• Journal of Power Electronics
• /
• 제9권5호
• /
• pp.757-771
• /
• 2009

This paper proposes a systematic approach to the modeling of the small-signal characteristics of three-phase bridge boost rectifiers under non-sinusoidal conditions. The main obstacle to the conventional synchronous d-q frame modeling approach is that it is unable to identify a steady-state under non-sinusoidal conditions. However, for most applications under non-sinusoidal conditions, the current loops of boost rectifiers are designed to have a bandwidth that is much higher than typical harmonics frequencies in order to achieve good current control for these harmonic components. Therefore a quasi-static method is applied to the proposed modeling approach. The converter small-signal characteristics developed from conventional synchronous frame modeling under different operating points are investigated and a worst case point is then located for the current loop design. Both qualitative and quantitative analyses are presented. It is observed that operating points influence the converter low frequency characteristics but hardly affect the dominant poles. The relationship between power stage parameters, system poles and zeroes is also presented which offers good support for the system design. Both the simulation and experimental results verified the analysis and proposed modeling approach. Finally, the practical case of a parallel active power filter is studied to present the modeling approach and the resultant regulator design procedure. The system performance further verifies the whole analysis.

• Communications for Statistical Applications and Methods
• /
• 제25권2호
• /
• pp.173-184
• /
• 2018

In medical or public health research, it is common to encounter clustered or longitudinal count data that exhibit excess zeros. For example, health care utilization data often have a multi-modal distribution with excess zeroes as well as a multilevel structure where patients are nested within physicians and hospitals. To analyze this type of data, zero-inflated count models with mixed effects have been developed where a count response variable is assumed to be distributed as a mixture of a Poisson or negative binomial and a distribution with a point mass of zeros that include random effects. However, no study has considered a situation where data are also censored due to the finite nature of the observation period or follow-up. In this paper, we present a weighted version of zero-inflated Poisson model with random effects accounting for variable individual follow-up times. We suggested two different types of weight function. The performance of the proposed model is evaluated and compared to a standard zero-inflated mixed model through simulation studies. This approach is then applied to Medicaid data analysis.

• Management Science and Financial Engineering
• /
• 제17권2호
• /
• pp.63-84
• /
• 2011

The ordinary multiple criteria decision making (MCDM) approach requires two types of input, alternative values and criterion weights, and employs two schemes of alternative prioritization, dominance and potential optimality. This paper allows for incomplete information on both types of input and gives rise to the dominance relationships and potential optimality of alternatives. Unlike the earlier studies, we emphasize that incomplete information frequently takes the form of strict inequalities, such as strict orders and strict bounds, rather than weak inequalities. Then the issues of rising importance include: (1) The standard mathematical programming approach to prioritize alternatives cannot be used directly, because the feasible region for the permissible decision parameters becomes an open set. (2) We show that the earlier methods replacing the strict inequalities with weak ones, by employing a small positive number or zeroes, which closes the feasible set, may cause a serious problem and yield unacceptable prioritization results. Therefore, we address these important issues and develop a useful and simple method, without selecting any small value for the strict preference information. Given strict information on both types of decision parameters, we first construct a nonlinear program, transform it into a linear programming equivalent, and finally solve it via a two-stage method. An application is also demonstrated herein.

• 응용통계연구
• /
• 제27권2호
• /
• pp.345-355
• /
• 2014

단위 영역의 결점수는 일반적으로 Poisson 분포를 가정한다. 이 Poisson 분포의 확장된 형태로 ZIP(zero-inflated Poisson) 분포를 고려할 수 있는데, 이 모형은 데이터에 0이 많이 관측되는 경우 잘 적합된다고 알려져 있다. 이 논문에서는 ZIP 분포를 따르는 공정을 관리하는 GLR(generalized likelihood ratio) 관리도 절차를 제안하고 있다. 또한 제안된 GLR 관리도의 효율을 기존에 제안된 CUSUM 관리도들과 비교하였다. 그 결과 제안된 GLR 관리도는 모수의 다양한 변화에 대해 효율이 좋거나 또는 효율이 크게 떨어지지 않았고, 특히 CUSUM 관리도에서 모수가 미리 설정한 방향과 다르게 변화했을 때 효율이 크게 나빠지는 문제를 해결할 수 있는 대안이라는 결론을 얻을 수 있었다.

• 전력전자학회논문지
• /
• 제21권6호
• /
• pp.457-464
• /
• 2016

This study proposes a bi-directional buck-boost converter controller design method for ESS using the MATLAB SISO tool. The conventional two-loop controller design is based on a continuous S-domain model that designs each controller independently. The demerit of the conventional method is that optimal performance is not easily achieved and extensive trials and errors are required because two-loop systems interact with one another. Using the MATLAB SISO tool based on the design method proposed in this work overcomes the disadvantages of the conventional method. In the proposed method, the SISO tool can select the location of the poles and zeroes of the open loop system, thereby facilitating the effective design of a high-performance controller. The design sequence is detailed systematically, and the performance of the method is verified with a computer simulation and 10 kW experimental system.

