检索结果-内蒙古大学图书馆

Deep reinforcement learning implementation on IC engine idle speed control

AIN SHAMS ENGINEERING JOURNAL 2024年第5期15卷

作者： Omran, Ibrahim Mostafa, Ahmed Seddik, Ahmed Ali, Mohamed Hussein, Mohand Ahmed, Youssef Aly, Youssef Abdelwahab, Mohamed Ain Shams Univ Fac Engn Publ Works Dept 1 El Sarayat StABBASSEYAEl Weili Cairo 11535 Egypt

Efficient control of automotive engine idle speed is crucial for achieving better fuel economy and smoother engine running. This paper presents a comparison between proportional-integral-derivative (PID) control and Reinforcement Learning (RL) using the Deep Q-Network (dqn) algorithm as a high-level control method for minimizing idle speed fluctuations caused by changes in engine irregularities, and the response time and accuracy of the throttle control mechanism. In addition to low-level PID control for the throttle valve position, MATLAB/Simulink was employed to build the simulation environment, incorporating an engine model and an electronic throttle body model, and observing the engine's current speed. The results demonstrated the superiority of RL-based control over PID in reducing idle speed fluctuations and enhancing engine performance in simulations and real -world experiments. This study advances automotive engine control strategies.

关键词： Engine idle speed control Reinforcement learning dqn algorithm Engine model PID Fuel economy

来源：评论

学校读者我要写书评

暂无评论

SAAS parallel task scheduling based on cloud service flow load algorithm

引用

COMPUTER COMMUNICATIONS 2022年 182卷 170-183页

作者： Zhu, Jian Li, Qian Ying, Shi Wuhan Univ Sch Comp Sci Wuhan 430072 Peoples R China Guangxi Vocat Normal Univ Sch Comp & Informat Engn Nanning 530007 Peoples R China

In cloud platform applications, the user's goal is to obtain high-quality application services, while the service provider's goal is to obtain revenue by performing the tasks submitted by the user. The platform built by the service provider's application resources needs to improve the mapping between service requests and resources to achieve higher value. Through the current situation of resource management in the cloud environment, it is found that many task scheduling and resource allocation algorithms are still affected by factors such as the diversity, dynamics, and multiple constraints of resources and tasks. This paper focuses on Software as a Service (SaaS) applications' task scheduling and resource configuration in a dynamic and uncertain cloud environment. It is a challenging online scheduling problem to automatically and intelligently allocate user task requests that continually reach SaaS applications to appropriate resources for execution. To this end, a real-time task scheduling method based on deep reinforcement learning is proposed, which automatically and intelligently allocates user task requests that continually reach SaaS applications to appropriate resources for execution. In this way, the limited virtual machine resources rented by SaaS providers can be used in a balanced and efficient manner. In the experiment, by comparing with other five task scheduling algorithms, it is proved that the algorithm proposed in this paper not only improves the execution efficiency of better deploying workflow in IaaS public cloud, but also makes the resources provided by SaaS are used in a balanced and efficient manner.

关键词： SAAS task scheduling Cloud service flow Network space status Parallel task management dqn algorithm

来源：评论

学校读者我要写书评

暂无评论

Improved Q-Learning-Based Motion Control for Basketball Intelligent Robots Under Multi-Sensor Data Fusion

引用

IEEE ACCESS 2024年 12卷 57059-57070页

作者： Lan, Jianping Dong, Xiujuan Macau Univ Sci & Technol Fac Innovat Engn Macau Peoples R China Hubei Univ Automot Technol Inst Automot Engineers Shiyan 442002 Peoples R China

