• Proceedings of the Korea Water Resources Association Conference
• /
• 2023.05a
• /
• pp.229-229
• /
• 2023

강화학습은 에이전트(agent)가 주어진 환경(environment)과의 상호작용을 통해서 상태(state)를 변화시켜가며 최대의 보상(reward)을 얻을 수 있도록 최적의 행동(action)을 학습하는 기계학습법을 의미한다. 최근 알파고와 같은 게임뿐만 아니라 자율주행 자동차, 로봇 제어 등 다양한 분야에서 널리 사용되고 있다. 상수도관망 분야의 경우에도 펌프 운영, 밸브 운영, 센서 최적 위치 선정 등 여러 문제에 적용되었으나, 설계에 강화학습을 적용한 연구는 없었다. 설계의 경우, 관망의 크기가 커짐에 따라 알고리즘의 탐색 공간의 크기가 증가하여 기존의 최적화 알고리즘을 이용하는 것에는 한계가 존재한다. 따라서 본 연구는 강화학습을 이용하여 상수도관망의 구성요소와 환경요인 간의 복잡한 상호작용을 고려하는 설계 방법론을 제안한다. 모델의 에이전트를 딥 강화학습(Deep Reinforcement Learning)으로 구성하여, 상태 및 행동 공간이 커 발생하는 고차원성 문제를 해결하였다. 또한, 해당 모델의 상태 및 보상으로 절점에서의 압력 및 수요량과 설계비용을 고려하여 적절한 수량과 수압의 용수 공급이 가능한 경제적인 관망을 설계하도록 하였다. 모델의 행동은 실제로 공학자가 설계하듯이 절점마다 하나씩 차례대로 다른 절점과의 연결 여부를 결정하는 것으로, 이를 통해 관망의 레이아웃(layout)과 관경을 결정한다. 본 연구에서 제안한 방법론을 규모가 큰 그리드 네트워크에 적용하여 모델을 검증하였으며, 고려해야 할 변수의 개수가 많음에도 불구하고 목적에 부합하는 관망을 설계할 수 있었다. 모델 학습과정 동안 에피소드의 평균 길이와 보상의 크기 등의 변화를 비교하여, 제안한 모델의 학습 능력을 평가 및 보완하였다. 향후 강화학습 모델을 통해 신뢰성(reliability) 또는 탄력성(resilience)과 같은 시스템의 성능까지 고려한 설계가 가능할 것으로 기대한다.

• Journal of Radiation Protection and Research
• /
• v.23 no.4
• /
• pp.243-249
• /
• 1998

The methodology for justification and optimization of the countermeasures related with contamination management of milk was designed based on the cost and benefit analysis. The application results were discussed for the deposition on August 15, when pasture is fully developed in Korean agricultural conditions. A dynamic food chain model DYNACON was used to estimate the time-dependent radioactivity of milk after the deposition. The considered countermeasures are (1) the ban of milk consumption (2) the substitution of clean fodder, which are effective in reducing the ingestion dose as well as simple and easy to carry out in the first year after the deposition. The total costs of the countermeasures were quantitatively estimated in terms of cost equivalent of doses and monetary costs. It is obvious that a fast reaction after the deposition is an important factor in cost effectiveness of the countermeasures. In most cases, the substitution of clean fodder was more effective countermeasure than the ban of consumption. A fast reaction after the deposition made longer justifiable/optimal duration of the countermeasure.

• Journal of Wetlands Research
• /
• v.26 no.1
• /
• pp.114-125
• /
• 2024

Wetlands in the living area are important ecological resources that are the basis for the daily life or farming activities of local residents, and have high ecological value such as ecosystem services and green infrastructure. This study was carried out to develop a functional evaluation methodology optimized for small-scale wetlands(RAMS). Based on on-site evaluation by experts, surveys and in-depth interviews, four functional items, including biodiversity, health, hydrophilic culture and ecology, water circulation, and carbon absorption, and 15 detailed indicators, and the evaluation grade for each detailed indicator, were developed on a 5-point scale. The evaluation methodology optimized for small-scale living areas wetlands (RAMS) proposed as a result of this study can be used as basic data for conservation and restoration and management of small-scale living areas wetlands at home and abroad.

• Journal of the Korea Society of Computer and Information
• /
• v.15 no.6
• /
• pp.147-152
• /
• 2010

The Internet has become a major channel of business marketing and sales, and there is a core competitive object between websites for a high position ranking in search engine results. There are various ways to maintain the high position ranking of website involving the development of componental coding or the expensive investment for the search engine optimization. The purpose of this paper is proposed to identify and rank the negative elements of website visibility to get rid of those elements when website designer designs the webpage. Website can be removed from indices of search engines when they are not satisfied for search engine optimization. The proposed experiments that are identified and ranked the negative elements of website visibility in this paper are based on the theories and experiments of existing website visibility models. The experimental analyses in this paper are scored and normalized based on methodologies of those models and 10 highest negative elements are ranked through the analyses. Therefore when website is designed, these highest negative elements should be avoided so website can not be removed in the indices of search engines.

