• Journal of the Computational Structural Engineering Institute of Korea
• /
• v.27 no.6
• /
• pp.605-613
• /
• 2014

Epistemic uncertainty, the lack of knowledge, is often more important than aleatory uncertainty, variability, in estimating reliability of a system. While the probability theory is widely used for modeling aleatory uncertainty, there is no dominant approach to model epistemic uncertainty. Different approaches have been developed to handle epistemic uncertainties using various theories, such as probability theory, fuzzy sets, evidence theory and possibility theory. However, since these methods are developed from different statistics theories, it is difficult to interpret the result from one method to the other. The goal of this paper is to compare different methods in handling epistemic uncertainty in the view point of calculating the probability of failure. In particular, four different methods are compared; the probability method, the combined distribution method, interval analysis method, and the evidence theory. Characteristics of individual methods are compared in the view point of reliability analysis.

• Journal of Navigation and Port Research
• /
• v.32 no.3
• /
• pp.207-213
• /
• 2008

Ship-handling simulation system has been used for maritime traffic safety assessment for harbor and fairway. There exist various environmental conditions under which ship may navigate along a fairway or in harbor. Due to the time and budget limitations, however, ship-handling simulations are usually carried out for very limited number of environmental conditions. In this paper, statistical method for effective and systematic determination of real time simulation conditions is suggested and applied to the maritime traffic safety assessment problems. In the empirical study, the principal component analysis method and the concept of empirical cumulative distribution function are suggested to estimate synthetic navigational difficulty and to select simulation conditions which would impose high difficulty on shiphandling.

• The Korean Journal of Financial Management
• /
• v.20 no.1
• /
• pp.213-231
• /
• 2003

This paper uses the Efficient Method of Moments(EMM) of Gallant and Tauchen to estimate continuous-time stochastic volatility diffusion model for the Korean Composite Stock Price Index, sampled daily over $1995\sim2002$. The estimates display non-normality of stock index return, leptokurtic distribution, and stochastic volatility. Funker, this study suggests that two factor stochastic volatility model will be more desirable than one factor stochastic volatility model to estimate daily Korean stock return and also suggests that the stochastic volatility diffusions should allow for Poisson jumps of time-varying intensity.

• Proceedings of the Computational Structural Engineering Institute Conference
• /
• 2009.04a
• /
• pp.101-104
• /
• 2009

최적설계는 설계자가 요구하는 제한조건을 만족시키는 범위에서 목적함수가 최소가 되는 설계점을 찾는 방법이다. 그러나 기존의 최적설계는 불확실성의 영향을 고려하지 않아 최적해가 제한조건의 경계에 위치하고 이것은 모델링과정이나 가공 등으로 인한 오차에 대한 영향을 고려하지 않는 문제점이 있다. 신뢰성 기반 최적설계는 불확실성을 정량화하면서 신뢰도를 계산하는 신뢰도 해석과정과 최적설계과정을 포함한다. 일반적으로 신뢰성 해석은 크게 추출법, 급속 확률 적분법, 모멘트 기반 신뢰성해석이 있다. 가장 널리 사용되는 급속 확률 적분법 중 최대 손상 가능점(MPP) 방법은 많은 MPP점이 존재하는 경우 수치적 비용이 증가하는 문제점과 표준 정규분포 공간으로 변환하는 과정에서 제한조건의 비선형성을 증가시켜 큰 오차를 발생시키는 문제점이 있다. 본 논문에서는 RBDO를 수행하기에 앞서 선행되어야 할 신뢰성해석 방법으로 곱분해기법을 사용하였고 이로부터 민감도 정보를 유도하여 기울기 기반 최적화 알고리즘을 적용하였다.

• Proceedings of the KAIS Fall Conference
• /
• 2010.11a
• /
• pp.367-369
• /
• 2010

본 논문은 열차제어시스템 유지보수업무의 최적화를 위한 정량적 주기산출방법에 관한 것으로, 특히 전장품과 같이 유지보수업무로 주로 교체업무가 선택되는 경우, 운영비용을 최적화하기 위한 교체주기를 정량적으로 산출하기위한 방법을 제시하고, 장치의 고장분포 데이터를 통해, 철도신호장치의 교체주기를 할당한다. 제시한 방법은 유지보수장치의 운영이력을 분석하여, 장치의 고장데이터를 바탕으로 고장분포를 확률적으로 모델링한 뒤, 정확한 LCC데이터가 적용될 때 새롭게 도입되는 장비 및 시스템의 유지보수업무주기를 할당이 가능하고, 이를 통해여 시스템의 운영안전을 보장하면서, 운용비용을 최적화 할 수 있다.

• Korean Journal of Construction Engineering and Management
• /
• v.19 no.5
• /
• pp.3-9
• /
• 2018

Although Life Cycle Cost (LCC) must be evaluated by experts, sometimes it may not allow a sufficient time for even the experienced LCC expert to make rational decisions. Therefore, it often ends with relatively comparing the final numbers. We have broken down 110 technical proposals that are actually bade and accepted for large construction projects, and then have analyzed the uncertainty of Maintenance and Energy (M&E) cost during building life cycle, which turns out be the most volatile factor in uncertainty of LCC. Also we suggest "Value Engineering Index (VEI)" - the reduced M&E cost that is normalized by the reduced first cost. It is analyzed that the most uncertain factors of the M&E cost include repair and replacement term differing from each project, duplicated repair and replacement, non-standard repair items, and site-specific energy cost. Eventually we propose a VEI population with a mean of 1.38 and a standard deviation of 1.19, which is obtained by individually and exclusively applying the uncertain factors of the M&E cost to the 35 standard sample of technical proposals. The LCC evaluators may be able to use the VEI population as the benchmark to select the technical proposal with the most reasonable LCC among many others in two suggested manners; the one is to deterministically calculate the probability of single VEIs, and the other is to stochastically calculate the probability of the VEIs where uncertainty is quantified.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2021.05a
• /
• pp.607-610
• /
• 2021