With the continuous development of science and technology, the application of artificial intelligence in robot technology, especially in the progress of sports robot technology, has received great attention. This development provides new opportunities for the development of intelligent basketball robots. However, achieving high accuracy and stability in motion control remains a challenge in the research field. The motion control technology of basketball intelligent robots is crucial. The research proposes a study on enhancing Q-learning-based action control technology for basketball robots with multi-sensor data fusion to address the problem of inaccuracy and instability. The first step involves examining how multi-sensor data fusion technology can offer more accurate and comprehensive environmental information that can aid the robot in making more precise action decisions. The second area of study involves the development of enhanced algorithms for controlling basketball robots through improved Q-learning techniques. The goal is to increase the accuracy and stability of the robots' actions. The final focus is on creating a model for controlling the basketball robots' actions, which is based on the improved Q-learning algorithm. This model effectively increases the accuracy and stability of the robots' actions. The experiment was conducted on a computer equipped with Intel Core i9-11900K processor and NVIDIA GeForce RTX 3080 graphics card. MATLAB R2023a was used for system parameterization and simulation, while Webots R2023a software was used to simulate robot actions. The sensor adopted Hokuyo URG-04LX-UG01 laser rangefinder, and the robot model was KUKA youBot. At the same time, in order to improve the control performance of basketball intelligent robots, comparisons were made with existing methods such as PID control, trajectory planning, traditional control methods, multi-sensor fusion technology, and visual perception basketball shooting correction technology.

关键词： Multi-sensor data fusion Q-learning algorithm motion control basketball robot dqn algorithm

来源：评论

学校读者我要写书评

暂无评论

An integration model for Texas Hold'em

引用

INTERNATIONAL JOURNAL OF COMPUTING SCIENCE AND MATHEMATICS 2023年第3期18卷 203-213页

作者： Wang, Yajie Han, Shengyu Wei, Zhihao Shi, Zhonghui Shenyang Aerosp Univ Engn Training Ctr Shenyang 110000 Liaoning Peoples R China Shenyang Aerosp Univ Dept Comp Sci Shenyang 110000 Liaoning Peoples R China

Texas Hold'em is a representative of an incomplete information game. Existing research on computing Nash equilibrium as a Texas Hold'em strategy has problems, including high resource consumption and conservative strategies. To solve the above problems, an integration model combining deep learning and reinforcement learning is proposed. Firstly, to reduce the storage resources consumed due to the large Texas Hold'em state space, a long short-term memory (LSTM) is designed to predict the game results. Since the win rate and historical action information are used as input data by the LSTM, a convolutional neural network (CNN) is designed to predict the current win rate. Secondly, in order to enable the strategy to have dynamic adjustment ability, the deep Q-network(dqn) is used to generate the strategy by using the results predicted by LSTM. Finally, an agent is implemented to provide training data for LSTM. The experimental results show that the model wins more chips, which proves that it can be used as a solution for incomplete information games.

关键词： Texas Hold' em reinforcement learning deep learning dqn algorithm integration model

来源：评论

学校读者我要写书评

暂无评论

Fault-tolerant control system for once-through steam generator based on reinforcement learning algorithm

引用

NUCLEAR ENGINEERING AND TECHNOLOGY 2022年第9期54卷 3283-3292页

作者： Li, Cheng Yu, Ren Yu, Wenmin Wang, Tianshu Naval Univ Engn Wuhan 430033 Peoples R China China Nucl Power Operat Technol Corp LTD Wuhan 430000 Peoples R China

Based on the Deep Q-Network(dqn) algorithm of reinforcement learning, an active fault-tolerance method with incremental action is proposed for the control system with sensor faults of the once -through steam generator(OTSG). In this paper, we first establish the OTSG model as the interaction environment for the agent of reinforcement learning. The reinforcement learning agent chooses an ac-tion according to the system state obtained by the pressure sensor, the incremental action can gradually approach the optimal strategy for the current fault, and then the agent updates the network by different rewards obtained in the interaction process. In this way, we can transform the active fault tolerant control process of the OTSG to the reinforcement learning agent's decision-making process. The com-parison experiments compared with the traditional reinforcement learning algorithm(RL) with fixed strategies show that the active fault-tolerant controller designed in this paper can accurately and rapidly control under sensor faults so that the pressure of the OTSG can be stabilized near the set-point value, and the OTSG can run normally and stably.(c) 2022 Korean Nuclear Society, Published by Elsevier Korea LLC. All rights reserved. This is an open access article under the CC BY-NC-ND license (http://***/licenses/by-nc-nd/4.0/).

