• Electronics and Telecommunications Trends
• /
• v.34 no.6
• /
• pp.87-99
• /
• 2019

Reinforcement learning is a type of machine learning paradigm that forces agents to repeat the observation-action-reward process to assess and predict the values of possible future action sequences. This allows the agents to incrementally reinforce the desired behavior for a given observation. Thanks to the recent advancements of deep learning, reinforcement learning has evolved into deep reinforcement learning that introduces promising results in various control and optimization domains, such as games, robotics, autonomous vehicles, computing, industrial control, and so on. In addition to this trend, a number of programming libraries have been developed for importing deep reinforcement learning into a variety of applications. In this article, we briefly review and summarize 10 representative deep reinforcement learning libraries and compare them from a development project perspective.

• Journal of the Korean Geotechnical Society
• /
• v.20 no.5
• /
• pp.109-116
• /
• 2004

The model tests are conducted to assess the behavior characteristics of geogrid reinforced soil walls according to different surcharge pressures and reinforcement spacings. The models are built in the box having dimension, 100cm tall, 140cm long, and 100cm wide. The reinforcement used is geogrid(tensile strength 2.26t/m). Decomposed ganite soil(SM) is used as a backfill material. The strain gauges and LVDTs are Installed to obtain the strain in the reinforcements and the displacements of the wall face. From the results, it can be concluded that the more the reinforcement tensile strength increases, the more the wall displacements and the geogrid strains decreases. The maximum wall displacements and geogrid strains of the model walls occur due to the uniform surcharge pressure at the 0.7H from the bottom of the wall. The horizontal displacements of the wall face nonlinearly increase with the increase of surcharge pressures, and this nonlinear behavior is significantly presented for larger surcharge due to the nonlinear tensile strength-strain relationship of the reinforcements.

• Journal of the Korean Society for Railway
• /
• v.10 no.2 s.39
• /
• pp.186-194
• /
• 2007

This study was performed to compare the load-carrying capacities of the reinforced concrete structure between the carbon fiber adhesion reinforcement method and the external post-tensioning method and further estimate the effective stress of the reinforced material by analyzing the experimental reinforcing effect of each method and the behavior resulting from each method. As a result, it was found out that the effective stress of the carbon fiber reinforcement according to the carbon fiber adhesion reinforcement method had an unexpected value, and also, bearing of the stress was found to be far from sharing thereof. That is to say, while the carbon fiber was bearing the whole stress to some limits, it got to be momentarily ruptured as soon as it went beyond such limits. On the other hand, the external post-tensioning method has the advantage of inducing an initial effective stress by introducing a strain, and thus, it was found that behavior or bearing of the stress was also found to be a solid behavior of the steel wire. This method was also found to be more efficient and excellent than the carbon fiber adhesion reinforcement method in the reinforcing effect or securing the effective stress. Accordingly, we were to discuss the effective stress as comparatively examined, focusing on deriving of the more enhanced reinforcing effect on the basis of the experiment to which the field characteristic is added.

• KCI Concrete Journal
• /
• v.14 no.3
• /
• pp.102-110
• /
• 2002

Results of an experimental study on the pullout behavior of the headed reinforcement are presented. A total of 48 pullout tests was performed to evaluate pullout strengths and load-displacement behaviors in pullout of the headed bars. The square steel heads had gross area of 4 $A_{b}$ and thickness of $d_{b}$ The test program consisted of three pullout test groups: Simple and Edge pullout tests using plain concrete slabs, comparison of pullout performances between the standard hooks and the headed reinforcement, and pullout tests of headed reinforcement using reinforced concrete columns. Test variables included concrete strengths ( $f_{c}$' = 27.1MPa, 39.1MPa), reinforcing bar diameters (D16~D29), embedment depths (6 $d_{b}$~12 $d_{b}$), edge conditions, column reinforcement, and single-vs.-multiple bar pullout. Test results revealed that the heads effectively provided the pullout resistances of the deformed bars in tension. The load-displacement behaviors were similar between the 90-degree hooks and the headed reinforcement. When a multiple number of headed bars installed with small head-to-head spacings was pulled out, reinforcement designed to run across the concrete failure surface in a direction parallel to the headed bars helped improve the pullout performances of the headed reinforcement.t.ement.t.

• Journal of the Korea institute for structural maintenance and inspection
• /
• v.10 no.3
• /
• pp.211-218
• /
• 2006

From the development of residential flat plate system, continuously bended shear reinforcement is developed for the prevention of punching shear. To know the punching shear capacity of developed shear reinforcement in slab-column joint, structural test is performed. The testing parameters are shear reinforcement types, such as no reinforcement, bended shear reinforcement, and head stud reinforcement. To verify the lateral capacity, cyclic load is applied under the constant vertical load condition. The results of tests are compared to as global displacement, slab-column joint strength. From the test results, the resisting capacity of developed shear reinforcement system has a good performance in the story drift ratio.

