• Nuclear Engineering and Technology
• /
• v.55 no.9
• /
• pp.3277-3290
• /
• 2023

Nuclear power plant (NPP) operations with multiple objectives and devices are still performed manually by operators despite the potential for human error. These operations could be automated to reduce the burden on operators; however, classical approaches may not be suitable for these multi-objective tasks. An alternative approach is deep reinforcement learning (DRL), which has been successful in automating various complex tasks and has been applied in automation of certain operations in NPPs. But despite the recent progress, previous studies using DRL for NPP operations have limitations to handle complex multi-objective operations with multiple devices efficiently. This study proposes a novel DRL-based approach that addresses these limitations by employing a continuous action space and straightforward binary rewards supported by the adoption of a soft actor-critic and hindsight experience replay. The feasibility of the proposed approach was evaluated for controlling the pressure and volume of the reactor coolant while heating the coolant during NPP startup. The results show that the proposed approach can train the agent with a proper strategy for effectively achieving multiple objectives through the control of multiple devices. Moreover, hands-on testing results demonstrate that the trained agent is capable of handling untrained objectives, such as cooldown, with substantial success.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.15 no.11
• /
• pp.4244-4274
• /
• 2021

The utilization of UAVs in various fields has led to the development of flying ad hoc network (FANET) technology. In a network environment with highly dynamic topology and frequent link changes, the traditional routing technology of FANET cannot satisfy the new communication demands. Traditional routing algorithm, based on geographic location, can "fall" into a routing hole. In view of this problem, we propose a geolocation routing protocol based on multi-agent reinforcement learning, which decreases the packet loss rate and routing cost of the routing protocol. The protocol views each node as an intelligent agent and evaluates the value of its neighbor nodes through the local information. In the value function, nodes consider information such as link quality, residual energy and queue length, which reduces the possibility of a routing hole. The protocol uses global rewards to enable individual nodes to collaborate in transmitting data. The performance of the protocol is experimentally analyzed for UAVs under extreme conditions such as topology changes and energy constraints. Simulation results show that our proposed QLGR-S protocol has advantages in performance parameters such as throughput, end-to-end delay, and energy consumption compared with the traditional GPSR protocol. QLGR-S provides more reliable connectivity for UAV networking technology, safeguards the communication requirements between UAVs, and further promotes the development of UAV technology.

• Proceedings of the Korea Inteligent Information System Society Conference
• /
• 2001.01a
• /
• pp.145-150
• /
• 2001

A Controlled Ecological Life Support System(CELSS) is essential for man to live a long time in a closed space such as a lunar base or a mars base. Such a system may be an extremely complex system that has a lot of facilities and circulates multiple substances,. Therefore, it is very difficult task to control the whole CELSS. Thus by regarding facilities constituting the CELSS as agents and regarding the status and action as information, the whole CELSS can be treated as multi-agent system(MAS). If a CELSS can be regarded as MAS the CELSS can have three advantages with the MAS. First the MAS need not have a central computer. Second the expendability of the CELSS increases. Third, its fault tolerance rises. However it is difficult to describe the cooperation protocol among agents for MAS. Therefore in this study we propose to apply reinforcement learning (RL), because RL enables and agent to acquire a control rule automatically. To prove that MAS and RL are effective methods. we have created the system in Java, which easily gives a distributed environment that is the characteristics feature of an agent. In this paper, we report the simulation results for material circulation control of the CELSS by the MAS and RL.

• Korean Journal of Cognitive Science
• /
• v.10 no.1
• /
• pp.1-11
• /
• 1999

Agent-based systems technology has generated lots of excitement in these years because of its promise as a new paradigm for conceptualizing. designing. and l implementing software systems Especially, there has been many researches for multi agent system because of the characteristics that it fits to the distributed and open Internet environments. In a multiagent system. agents must cooperate with each other through a Coordination procedure. when the conflicts between agents arise. where those are caused b by the point that each action acts for a purpose separately without coordination. But P previous researches for coordination methods in multi agent system have a deficiency that they can not solve correctly the cooperation problem between agents which have different goals in dynamic environment. In this paper. we solve the cooperation problem of multiagent that has multiple goals in a dynamic environment. with an automatic cooperative coordination model using I reinforcement learning. We will show the two pursuit problems that we extend a traditional problem in multi agent systems area for modeling the restriction in the multiple goals in a dynamic environment. and we have verified the validity of the proposed model with an experiment.