When learning a model in supervised learning, input data and the label of the data are required. However, labeling is high cost task and if automated, there is no guarantee that the label will always be correct. In the case of supervised learning in such a noisy labels environment, the accuracy of the model increases at the initial stage of learning, but decrease significantly after a certain period of time. There are various methods to solve the noisy label problem. But in most cases, the probability predicted by the model is used as the pseudo label. So, we proposed a method to predict the true label more quickly by refining the probabilities predicted by the model. Result of experiments on the same environment and dataset, it was confirmed that the performance improved and converged faster. Through this, it can be applied to methods that use the probability distribution predicted by the model among existing studies. And it is possible to reduce the time required for learning because it can converge faster in the same environment.

• Journal of Korean Society of Transportation
• /
• v.19 no.5
• /
• pp.109-117
• /
• 2001

The median encroachment accident model proposed in this paper is the first step to develop cost-effective criteria about installing facilities preventing traffic accidents by median encroachment. This model consists of expected annual number of median encroachment on roadway and conditional probability to collide with vehicles on opposite lane after encroachment. Expected encroachment number is related to traffic volume and quote from a study of Hutchinson & Kennedy(1966). The probability of vehicle collision is composed of assumed headway distribution of opposite directional vehicles (negative exponential distribution), driving time of encroaching vehicle and Gap & Gap acceptance model. By using expected accident number yielded from the presented model, it will be able to calculate the benefit of reduced accident and to analyze the cost of installing facilities. Therefore this will help develop cost-effective criteria of what, to install in the median.

• Journal of Internet Computing and Services
• /
• v.21 no.2
• /
• pp.49-55
• /
• 2020

Information theoretic learning (ITL) methods based on random symbols (RS) use a set of random symbols generated according to a target distribution and are designed nonparametrically to minimize the cost function of the Euclidian distance between the target distribution and the input distribution. One drawback of the learning method is that it can not utilize the input power statistics by employing a constant stepsize for updating the algorithm. In this paper, it is revealed that firstly, information potential input (IPI) plays a role of input in the cost function-derivative related with information potential output (IPO) and secondly, input itself does in the derivative related with information potential error (IPE). Based on these observations, it is proposed to normalize the step-size with the statistically varying power of the two different inputs, IPI and input itself. The proposed algorithm in an communication environment of impulsive noise and multipath fading shows that the performance of mean squared error (MSE) is lower by 4dB, and convergence speed is 2 times faster than the conventional methods without step-size normalization.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2018.05a
• /
• pp.288-288
• /
• 2018

점착성 유사는 응집 현상을 겪는 유사로, 응집 현상(Flocculation Process)는 응집 과정(Aggregation Process)와 파괴 과정(Breakup Process)의 경쟁으로 이루어진다고 여겨진다. 응집 현상을 통해 점착성 유사는 물과 점착성을 띠는 작은 입자들의 덩어리인 플럭(Floc)을 형성하여 흐름 내에서는 대부분이 플럭의 형태로 이동한다. 점착성 유사의 응집 모형 중 하나인 플럭 성장모형(Floc Growth Model, FGM)은 상미분 방정식으로 시간에 따른 플럭의 크기를 계산하는 모형이다. 응집과 파괴의 평형 상태에서 평균 입경을 얻는다. 이러한 FGM은 낮은 수치 계산 비용으로 합리적인 계산 결과를 얻을 수 있으며, 유사 이동 모형 혹은 흐름 모형과의 결합이 수월한 장점을 가진다. 또한, 닫힌 계(Closed System)에서 질량이 보존되는 특징이 있다. 반면, 결정론적인 특성을 띠면서 특정 플럭 크기만을 계산하기 때문에 점착성 유사의 입도 분포에 대한 정보를 얻을 수 없다. 결정론적 특성을 띠는 FGM에 추계학적 방법을 적용함으로써 특정 확률 분포형을 가지는 플럭의 입도 분포를 얻을 수 있다. 본 연구에서는 기 개발된 추계학적 FGM과 유사 이동 모형의 결합을 통해 변화하는 유수동역학적 조건에서 플럭의 입도 분포를 산정하고자 한다. 이전의 많은 실험실 실험 결과들은 부유가 발생한 상태를 유지하면서 수행되는 것으로, 특정 난류 특성(난류 소산 매개변수)와 특정 유사 농도 조건에서의 입도 분포를 얻는다. 그러나 하구부 및 하천의 하류는 조류의 영향을 받는 구간으로, 점착성 유사의 특성을 분석하기 위해서는 변화하는 유수동역학적 특성에 관한 고려가 필수적이라 할 수 있다. 결합된 점착성 유사 입도 분포 모형은 플럭의 침강과 재부유를 고려할 수 있는 특징을 가지며, 실측자료와의 비교를 통해 입도 분포를 합리적으로 모의하는 것으로 나타난다. 본 연구에서 개발된 점착성 유사 입도 분포 모형은 나아가 비점착성 유사의 입도 분포 모형과의 결합을 통해 두 종류의 유사가 혼재하는 구간에서도 합리적인 입도분포와 유사의 이동을 모의할 수 있을 것으로 예측된다.