• 대한전기학회논문지
• /
• 제39권7호
• /
• pp.729-739
• /
• 1990

This paper deals with the fault detection problem in uncertain linear/non-linear systems having both undermodelling and noise. A robust fault detection method is presented which accounts for the effects of noise, model mismatch and nonlinearities. The basic idea is to embed the unmodelled dynamics in a stochastic process and to use the nominal model with a predetermined fixed denominator. This allows the input /output relationship to be represented as a linear function of the system parameters and also facilitate the quatification of the effect of noise, model mismatch and linearization errors on parameter estimation by the Bayesian method. Comparisons are made via simulations with traditional fault detection methods which do not account for model mismatch or linearization errors. The new method suggested in this paper is shown to have a marked improvement over traditional methods on a number of simulations, which is a consequence of the fact that the new method explicitly for the effects of undermodelling and linearization errors.

• 대한산업공학회지
• /
• 제39권3호
• /
• pp.183-191
• /
• 2013

We address the power generation mix problem that considers not only nuclear and fossil fuels such as oil, coal and LNG but also renewable energy technologies. Unlike nuclear or other generation technologies, the expansion plan of renewable energy is highly uncertain because of its dependency on the government policy and uncertainty associated with technology improvements. To address this issue, we conduct a delphi survey and forecast the capacity of renewable energy. We further propose a stochastic mixed integer programming model that determines an optimal capacity expansion and the amount of power generation using each generation technology. Using the proposed model, we test eight generation mix scenarios and particularly evaluate how much the expansion of renewable energy contributes to the total costs for power generation in Korea. The evaluation results show that the use of renewable energy incurs additional costs.

• 상하수도학회지
• /
• 제11권3호
• /
• pp.87-95
• /
• 1997

A river water quality management model was made by Dynamic programming. This model optimizes the wastewater treatment cost of the application area, and computed water quality with it must meet the water quality standard. And this model takes into consideration tributary input, wastewater treatment plant effluent, withdrawls for several purposes. Modified Streeter-Phelps equation was used to calculate BOD and DO. Optimization problem was solved with particular exceedance probability flow, and the water quality of each point was calculated with the decided treatment efficiencies. At that time, the probability satisfying the water quality standard of constraints to the exceedance probability of the flow. The developed model was applied to the lower part of the Han-River. The reliability to meet the water quality standard is 70 % when 4 wastewater treatment plants of Seoul City are operated by activated sludge system at autumn of the year 2001. Treatment cost of this case is 121.288 billion won per year.

• 한국국방경영분석학회지
• /
• 제28권2호
• /
• pp.56-69
• /
• 2002

A technique of finding both probability density and distribution function for interkilling times is considered and demonstrated. An important result is that any arbitrary interfiring time random variables fit to this study, The interfiring renewal density function given a certain interfiring probability density function can be applied to obtain the corresponding interkilling renewal density function which helps us to estimate the expected number of killing events in a time period. The numerical inversion of Laplace transformation makes these possible and the results appear to be excellent. In case of ammunition supply is limited, an alternative way of getting the probability density function of time to the killing is investigated. The convolution technique may give us a means of settling for this new problem.

• 로봇학회논문지
• /
• 제17권1호
• /
• pp.40-47
• /
• 2022

A reward function suitable for a task is required to manipulate objects through reinforcement learning. However, it is difficult to design the reward function if the ample information of the objects cannot be obtained. In this study, a demonstration-based object manipulation algorithm called stochastic exploration guided by demonstration (SEGD) is proposed to solve the design problem of the reward function. SEGD is a reinforcement learning algorithm in which a sparse reward explorer (SRE) and an interpolated policy using demonstration (IPD) are added to soft actor-critic (SAC). SRE ensures the training of the critic of SAC by collecting prior data and IPD limits the exploration space by making SEGD's action similar to the expert's action. Through these two algorithms, the SEGD can learn only with the sparse reward of the task without designing the reward function. In order to verify the SEGD, experiments were conducted for three tasks. SEGD showed its effectiveness by showing success rates of more than 96.5% in these experiments.

• Earthquakes and Structures
• /
• 제22권4호
• /
• pp.421-430
• /
• 2022

This paper addresses the stochastic control problem of robots within the framework of parameter uncertainty and uncertain noise covariance. First of all, an open circle deterministic trajectory optimization issue is explained without knowing the unequivocal type of the dynamical framework. Then, a Linear Quadratic Gaussian (LQG) controller is intended for the ostensible trajectory-dependent linearized framework, to such an extent that robust hereditary NN robotic controller made out of the Kalman filter and the fuzzy controller is blended to ensure the asymptotic stability of the non-continuous controlled frameworks. Applicability and performance of the proposed algorithm shown through simulation results in the complex systems which are demonstrate the feasible to improve the performance by the proposed approach.

• Advances in Computational Design
• /
• 제7권1호
• /
• pp.37-56
• /
• 2022

Phase-type distributions are the distributions of the time to absorption in finite and absorbing Markov chains. They generalize, while at the same time, retain the tractability of the exponential distributions and their family. They are widely used as stochastic models from queuing theory, reliability, dependability, and forecasting, to computer networks, security, and computational design. The ability to fit phase-type distributions to intractable or empirical distributions is, therefore, highly desirable for many practical purposes. Many methods and tools currently exist for this fitting problem. In this paper, we present the results of our investigation on using orthogonal-distance fitting as a method for fitting phase-type distributions, together with a comparison to the currently existing fitting methods and tools.

