• Journal of the Korea Institute of Military Science and Technology
• /
• v.27 no.4
• /
• pp.474-484
• /
• 2024

Reinforcement learning, which are also studied in the field of defense, face the problem of sample efficiency, which requires a large amount of data to train. Transfer learning has been introduced to address this problem, but its effectiveness is sometimes marginal because the model does not effectively leverage prior knowledge. In this study, we propose a stochastic initial state randomization(SISR) method to enable robust knowledge transfer that promote generalized and sufficient knowledge transfer. We developed a simulation environment involving a cooperative robot transportation task. Experimental results show that successful tasks are achieved when SISR is applied, while tasks fail when SISR is not applied. We also analyzed how the amount of state information collected by the agents changes with the application of SISR.

• International Journal of Control, Automation, and Systems
• /
• v.1 no.1
• /
• pp.142-148
• /
• 2003

Reinforcement learning is considered as an important tool for robotic learning in unknown/uncertain environments. In this paper, we propose an evaluation function expressed in a vector form to realize multi-dimensional reinforcement learning. The novel feature of the proposed method is that learning one behavior induces parallel learning of other behaviors though the objectives of each behavior are different. In brief, all behaviors watch other behaviors from a critical point of view. Therefore, in the proposed method, there is cross-criticism and parallel learning that make the multi-dimensional learning process more efficient. By ap-plying the proposed learning method, we carried out multi-dimensional evaluation (reward) and multi-dimensional learning simultaneously in one trial. A special neural network (Q-net), in which the weights and the output are represented by vectors, is proposed to realize a critic net-work for Q-learning. The proposed learning method is applied for behavior planning of mobile robots.

• Journal of KIISE:Computing Practices and Letters
• /
• v.14 no.5
• /
• pp.502-506
• /
• 2008

The recently proposed "Potential-based" reinforcement learning (RL) method made it possible to combine multiple learnings and expert advices as supervised knowledge within an RL framework. The effectiveness of the approach has been established by a theoretical convergence guarantee to an optimal policy. In this paper, the potential-based RL method is applied to a dynamic channel assignment (DCA) problem in a cellular networks. It is empirically shown that the potential-based RL assigns channels more efficiently than fixed channel assignment, Maxavail, and Q-learning-based DCA, and it converges to an optimal policy more rapidly than other RL algorithms, SARSA(0) and PRQ-learning.

• Journal of Korean Society of Disaster and Security
• /
• v.13 no.1
• /
• pp.13-23
• /
• 2020

In 2005, Urban railway seismic design was introduced in Rep. Of Korea, and many studies on seismic performance evaluation and reinforcement methods were conducted. In accordance with the Enforcement Decree of the Earthquake Disaster Countermeasures Act issued in March 2009, during April 2010 to October 2013, some of local governments established detailed evaluation and reinforcement measures for seismic performance of the urban railway underground structure. Afterwards, the seismic performance reinforcement of the existing urban railway structures was conducted for the sections that a long period of used until the end of 2018, and the reinforcement works are carried out by various methods using the previously studied methods. However, various reinforcing materials and construction methods using have been studied, but the classification research on the construction methods currently applied to reinforcement construction of urban railways is insufficient. The purpose of study is to analysis the cases currently applied to seismic reinforcement construction and to show the characteristics of each construction method, the reasons for its application and problems.

• Journal of the Korean Geotechnical Society
• /
• v.21 no.8
• /
• pp.107-116
• /
• 2005

Generally, when the stability of the geosynthetic reinforced soil slopes is analyzed, the required tensile forces of each reinforcement layers are calculated from total reinforcement forces which are necessary to retain the equilibrium state of slip mass in which the slip surfaces are assumed to be a linear or bilinear. It is assumed that the reinforcement forces are increased or constant with depth. However, according to the instrumented field data and laboratory model test results, the maximum tensile strain of reinforcement in a reinforced soil slope is developed in a certain elevation, not a bottom of the slope. In the concept of reinforced soil, postulated failure surfaces are the traces of the position in which the reinforcement tensile forces are maximum in the layer, and the reinforcement tensile forces are related to the stress state on the postulated failure surface. Therefore, in this study, based on the distribution of normal stress on the slip surface, a new method for the evaluation of required tensile forces is suggested and a number of the instrumented field data are analyzed by newly suggested method. As a result, it is shown that the newly suggested method produces relatively accurate reinforcement tension forces.

• Journal of Korean Tunnelling and Underground Space Association
• /
• v.21 no.4
• /
• pp.455-478
• /
• 2019

In this paper, we conducted experimental investigation on the field applicability through the verification of reinforcement effect of the steel pipe reinforcement grouting using high strength steel pipe. SGT275 (formerly known as STK400) steel pipe is generally applied to the traditional steel pipe reinforcement grouting method. However, the analysis of tunnel collapse cases applying the steel pipe reinforcement grouting shows that there are cases where the excessive bending and breakage of steel pipe occur. One of the reasons causing these collapses is the lack of steel pipe stiffness responding to the loosening load of tunnels caused by excavation. The strength of steel pipe has increased due to the recent development of high strength steel pipe (SGT550). However, since research on the reinforcement method considering strength increase is insufficient, there is a need for research on this. Therefore, in this study, we conducted experiments on the tensile and bending strength based on various conditions between high strength steel pipe, and carried out basic research on effective field application depending on the strength difference of steel pipe through the conventional design method. In particular, we verified the reinforcement effect of high strength steel pipe through the measurement results of deformed shape and stress of steel pipe arising from excavation after constructing high strength steel pipe and general steel pipe at actual sites. The research results show that high strength steel pipe has excellent bending strength and the reinforcement effect of reinforced grouting. Further, it is expected that high strength steel pipe will have an arching effect thanks to strength increase.

