• Journal of Information Processing Systems
• /
• 제16권5호
• /
• pp.1223-1230
• /
• 2020

Recently, extensive studies have been conducted to apply deep learning to reinforcement learning to solve the state-space problem. If the state-space problem was solved, reinforcement learning would become applicable in various fields. For example, users can utilize dance-tutorial systems to learn how to dance by watching and imitating a virtual instructor. The instructor can perform the optimal dance to the music, to which reinforcement learning is applied. In this study, we propose a method of reinforcement learning in which the action space is dynamically adjusted. Because actions that are not performed or are unlikely to be optimal are not learned, and the state space is not allocated, the learning time can be shortened, and the state space can be reduced. In an experiment, the proposed method shows results similar to those of traditional Q-learning even when the state space of the proposed method is reduced to approximately 0.33% of that of Q-learning. Consequently, the proposed method reduces the cost and time required for learning. Traditional Q-learning requires 6 million state spaces for learning 100,000 times. In contrast, the proposed method requires only 20,000 state spaces. A higher winning rate can be achieved in a shorter period of time by retrieving 20,000 state spaces instead of 6 million.

• 한국지능시스템학회논문지
• /
• 제15권4호
• /
• pp.395-399
• /
• 2005

본 논문에서는 특정 주제에 관한 웹 문서들을 더욱 빠르고 정확하게 탐색하기 위하여 강화학습을 이용한 HIGH-Q 학습 알고리즘을 제안한다. 강화학습의 목적은 환경으로부터 주어지는 보상(reward)을 최대화하는 것이며 강화학습 에이전트는 외부에 존재하는 환경과 시행착오를 통하여 상호작용하면서 학습한다. 제안한 알고리즘이 주어진 환경에서 빠르고 효율적임을 보이기 위하여 넓이 우선 탐색과 비교하는 실험을 수행하고 이를 평가하였다. 실험한 결과로부터 우리는 미래의 할인된 보상을 이용하는 강화학습 방법이 정답을 찾기 위한 탐색 페이지의 수를 줄여줌으로써 더욱 정확하고 빠른 검색을 수행할 수 있음을 알 수 있었다.

• 한국지반공학회:학술대회논문집
• /
• 한국지반공학회 2008년도 추계 학술발표회
• /
• pp.1232-1238
• /
• 2008

From the result of analysis using finite element method for the Pile Slab reinforcement length through embankment of height, soft ground and the change of cohesion following results were acquired. 1. The higher embankment of height is, the deeper depth of soft ground is, the smaller cohesion is, Pile Slab reinforcement length increased almost straight. 2. The reinforcement length is controlled by the depth of soft ground, cohesion, embankment of height and the like. Among these, cohesion of soft ground is affected the most. 3. The reinforcement length of Pile Slab is determined using by calculated formula.

• Computers and Concrete
• /
• 제11권5호
• /
• pp.383-397
• /
• 2013

Two-dimensional tensile stresses are occurring at the back of the anchorage of the tendons of prestressed concrete bridges. A new method named "tensile stress region" for the design of the reinforcement is presented in this paper. The basic idea of this approach is the division of an anchor block into several slices, which are described by the tensile stress region. The orthogonal reinforcing wire mesh can be designed in each slice to resist the tensile stresses. Additionally the sum of the depth of every slice defined by the tensile stress region is used to control the required length of the longitudinal reinforcement bars. An example for the reinforcement design of an anchorage block of an external prestressed concrete bridge is analyzed by means of the new presented method and a finite element model is established to compare the results. Furthermore the influence of the transverse and vertical prestressing on the ordinary reinforcement design is taken into account. The results show that the amount of reinforcement bars at the anchorage block is influenced by the layout of the transverse and the vertical prestressing tendons. Using the "tensile stress region" method, the ordinary reinforcement bars can be designed more precisely compared to the design codes, and arranged according to the stress state in every slice.

• 한국농촌건축학회논문집
• /
• 제24권2호
• /
• pp.13-20
• /
• 2022

In this study, a general low-rise building was selected to compare the new shear wall reinforcement method, which is a general method among the existing reinforcement methods, and the reinforcement method using sinusoidal corrugated web reinforcement. And it was confirmed that the following effects can be expected. Sinusoidal corrugated web members can be carried out in a short period of time as it does not require the removal of the masonry filling wall, the reinforcement of reinforcing bars, and the curing period of the concrete. It is effective in preventing damage that may occur when masonry filling wall is overturned in the out-of-plane direction, and the burden of the foundation is also reduced, and thus the construction period and cost required for reinforcement can greatly be reduced. By adjusting the number of sinusoidal corrugated web member, details of joints, and reinforcement positions, the flow of load can be induced to have an advantageous effect on the building. It can be considered as the most suitable reinforcement plan in terms of life safety. Unlike the shear wall that fills between the columns, the sinusoidal corrugated web members, which has a width of 1.5m, can install openings between two columns depending on the purpose of use, and can be expected to have a great effect in terms of usability due to its free installation location. As mentioned above, the seismic reinforcement using a sinusoidal corrugated web members, can expect great effect compared to conventional reinforcement methods in terms of usability, economic feasibility, and stability.