• Journal of Information Processing Systems
• /
• 제20권2호
• /
• pp.226-238
• /
• 2024

Recently, multi-access edge computing (MEC) has emerged as a promising technology to alleviate the computing burden of vehicular terminals and efficiently facilitate vehicular applications. The vehicle can improve the quality of experience of applications by offloading their tasks to MEC servers. However, channel conditions are time-varying due to channel interference among vehicles, and path loss is time-varying due to the mobility of vehicles. The task arrival of vehicles is also stochastic. Therefore, it is difficult to determine an optimal offloading with resource allocation decision in the dynamic MEC system because offloading is affected by wireless data transmission. In this paper, we study computation offloading with resource allocation in the dynamic MEC system. The objective is to minimize power consumption and maximize throughput while meeting the delay constraints of tasks. Therefore, it allocates resources for local execution and transmission power for offloading. We define the problem as a Markov decision process, and propose an offloading method using deep reinforcement learning named deep deterministic policy gradient. Simulation shows that, compared with existing methods, the proposed method outperforms in terms of throughput and satisfaction of delay constraints.

• 한국ITS학회 논문지
• /
• 제3권1호
• /
• pp.31-44
• /
• 2004

최근 몇 년간 도시교통문제의 해결책으로 부각되어온 지능형교통체계(ITS : Intelligent Transport System)의 한 분야로 첨단여행자 정보체계(ATIS : Advanced Travellers Information System)는 자동차에 장착된 항법장치(CNS)를 통해 운전자에게 원하는 목적지까지 최적경로를 제공하거나 경로에 대한 통행시간 정보를 제공 또는 예측해 주는 시스템이다. 본 연구에서는 이러한 최적경로 제공이나 통행시간 예측에 있어 좀 더 효율적인 통행시간 예측모형을 개발하고자 하였다. 현재까지의 통행시간 예측은 운전자가 통행을 시작할 때의 교통상황에 대한 정보이기 때문에 운전 중에 달라지는 교통상황을 반영할 수 없어 이로 인해 운전자가 경험하는 통행시간과 큰 차이를 발생시킬 수 있다. 본 연구에서는 이러한 불합리적인 예측시스템을 개선시킬 수 있는 예측된(predicted) 통행시간 예측 모형을 개발하고자 하였다. 이를 위해 우선 통행시간 예측모형을 특정링크에 적용시켜 모형들의 예측치와 실제 통행시간을 비교하여 교통량 흐름 패턴에 따라 어느 모형이 적합한지, 또 예측시간이 달라짐에 따라 모형들의 적합도와 첨두와 비첨두시 예측시간 간격에 따라 예측치와 실측치의 오차율을 알아보았다, 이를 통해 선정된 확률과정 모형과 칼만 필터링 예측모형을 서울시의 4개축에 대해서 다시 적용해 보았다. 그 결과 단기통행시간 예측에 있어서는 칼만필터링모형이, 장기 통행시간 예측에 있어서는 확률과정 모형이 통행시간 예측에 있어 우수한 모형임을 밝혀냈다. 마지막으로 서울시 28개 교통축의 5분 후 통행시간 예측에 칼만필터링 모형을 이용하여 오차분석을 적용하여 보았다. 그 결과 칼만필터링 모형이 신뢰할 만한 오차율을 보였다.

• 한국수자원학회논문집
• /
• 제43권1호
• /
• pp.67-79
• /
• 2010

한국수자원공사는 낙동강 및 금강수계에 대해 실시간 물관리 시스템을 개발하여 수계 내 다목적 댐의 효율적인 운영을 도모하였다. 본 연구에서는 이를 확장하여 최적화 기법인 표본 추계학적 동적계획법을 사용하여 한강수계 다목적댐의 효율적인 월말 목표저수량을 산정하고 이를 통해 실시간 물관리시스템의 효율을 극대화하고자 수행되었다. 수계 내 댐중 저수용량과 용수공급 측면에서 중요도가 높은 소양강댐, 충주댐, 화천댐 등 3개 댐만을 대상으로 모형을 개발하였으며 저수량 상태변수의 민감도 분석 결과를 통해 저수량 상태변수 개수를 설정하였다. 본 연구를 통해 제시된 최적화모형을 통한 운영률과 한국수자원공사의 운영목표수위에 의한 저수지 운영방식과 비교를 실시한 결과 연평균 37.22 MCM의 수계 내 요구용수 부족량 감소효과와 더불어 발전량 측면에서도 연간 171 GWh가 증가하는 것을 확인하였다. 또한 이수기 저수지 운영계획 시스템을 실제 적용한 결과 2007년~2008년 이수기 동안 전력생산, 수질개선 등을 위한 추가적인 용수공급이 가능하다는 것을 확인하였다. 이러한 실제 적용사례를 통해 SSDP 모형과 이수기 저수지운영 시스템의 그 유용성을 타진할 수 있었으며 다른 수계로 확장을 실시한다면 보다 합리적인 저수지 운영계획이 가능할 것으로 기대한다.