• Proceedings of the Korea Inteligent Information System Society Conference
• /
• 2001.01a
• /
• pp.65-70
• /
• 2001

We consider a hyper-redundant system that consists of many uniform units. The hyper-redundant system has many degrees of freedom and it can accomplish various tasks. Applysing the reinforcement learning to the hyper-redundant system is very attractive because it is possible to acquire various behaviors for various tasks automatically. In this paper we present a new reinforcement learning algorithm "Q-learning with propagation of motion". The algorithm is designed for the multi-agent systems that have strong connections. The proposed algorithm needs only one small Q-table even for a large scale system. So using the proposed algorithm, it is possible for the hyper-redundant system to learn the effective behavior. In this algorithm, only one leader agent learns the own behavior using its local information and the motion of the leader is propagated to another agents with time delay. The reward of the leader agent is given by using the whole system information. And the effective behavior of the leader is learned and the effective behavior of the system is acquired. We apply the proposed algorithm to a snake-like hyper-redundant robot. The necessary condition of the system to be Markov decision process is discussed. And the computer simulation of learning the locomotion is demonstrated. From the simulation results we find that the task of the locomotion of the robot to the desired point is learned and the winding motion is acquired. We can conclude that our proposed system and our analysis of the condition, that the system is Markov decision process, is valid.

• Steel and Composite Structures
• /
• v.52 no.6
• /
• pp.627-645
• /
• 2024

Punching shear is a brittle failure that occurs within the RC flat slabs where stresses are concentrated within small regions, resulting in a catastrophic and unfavorable progressive collapse. However, increasing the slab slenderness ratio is believed to significantly affect the slab's behavior by the induced strain values throughout the slab depth. This study examines the punching shear behavior of flat slabs by the nonlinear finite element analysis approach using ABAQUS software, where 72 models were investigated. The parametric study includes the effect of opening existence, opening-to-column ratio (O/C), temperature level, slenderness ratio (L/d), and flexural reinforcement rebar diameter. The behavior of the punching shear failure was fully examined under elevated temperatures which was not previously considered in detail along with the combined effect of the other sensitive parameters (opening size, slab slenderness, and reinforcement rebar size). It has been realized that increasing the slab slenderness has a major role in affecting the slab's structural behavior, besides the effect of the flexural reinforcement ratio. Reducing the slab's slenderness from 18.27 to 5.37 increased the cracking load by seven times for the slab without openings compared to nine times for the initial stiffness value. In addition, the toughness capacity is reduced up to 80% upon creating an opening, where the percentage is further increased by increasing the opening size by about an additional 10%. Finally, the ultimate deflection capacity of flat slabs with an opening is increased compared to the solid slab with the enhancement being increased for openings of larger size, larger depths, and higher exposure temperature.

• Advances in concrete construction
• /
• v.5 no.3
• /
• pp.257-270
• /
• 2017

An experimental investigation was carried out to study the strength and behavior of reinforced cement concrete (RCC) frames with ferrocement and fiber reinforced concrete infill panel. Seven numbers of $1/4^{th}$ scaled down model of one bay-three storey frames were tested under reverse cyclic loading. Ferrocement infilled frames and fiber reinforced concrete infilled frames with varying volume fraction of reinforcement in infill panels viz; 0.20%, 0.30%, and 0.40% were tested and compared with the bare frame. The experimental results indicate that the strength, stiffness and energy dissipation capacity of infilled frames were considerably improved when compared with the bare frame. In the case of infilled frames with equal volume fraction of reinforcement in infill panels, the strength and stiffness of frames with fiber reinforced concrete infill panels were slightly higher than those with ferrocement infill panels. Increase in volume fraction of reinforcement in the infill panels exhibited only marginal improvement in the strength and behavior of the infilled frames.

• Computers and Concrete
• /
• v.18 no.1
• /
• pp.139-154
• /
• 2016

Since the concrete strength around the reinforcement rebar affects the tension stiffening, the tension stiffening effect of ultra high performance concrete on the concrete members reinforced by steel rebar is examined by testing the specimens with circular cross section with the length 850 mm reinforced by a steel rebar at the center of a specimen's cross section in this research. Conducting a tensile test on the specimens, the cracking behavior is evaluated and a curve with an exponential descending branch is obtained to explain the post-cracking zone. In addition, this paper proposes an equation for this branch and parameters of equation is obtained based on the ratio of cover thickness to rebar diameter (c/d) and reinforcement percentage (${\rho}$).

• Structural Engineering and Mechanics
• /
• v.49 no.4
• /
• pp.449-461
• /
• 2014

The use of low-ductility welded wire fabric (WWF) as a main tensile reinforcement in concrete slabs compromises the ductility of concrete structures. Lower ductility in concrete structures can lead to brittle and catastrophic failure of the structures. This paper presents the experimental study carried out on eight simply supported one-way slabs to study the structural behavior of concrete slabs reinforced with low-ductility WWF and steel fibers. The different types of steel fibers used were crimped fiber, hooked-end fiber and twincone fiber. The experimental results show that the ductility behavior of the slab specimens with low-ductility reinforcement was significantly improved with the inclusion of $40kg/m^3$ of twincone fiber. Distribution of cracks was prominent in the slabs with twincone fiber, which also indicates the better distribution of internal forces in these slabs. However, the slab reinforced only with low-ductility reinforcement failed catastrophically with a single minor crack and without appreciable deflection.