• ETRI Journal
• /
• 제37권5호
• /
• pp.945-950
• /
• 2015

A novel microstrip low-pass filter is presented to achieve an ultra-wide stopband with 11 harmonic suppression and very sharp skirt characteristics. The filter is composed of a modified U-shaped resonator (which creates two fully adjustable transmission zeroes), a T-shaped resonator (which determines a cut-off frequency), and four radial stubs (which provide a wider stopband). The operating mechanism of the filter is investigated based on a proposed equivalent-circuit model, and the role of each section of the proposed filter in creating null points is theoretically discussed in detail. The presented filter with 3 dB cut-off frequency ($f_c=2.35GHz$) has been fabricated and measured. Results show that a relative stopband bandwidth of 164% (referred to as a 22 dB suppression) is obtained while achieving a high figure-of-merit of 15,221.

• Structural Engineering and Mechanics
• /
• 제63권5호
• /
• pp.659-667
• /
• 2017

Rail support failure is inevitably subjected to track geometric deformations. Due to the randomness and evolvements of track irregularities, it is naturally a hard work to grasp the trajectories of dynamic responses of railway systems. This work studies the influence of rail fastener failure on dynamic behaviours of wheel/rail interactions and the railway tracks by jointly considering the effects of track random irregularities. The failure of rail fastener is simulated by setting the stiffness and damping of rail fasteners to be zeroes in the compiled vehicle-track coupled model. While track random irregularities will be transformed from the PSD functions using a developed probabilistic method. The novelty of this work lays on providing a method to completely reveal the possible responses of railway systems under jointly excitation of track random irregularities and rail support failure. The numerical results show that rail fastener failure has a great influence on both the wheel/rail interactions and the track vibrations if the number of rail fastener failure is over three. Besides, the full views of time-dependent amplitudes and probabilities of dynamic indices can be clearly presented against different failing status.

• 응용통계연구
• /
• 제34권3호
• /
• pp.329-345
• /
• 2021

추천 시스템은 고객의 데이터를 이용하여 개인 맞춤화된 상품을 추천한다. 추천 시스템은 협업 필터링, 콘텐츠 기반 필터링 그리고 이 두 가지를 합친 하이브리드 방법의 세 가지로 크게 나누어진다. 이 연구에서는 딥러닝 방법론에 기초한 오토인코더를 이용한 추천 시스템에 대한 소개와 그 모형들의 비교 연구를 진행한다. 오토인코더는 데이터 행렬에 0이 많은 경우의 문제를 효과적으로 다룰 수 있는 딥러닝 기반의 비지도학습 모형이다. 이 연구에서는 세 개의 실제 데이터를 이용하여 다섯 가지 종류의 오토인코더 기반 모형들을 비교한다. 처음의 세 개 모형은 협업 필터링에 속한 모형이고 나머지 두 개의 모형은 하이브리드 모형이다. 실제 데이터는 고객의 평점 데이터이고, 대부분의 평점이 없어서 희박성 비율이 높다는 특징이 있다.

• Acta Via Serica
• /
• 제7권1호
• /
• pp.105-130
• /
• 2022

This paper argues that the war in Syria is partly the result of a global Islamist wave that contributed to fuelling conflict across large regions of Asia and Africa. Of course, the war that has consumed Syria since 2011 most certainly has multiple interrelated causes and driving forces, and any attempt to isolate one or even two or three runs the risk of advancing an overly simplistic interpretation of history. This essay, therefore, does not aim to offer an appraisal of the multiple variables that contributed to the war in Syria. Instead, it zeroes in on how political Islam came to impact Syria and its people. In doing so, it demonstrates how competing varieties of political Islam represented leading causes of conflict. Indeed, different Islamist movements contributed to the outbreak of the war in 2011, fuelled the conflict for years on end, and to this day represent major obstacles to the achievement of sustainable peace. Four broad Islamist currents are especially relevant to the case of Syria: the Muslim Brotherhood; the Shia revivalist movement at the nexus of the alliance between Iran, Hezbollah, and Syria; Salafi jihadism and its volatile and fractious underworld of competing armed groups, from Al-Qaeda to the Islamic State; and Recep Tayyip Erdoğan's market-friendly Islamism, which induced Turkey to intervene in Syria's civil war.

• 응용통계연구
• /
• 제28권6호
• /
• pp.1171-1180
• /
• 2015

다차원척도법(multidimensional scaling)이란 개체간의 비유사성을 저차원 공간에 기하적으로 나타내려는 다변량 분석의 그래프적 기법이다. 일반적으로 다차원척도법은 계량형 다차원척도법과 비계량형 다차원척도법으로 분류할 수 있는데, 계량형 다차원척도법은 양적자료에 적용하게 된다. 그러나 이를 통해서는 개체들에 대한 군집화 정보만을 파악할 수 있으며, 개별 군집의 특징을 파악하기 위해서는 가상점(pseudo-points)을 활용한 변수들의 정보에 대한 추가적인 표현이 요구된다. 이러한 이유로 Gower (1992)는 연속형 변수에 대한 가상점들의 궤적을 표현함으로서 계량형 다차원척도법의 공간 상에 변수 정보를 나타내는 '대체법(replacement method)'을 제안한 바 있다. 그러나 이진수 자료는 계량형 다차원척도법을 적용할 수 있음에도 불구하고 대체법을 적용하면 가상점의 궤적을 표현할 수 없다. 따라서 본 연구에서는 이진수 자료에 대한 다차원척도법의 공간 상에 가상점을 이용하여 변수 정보를 표현하는 '분할법(partition method)'을 제안하려한다. 분할법은 0과 1의 비율을 모두 고려하여 가상점을 결정한다. 따라서 분할법에 의한 가상점을 활용한 계량형 다차원척도법을 통해 이진수 자료에서 변수와 개체간의 관계를 파악할 수 있게 해준다.