• Journal of KIISE:Software and Applications
• /
• v.35 no.3
• /
• pp.179-188
• /
• 2008

An important issue in multiagent reinforcement learning is how an agent should team its optimal policy through trial-and-error interactions in a dynamic environment where there exist other agents able to influence its own performance. Most previous works for multiagent reinforcement teaming tend to apply single-agent reinforcement learning techniques without any extensions or are based upon some unrealistic assumptions even though they build and use explicit models of other agents. In this paper, basic concepts that constitute the common foundation of multiagent reinforcement learning techniques are first formulated, and then, based on these concepts, previous works are compared in terms of characteristics and limitations. After that, a policy model of the opponent agent and a new multiagent reinforcement learning method using this model are introduced. Unlike previous works, the proposed multiagent reinforcement learning method utilize a policy model instead of the Q function model of the opponent agent. Moreover, this learning method can improve learning efficiency by using a simpler one than other richer but time-consuming policy models such as Finite State Machines(FSM) and Markov chains. In this paper. the Cat and Mouse game is introduced as an adversarial multiagent environment. And effectiveness of the proposed multiagent reinforcement learning method is analyzed through experiments using this game as testbed.

• Electronics and Telecommunications Trends
• /
• v.34 no.4
• /
• pp.1-14
• /
• 2019

Recent trends in deep reinforcement learning (DRL) have revealed the considerable improvements to DRL algorithms in terms of performance, learning stability, and computational efficiency. DRL also enables the scenarios that it covers (e.g., partial observability; cooperation, competition, coexistence, and communications among multiple agents; multi-task; decentralized intelligence) to be vastly expanded. These features have cultivated multi-agent reinforcement learning research. DRL is also expanding its applications from robotics to natural language processing and computer vision into a wide array of fields such as finance, healthcare, chemistry, and even art. In this report, we briefly summarize various DRL techniques and research directions.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.21 no.3
• /
• pp.175-181
• /
• 2021

This paper analyzes the artificial intelligence-based approach for short-term energy consumption prediction. In this paper, we employ the reinforcement learning algorithms to improve the limitation of the supervised learning algorithms which usually utilize to the short-term energy consumption prediction technologies. The supervised learning algorithm-based approaches have high complexity because the approaches require contextual information as well as energy consumption data for sufficient performance. We propose a deep reinforcement learning algorithm based on multi-agent to predict energy consumption only with energy consumption data for improving the complexity of data and learning models. The proposed scheme is simulated using public energy consumption data and confirmed the performance. The proposed scheme can predict a similar value to the actual value except for the outlier data.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2000.10a
• /
• pp.285-288
• /
• 2000

본 논문에서는 다중 에이전트(multi-agent) 환경에서 에이전트들의 행동을 효율적으로 조정 (coordination)하기 위해 강화 학습(reinforcement learning)을 이용하였다. 제안된 방법은 각 에이전트가 목표(goal)와의 거리 관계(distance relationship)와 인접 에이전트들과의 공간 관계(spatial relationship)를 이용하였다. 그러므로 각 에이전트는 다른 에이전트와 충돌(collision) 현상이 발생하지 않으면서, 최적의 다음 상태를 선택할 수 있다. 또한, 상태 공간으로부터 입력되는 강화 값이 0과 1 사이의 값을 갖기 때문에 각 에이전트가 선택한 (상태, 행동) 쌍이 얼마나 좋은가를 나타낼 수 있다. 제안된 방법을 먹이 포획 문제(prey pursuit problem)에 적용한 결과 지역 제어(local control)나. 분산 제어(distributed control) 전략을 이용한 방법보다 여러 에이전트들의 행동을 효율적으로 조정할 수 있었으며, 매우 빠르게 먹이를 포획할 수 있음을 알 수 있었다.

• Journal of KIISE:Software and Applications
• /
• v.30 no.7_8
• /
• pp.672-680
• /
• 2003

A shopbot is a software agent whose goal is to maximize buyer´s satisfaction through automatically gathering the price and quality information of goods as well as the services from on-line sellers. In the response to shopbots´ activities, sellers on the Internet need the agents called pricebots that can help them maximize their own profits. In this paper we adopts Q-learning, one of the model-free reinforcement learning methods as a price-setting algorithm of pricebots. A Q-learned agent increases profitability and eliminates the cyclic price wars when compared with the agents using the myoptimal (myopically optimal) pricing strategy Q-teaming needs to select a sequence of state-action fairs for the convergence of Q-teaming. When the uniform random method in selecting state-action pairs is used, the number of accesses to the Q-tables to obtain the optimal Q-values is quite large. Therefore, it is not appropriate for universal on-line learning in a real world environment. This phenomenon occurs because the uniform random selection reflects the uncertainty of exploitation for the optimal policy. In this paper, we propose a Mixed Nonstationary Policy (MNP), which consists of both the auxiliary Markov process and the original Markov process. MNP tries to keep balance of exploration and exploitation in reinforcement learning. Our experiment results show that the Q-learning agent using MNP converges to the optimal Q-values about 2.6 time faster than the uniform random selection on the average.

• 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.