurement errors and a large variation in a DC-link capacitor value, the system identification is needed to provide the accurate parameters. Therefore, the paper also presents the identification of system parameters using the adaptive Tabu search technique. The stability margins can be then predicted via the eigenvalue theorem with the resulting dynamic model. The intensive time-domain simulations and experimental results are used to support the theoretical results.

• Communications for Statistical Applications and Methods
• /
• v.14 no.3
• /
• pp.687-698
• /
• 2007

A normal mixture model with which dependence between classes is incorporated is proposed in order to detect differentially expressed genes. Gene clustering approaches suffer from the high dimensional column of microarray expression data matrix which leads to the over-fit problem. Various methods are proposed to solve the problem. In this paper, use of simple averaging data within each class is proposed to overcome the various problems due to high dimensionality when the normal mixture model is fitted. Some experiments through simulated data set and real data set show its availability in actuality.