• Journal of the Korea Society of Computer and Information
• /
• v.26 no.7
• /
• pp.1-7
• /
• 2021

In this paper, we propose a feature selection technique for multi-label classification. Many existing feature selection techniques have selected features by calculating the relation between features and labels such as a mutual information scale. However, since the mutual information measure requires a joint probability, it is difficult to calculate the joint probability from an actual premise feature set. Therefore, it has the disadvantage that only a few features can be calculated and only local optimization is possible. Away from this regional optimization problem, we propose a feature selection technique that constructs a low-rank space in the entire given feature space and selects features with sparsity. To this end, we designed a regression-based objective function using Nuclear norm, and proposed an algorithm of gradient descent method to solve the optimization problem of this objective function. Based on the results of multi-label classification experiments on four data and three multi-label classification performance, the proposed methodology showed better performance than the existing feature selection technique. In addition, it was showed by experimental results that the performance change is insensitive even to the parameter value change of the proposed objective function.

• Journal of Korean Society of Transportation
• /
• v.23 no.1
• /
• pp.83-92
• /
• 2005

This study if the first trial toward realizing a new methodology for vehicle re-identification based on heterogeneous sensor systems. A major interest of the author is how to effectively utilize information obtained from different sensors to derive accurate and reliable section travel times. The 'blade' sensor that is a newly developed sensor for capturing vehicle wheel information and the existing square loop sensor are employed to extract the inputs of the proposed vehicle re-identification algorithm. The fundamental idea of the algorithm developed in this study, which is so called 'anonumous vehicle re-identification,' it to match vehicle features obtained from both sensors. The results of the algorithm evaluation reveal that the proposed methodology could be successfully implemented in the field. The proposed methodology would be an invaluable tool for operating agencies in support of traffic monitoring systems and traveler information systems.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2012.04a
• /
• pp.348-351
• /
• 2012

본 논문에서는 많은 양의 외부 데이터를 활용할 수 있는 예제기반 초해상도(example-based super-resolution) 방법을 보다 효율적으로 할 수 있는 예제선정과 그를 위한 최적화기반의 방법론을 제안한다. 외부 데이터베이스 전체에 의존하는 것이 아니라, 예제선정을 위해 영상검색 (image retrieval) 방법을 도입하여 입력 영상과 관련 있는 영상을 외부 데이터베이스로부터 찾고 영상들로부터 초해상도 영상을 얻는다. 기존의 방법은 외부 데이터베이스를 모두 사용하기 때문에 입력영상에 불필요한 정보들이 복원되어 초해상도 결과의 질을 저하시킨다. 하지만 제안하는 방법에서는 영상검색을 통해 불필요한 정보들을 미리 제거하여 좋은 결과를 얻을 수 있다. 또한 외부 데이터베이스를 크기에 상관없이 검색된 몇 장의 영상을 사용하기 때문에 기존의 방법에 비해서 속도가 향상되었다.

• Journal of Korea Water Resources Association
• /
• v.41 no.12
• /
• pp.1211-1218
• /
• 2008

This study is for design of the detention system distributed in a watershed by the Multi-Objective Genetic Algorithms(MOGAs). A new model is developed to determine optimal size and location of detention. The developed model has two primary interfaced components such as a rainfall runoff model to simulate water surface elevation(or flowrate) and MOGAs to get the optimal solution. The objective functions used in this model depend on the peak flow and storage of detention. With various constraints such as structural limitations, capacities of storage and operational targets. The developed model is applied at Gwanyang basin within Anyang watershed. The simulation results show the maximum outlet reduction is occurred at detention facilities located in upper reach of watershed in the peak discharge rates. It is also reviewed the simultaneous construction of an off-line detention and an on-line detention. The methodologies obtained from this study will be used to control the flood discharges and to reduce flood damage in urbanized watershed.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.20 no.9
• /
• pp.133-141
• /
• 2019

Crew rostering in railway operations is usually done by arranging a crew diagram in accordance with working standards every month. This study was done to identify the problems related to the creation of crew rosters in railway operations and to suggest an optimum crew rostering method that can be applied in railway operations planning. To do this, the work standards of a railway company were identified, and a genetic algorithm was used to develop an optimal roster with equal working time while considering actual working patterns. The optimization process is composed of analysis of the input data, creation of work patterns, creation of a solution, and optimization steps. To verify the method, the roster derived from the proposed process was compared with a manually created roster. The results of the study could be used to reduce the deviation of business hours when generating a roster because the standard deviation of working time is the objective function.

• The Transactions of the Korea Information Processing Society
• /
• v.13 no.2
• /
• pp.26-33
• /
• 2024

There has been a recent surge of interest in the Digital Twin technology. The Digital Twin is technique for optimizing objects by simulating physical phenomena or objects through computer-based simulations. Currently, single Digital Twin is being developed to optimize processes limited to specific fields, but there is a limitation in that the independent Digital Twins cannot analyze the vast and complex processes of the real world. To overcome this, the concept of federated Digital Twin has been introduced. To date, the federated Digital Twin research has primarily focused on how to optimize macroscopic objects such as cities. However, by leveraging the interconnected nature of twins, existing implementations of the single Digital Twins can be modularized. In this study, we define the concepts and interrelationships of the single Digital Twin and the federated Digital Twin from a functional perspective related to process optimization and design a modularization technique for the single Digital Twin using the federated Digital Twin. Furthermore, this study aims to discuss the proposed methodology's efficacy by designing a model applying modularization to a real-world fabric manufacturing case.