• 한국구조물진단유지관리공학회 논문집
• /
• 제21권2호
• /
• pp.60-68
• /
• 2017

본 연구에서는 지내력이 기초판에 미치는 영향을 충분히 고려할 수 있도록 현장여건과 동일한 옥외의 지반에서 실험할 수 있는 시스템을 구축하였으며, 대상 실험체는 경제성 및 시공성 향상을 위하여 강판을 "ㄷ"자형으로 절곡하여 단면 2차모멘트를 극대화 하고 현장조립이 가능하도록 제안 하였다. 대상 실험체는 무보강 실험체 1개, 강판 두께를 동일하게 하여 보강 길이를 달리한 실험체 3개, 강판 두께를 달리하고 위험단면 부근에 스티프너 보강한 실험체 2개 총 6개의 실험체를 대상으로 비교 검토 한다. 실험 결과 스티프너 보강에 의한 효과는 없는 것으로 나타났으며, 전단보강재의 보강길이는 확장된 위험단면에서 전단력을 지내력으로 나타낸 값과 위험단면에서 보강재가 받을 수 있는 전단내력을 지내력으로 환산여하여 두 선의 교차점을 유효보강 길이로 산정하는 강판두께별 유효보강길이 산정방법을 제안하였다.

• 한국철도학회:학술대회논문집
• /
• 한국철도학회 2011년도 정기총회 및 추계학술대회 논문집
• /
• pp.1577-1582
• /
• 2011

Recently rapid construction of bridge is a main interest in construction. A research on rapid construction of pier coping is urgently needed because pier, which is a bridge understructure, directly affect lane reduction and increase of social cost. Precast assembly method and pre-assembly method are the main subjects of rapid construction. But these researches have focused not on reduction of reinforcement amount, but on modifying production method of coping. Reinforcement amount of design specification is as much as that of coping under constructing. So different approach is needed for reduction of reinforcement amount. In this paper, design of pier coping using strut-tie model was proposed for reduction of reinforcement amount and improvement of constructability. Railway bridge pier coping under constructing was analyzed using a finite element method and designed using strut-tie model.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• 제1권1호
• /
• pp.125-130
• /
• 2001

In this paper, we propose a new type of evolution strategies combined with reinforcement learning. We use the variances of fitness occurred by mutation to make the reinforcement signals which estimate and control the step length of mutation. With this proposed method, the convergence rate is improved. Also, we use cauchy distributed mutation to increase global convergence faculty. Cauchy distributed mutation is more likely to escape from a local minimum or move away from a plateau. After an outline of the history of evolution strategies, it is explained how evolution strategies can be combined with the reinforcement learning, named reinforcement evolution strategies. The performance of proposed method will be estimated by comparison with conventional evolution strategies on several test problems.

• 한국정보처리학회논문지
• /
• 제6권3호
• /
• pp.633-640
• /
• 1999

Reinforcement learning performs learning through interacting with trial-and-error in dynamic environment. Therefore, in dynamic environment, reinforcement learning method like Q-learning and TD(Temporal Difference)-learning are faster in learning than the conventional stochastic learning method. However, because many of the proposed reinforcement learning algorithms are given the reinforcement value only when the learning agent has reached its goal state, most of the reinforcement algorithms converge to the optimal solution too slowly. In this paper, we present COMREL(COMpressed REinforcement Learning) algorithm for finding the shortest path fast in a maze environment, select the candidate states that can guide the shortest path in compressed maze environment, and learn only the candidate states to find the shortest path. After comparing COMREL algorithm with the already existing Q-learning and Priortized Sweeping algorithm, we could see that the learning time shortened very much.

• 방사성폐기물학회지
• /
• 제21권3호
• /
• pp.329-345
• /
• 2023

This paper described a method for analyzing the structural performance of a metal container used for disposing radioactive waste generated during the decommissioning of a nuclear power plant, and numerical analysis results of a method for reinforcing the container. The containers to be analyzed were those that can be used in near-surface and landfill disposal facilities scheduled to be operated at the Gyeongju radioactive waste disposal facility. Structural reinforcement of the container was performed by lattice reinforcement, column reinforcement, and bottom plate reinforcement. Accordingly, a total of 14 reinforcement cases were modeled. The external force causing damage to the container was set equivalent to the impact of a 9-m fall, accounting for the height of the vault at the near-surface disposal facility. The reinforcement methods with a high contribution to the structural performance of the container were concluded to be lattice and column reinforcements.