关键词： Once -through steam generator Reinforcement learning dqn algorithm Incremental action

来源：评论

学校读者我要写书评

暂无评论

A Dynamic Configuration Design Method Based on Cloud Model and Deep Reinforcement Learning 29

A Dynamic Configuration Design Method Based on Cloud Model a...

引用

29th International Conference on Automation and Computing (ICAC)

作者： Li, Rong Zhang, Haizhu Qin, Shengfeng Yang, Hao Zhang, Yongjie Southwest Jiaotong Univ Sch Mech Engn Chengdu Peoples R China Northumbria Univ Sch Design Newcastle Upon Tyne Tyne & Wear England

ISBN: (纸本)9798350360882;9798350360899

Rapidly developing technology and changing market make customer requirement (CR) change rapidly. In the early stage of product development, timely access to CRs is crucial. However, the previous design methods cannot grasp the changes of customers requirement preferences in time and give a better design scheme. Therefore, a dynamic configuration design method based on cloud model dealing with requirement uncertainty and obtaining CR preference is proposed to quickly solve the design scheme. Firstly, the reverse cloud conversion algorithm named Multiple Backward Cloud Transformation based on Sampling with Replacement (MBCT-SR) is used to convert the CR data into a multi-level evaluation cloud model, and the CR preference is calculated. Secondly, the design structure matrix is constructed to map the CR preference to the configuration instance. Then, the Deep Q Network (dqn) algorithm model is established and trained, and the updated CR preference is input to quickly solve the product configuration scheme. Finally, the effectiveness of the proposed method is verified by an example analysis of high-speed train bogies.

关键词： Complex product configuration design Cloud model Requirement preference dqn algorithm

来源：评论

学校读者我要写书评

暂无评论

QoE Estimation of DASH-Based Mobile Video Application Using Deep Reinforcement Learning 20th

QoE Estimation of DASH-Based Mobile Video Application Using ...

引用

20th International Conference on algorithms and Architectures for Parallel Processing (ICA3PP)

作者： Hou, Biao Zhang, Junxing Inner Mongolia Univ Hohhot Peoples R China

ISBN: (纸本)9783030602390;9783030602383

An increasing number of video content providers have adopted adaptive bitrate (ABR) streaming via the HTTP protocol. The client players usually run an ABR algorithm to determine the optimal quality of video playback in the next few seconds. Faced with unpredictable bandwidth variability, the latest ABR algorithm attempts to achieve the best balance between competing goals of high bitrate, less rebuffering and high smoothness. However, there is no guarantee that optimal resource utilization ensures a high quality of experience (QoE). QoE is also affected by users' preferences for video content. Even for the same movie clip, different users have varied preferences for characters, scenes, plots and other content. In this paper, we propose a Deep-Q Learning Network (dqn) based ABR algorithm to optimize the use of network and client resources in video playback and also improve QoE of users.

关键词： MPEG-DASH protocol ABR algorithm QoE model dqn algorithm

来源：评论

学校读者我要写书评

暂无评论

Security and 5G: Attack mitigation using Reinforcement Learning in SDN networks

Security and 5G: Attack mitigation using Reinforcement Learn...

引用

IEEE Future Networks World Forum (FNWF)

作者： Alvaro Fernandez-Carrasco, Jose Segurola-Gil, Lander Zola, Francesco Orduna-Urrutia, Raul BRTA Fdn Vicomtech Mikeletegi 57 Donostia San Sebastian 20009 Spain