• International Journal of Internet, Broadcasting and Communication
• /
• v.12 no.1
• /
• pp.144-150
• /
• 2020

Reinforcement learning is a technology that can present successful and creative solutions in many areas. This reinforcement learning technology was used to deploy containers from cloud servers to fog servers to help them learn the maximization of rewards due to reduced traffic. Leveraging reinforcement learning is aimed at predicting traffic in the network and optimizing traffic-based fog computing network environment for cloud, fog and clients. The reinforcement learning system collects network traffic data from the fog server and IoT. Reinforcement learning neural networks, which use collected traffic data as input values, can consist of Long Short-Term Memory (LSTM) neural networks in network environments that support fog computing, to learn time series data and to predict optimized traffic. Description of the input and output values of the traffic-based reinforcement learning LSTM neural network, the composition of the node, the activation function and error function of the hidden layer, the overfitting method, and the optimization algorithm.

• Proceedings of the Korean Institute of Building Construction Conference
• /
• 2019.05a
• /
• pp.85-86
• /
• 2019

Recently, the educational facilities were 45% among total damaged facilities from the Po-Hang earthquake. Consequently, the seismic reinforcement of existing educational facilities were ended up attracting people's interesting. However, research is insufficient to consider that how far technology has been developed. Therefore, the purpose of this study is to investigate the level of domestic guideline research by comparing to foreign guidelines in regard to seismic reinforcement methods applicable to existing educational facilities. To achieve the objective, first, the current status of seismic reinforcement projects in domestic educational facilities was analyzed. Second, the domestic and foreign seismic reinforcement methods guidelines for structural, non-structural was compared. Third, the improvement directions for future guidelines were suggested. As a results, the improvement directions of domestic seismic reinforcement methods guidelines were proposed. First, the structural seismic reinforcement strategy needs to be segmented. Second, it is necessary to analyze about non-structural guidelines based on additional cases. In the future, this study can be used as a basic material for developing seismic reinforcement methods guidelines in domestic existing educational facilities.

• Geomechanics and Engineering
• /
• v.11 no.6
• /
• pp.787-803
• /
• 2016

DDARF (Discontinuous Deformation Analysis for Rock Failure) is a numerical algorithm for simulating jointed rock masses' discontinuous deformation. While its reinforcement simulation is only limited to end-anchorage bolt, which is assumed to be a linear spring simply. Here, several new reinforcement modes in DDARF are proposed, including lining reinforcement, full-length anchorage bolt and equivalent reinforcement. In the numerical simulation, lining part is assigned higher mechanical strength than surrounding rock masses, it may include multiple virtual joints or not, depending on projects. There must be no embedding or stretching between lining blocks and surrounding blocks. To realize simulation of the full-length anchorage bolt, at every discontinuity passed through the bolt, a set of normal and tangential spring needs to be added along the bolt's axial and tangential direction. Thus, bolt's axial force, shearing force and full-length anchorage effect are all realized synchronously. And, failure criterions of anchorage effect are established for different failure modes. In the meantime, from the perspective of improving surrounding rock masses' overall strength, a new equivalent and tentative simulation method is proposed, it can save calculation storage and improve efficiency. Along the text, simulation algorithms and applications of these new reinforcement modes in DDARF are given.

• Advances in concrete construction
• /
• v.13 no.3
• /
• pp.243-254
• /
• 2022

Reinforced concrete (RC) deep beams are critical structural elements used in offshore pile caps, rectangular cross-section water tanks, silo structures, transfer beams in high-rise buildings, and bent caps. As a result of the low shear span ratio to effective depth (a/d) in deep beams, arch action occurs, which leads to shear failure. Several studies have been carried out to improve the shear resistance of RC deep beams and avoid brittle fracture behavior in recent years. This study was performed to investigate the behavior of RC deep beams numerically and experimentally with different reinforcement arrangements. Deep beams with four different reinforcement arrangements were produced and tested under monotonic static loading in the study's scope. The horizontal and vertical shear reinforcement members were changed in the test specimens to obtain the effects of different reinforcement arrangements. However, the rebars used for tension and the vertical shear reinforcement ratio were constant. In addition, the behavior of each deep beam was obtained numerically with commercial finite element analysis (FEA) software ABAQUS, and the findings were compared with the experimental results. The results showed that the reinforcements placed diagonally significantly increased the load-carrying and energy absorption capacities of RC deep beams. Moreover, an apparent plastic plateau was seen in the load-displacement curves of these test specimens in question (DE-2 and DE-3). This finding also indicated that diagonally located reinforcements improve displacement ductility. Also, the numerical results showed that the FEM method could be used to accurately predict RC deep beams'behavior with different reinforcement arrangements.