ISBN: (纸本)9781665462501

5G ecosystem is shaping the future of communication networks enabling innovation and digital transformation not only for individual users but also for companies, industries, and communities. In this scenario, technologies such as Software Defined Networking (SDN) represent a solution for telecommunications providers to create agile, scalable, efficient platforms capable of meeting the requirements in the 5G ecosystem. However, as network environments and systems become increasingly complex, both in terms of size and dynamic behavior, the number of vulnerabilities in them can be very high. In addition, hackers are continuously improving intrusion methods, which are becoming more difficult to detect. For this reason, in this study, we deploy a system based on a Reinforcement Learning (RL) agent capable of applying different countermeasures to defend a network against intrusion and DDoS attacks using SDN. The approach is drawn like a serious game in which a defender and an attacker carry out actions based on the observations they get from the environment, i.e., network current status. In this study, defenders and attackers are trained using the Deep Q-Learning (dqn) algorithm with some variations, like Prioritized Replay, Dueling, and Double dqn, comparing their results in order to get the best strategy for attack mitigation. The results of this paper show that RL algorithms can be successfully used to create more versatile agents able of interpreting and adapting themselves to different situations and so run the best countermeasure to protect the network. According to the results, it is also shown that the Complete strategy, which includes the three dqn variations analyzed, is the one that allows obtaining agents with the best decision making to respond to attacks.

关键词： 5G SDN Reinforcement Learning dqn algorithm Cybersecurity

来源：评论

学校读者我要写书评

暂无评论

A Constraint Satisfaction Service Composition Method Supporting One to Many Task Pattern

A Constraint Satisfaction Service Composition Method Support...

引用

IEEE International Conference on Services Computing (SCC)

作者： Chu, Weijie Wang, Yuxuan Mo, Tong Li, Weiping Peking Univ Sch Software & Microelect Beijing Peoples R China

ISBN: (纸本)9781665416832

The current service composition problem is usually aimed at the constrained service composition problem of Web services. However, this kind of scheme is not suitable for the situation that one service corresponds to multiple tasks. In order to solve the problem of service composition meeting business constraints and users' needs in a specific field, this paper uses Markov decision process to model the problem, proposes a method based on deep Q-learning to solve the problem, and uses random sampling method for the inference. This method calculates the candidate services that can be used for composition, and takes the maximum cumulative rewards represented by the degree of constraint satisfaction as the optimization objective, and globally optimizes the results to meet the needs of users to the greatest extent. The experimental results show that: in this problem, compared with the existing methods, this method has higher combination efficiency.

关键词： dqn algorithm Service Composition Constraint Satisfaction MDP Deep Exploration

来源：评论

学校读者我要写书评

暂无评论

Research and Optimization of Urban Rail Transit Handover Based on WLAN 9

Research and Optimization of Urban Rail Transit Handover Bas...

引用

9th International Conference on Traffic and Logistic Engineering (ICTLE)

作者： Wang ChengLong Wang YeLi Wang XinJi Lanzhou Jiaotong Univ Key Lab Optoelect Technol & Intelligent Control Natl Engn Res Ctr Technol & Equipment Technol Env Minist Educ Lanzhou Peoples R China

ISBN: (纸本)9781665427524

In order to adapt to the development trend of urban rail transit systems, the traditional IEEE 802.11 standard handover method is aimed at the problems of high handover delay and ping-pong handover. The train control system CBTC (Communication Based Train Control) handover is studied. This paper studies the WLAN (Wireless Local Area Network) handover strategy combined with dqn (Deep Q-Network) under the WLAN urban rail transit model. This strategy extracts and inputs train driving characteristic status information, and performs switching actions based on the signal strength of the serving AP and the target AP. Finally, after deep neural network training, the best switching points at different speeds are obtained. The optimal value obtained by the dqn algorithm is substituted into the model for simulation verification. The results show that compared with the traditional method, the throughput of this strategy is increased by 36%, and the packet delay is reduced by 55% Prove the effectiveness of dqn algorithm in WLAN urban rail transit handover. At the same time, this strategy can obtain the optimal switching position after training at different speeds and different AP distances. This makes the strategy widely applicable and makes the WLAN urban rail transit switching system safer and more reliable.

关键词： communication-based train control handover IEEE802.11 standard dqn algorithm

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：