检索结果-南通市图书馆

End-to-End Autonomous Driving Through Dueling Double deep q-network

在线全文

学校读者我要写书评

暂无评论

Automotive Innovation 2021年第3期4卷 328-337页

作者： Baiyu Peng qi Sun Shengbo Eben Li Dongsuk Kum Yuming Yin Junqing Wei Tianyu Gu State Key Lab of Automotive Safety and Energy School of Vehicle and MobilityTsinghua UniversityBeijing 100084China Korea Advanced Institute of Science and Technology 193Munji-roYuseong-guDaejeonKorea DiDi Autonomous Driving Company Beijing 100084China

Recent years have seen the rapid development of autonomous driving systems,which are typically designed in a hierarchical architecture or an end-to-end architecture.The hierarchical architecture is always complicated and hard to design,while the end-to-end architecture is more promising due to its simple structure.This paper puts forward an end-to-end autonomous driving method through a deep reinforcement learning algorithm Dueling Double deep q-network,making it possible for the vehicle to learn end-to-end driving by itself.This paper firstly proposes an architecture for the end-to-end lane-keeping task.Unlike the traditional image-only state space,the presented state space is composed of both camera images and vehicle motion information.Then corresponding dueling neural network structure is introduced,which reduces the variance and improves sampling efficiency.Thirdly,the proposed method is applied to The Open Racing Car Simulator(TORCS)to demonstrate its great performance,where it surpasses human drivers.Finally,the saliency map of the neural network is visualized,which indicates the trained network drives by observing the lane lines.A video for the presented work is available online,https://youtu.be/76ciJ mIHMD8 or https://***/v_show/id_XNDM4 ODc0M TM4NA==.html.

关键词： End-to-end autonomous driving Reinforcement learning deep q-network Neural network

UAV Autonomous Navigation for Wireless Powered Data Collection with Onboard deep q-network

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

ZTE Communications 2023年第2期21卷 80-87页

作者： LI Yuting DING Yi GAO Jiangchuan LIU Yusha HU Jie YANG Kun School of Information and Communication Engineering University of Electronic Science and Technology of ChinaChengdu 611731China China Mobile Communications Group Jilin Co. Ltd.Changchun 130061China School of Computer Science and Electronic Engineering University of EssexColchester CO43SQUnited Kingdom

In a rechargeable wireless sensor network,utilizing the unmanned aerial vehicle(UAV)as a mobile base station(BS)to charge sensors and collect data effectively prolongs the network’s lifetime.In this paper,we jointly optimize the UAV’s flight trajectory and the sensor selection and operation modes to maximize the average data traffic of all sensors within a wireless sensor network(WSN)during finite UAV’s flight time,while ensuring the energy required for each sensor by wireless power transfer(WPT).We consider a practical scenario,where the UAV has no prior knowledge of sensor locations.The UAV performs autonomous navigation based on the status information obtained within the coverage area,which is modeled as a Markov decision process(MDP).The deep q-network(DqN)is employed to execute the navigation based on the UAV position,the battery level state,channel conditions and current data traffic of sensors within the UAV’s coverage area.Our simulation results demonstrate that the DqN algorithm significantly improves the network performance in terms of the average data traffic and trajectory design.

关键词： unmanned aerial vehicle wireless power transfer deep q-network autonomous navigation

A New Feature Selection Algorithm Based on deep q-network

在线全文

学校读者我要写书评

暂无评论

A New Feature Selection Algorithm Based on Deep Q-Network

Improving SINR via Joint Beam and Power Management for GEO and LEO Spectrum-Sharing Satellite Communication Systems

第40届中国控制会议

作者： Xinqian Li Jie Yao Jia Ren Liqiang Wang Faculty of Mechanical Engineering and Automation Zhejiang Sci-Tech University College of Mechanical & Automotive Engineering Zhejiang University of Water Resources & Electric Power

In machine learning tasks, feature selection is an important data preprocessing step. It improves the efficiency and prediction accuracy by removing redundant and irrelevant features in a dataset. Feature selection is essentially a search process for the optimal feature subset. Although the traditional feature selection methods can search feature subset effectively, they still have the problems of large scale and low classification accuracy. In this paper, a new automatic feature selection method based on deep q-network（DqN-FS） is proposed to solve the shortcomings of traditional feature selection. Firstly, the feature selection problem is formulated as a Markov decision process. Subsequently, the q network is used to optimize the search strategy of the optimal feature subset. Finally, the parameters of the q network are updated by stochastic gradient descent. This paper also considers an extensive study of the feature sequence setting for the proposed technique. The proposed methods are tested on 14 UCI standard datasets and then compared with mainstream feature selection methods. The results show the efficiency of DqNFS in searching for the optimal feature subsets.

关键词： Feature Selection deep q-network deep Reinforcement Learning Feature Ranking

来源： cnki会议评论

在线全文

cnki会议

学校读者我要写书评

暂无评论

China Communications 2022年第7期19卷 25-36页

作者： Xiaojin Ding Zhuangzhuang Ren Huanbin Lu Gengxin Zhang Telecommunication and networks National Engineering Research Center Nanjing University of Posts and TelecommunicationsNanjing 210003China

In this paper,we investigate a geosynchronous earth orbit(GEO)and low earth orbit(LEO)coexisting satellite communication system.To decrease the interference imposed on the GEO user caused by LEO satellites,we propose a joint beammanagement and power-allocation(JBMPA)scheme to maximize signal-to-interference plus noise ratio(SINR)at the GEO user,whilst maintaining the ongoing wireless links spanning from LEO satellites to their corresponding users.Specifically,we first analyze the overlapping coverage among GEO and LEO satellites,to obtain the LEO-satellite set in which their beams impose interference on the GEO user.Then,considering the traffic of LEO satellites in the obtained set,we design a beam-management method to turn off and switch interference beams of LEO satellites.Finally,we further propose a deep q-network(DqN)aided power allocation algorithm to allocate the transmit power for the ongoing LEO satellites in the obtained set,whose beams are unable to be managed.Numerical results show that comparing with the traditional fixed beam with power allocation(FBPA)scheme,the proposed JBMPA can achieve a higher SINR and a lower outage probability,whilst guaranteeing the ongoing wireless transmissions of LEO satellites.

关键词： beam management deep q-network GEO-LEO coexisting power allocation

Adaptive Electrospinning System Based on Reinforcement Learning for Uniform‑Thickness Nanofiber Air Filters

在线全文

学校读者我要写书评

暂无评论

Advanced Fiber Materials 2023年第2期5卷 617-631页

作者： Seok Hyeon Hwang Jin Yeong Song Hyun Il Ryu Jae Hee Oh Seungwook Lee Donggeun Lee Dong Yong Park Sang Min Park School of Mechanical Engineering Pusan National University63‑2 Busan University‑roGeumjeong‑guBusan 46241South Korea Smart Manufacturing Technology R&D Group Korea Institute of Industrial Technology320 Techno Sunhwan‑roYuga‑eupDalseong‑gunDaegu 42994Republic of Korea

Electrospinning is a simple and versatile method to produce nanofiber filters.However,owing to bending instability that occurs during the electrospinning process,electrospinning has frequently produced a non-uniform-thickness nanofiber filter,which deteriorates its air filtration.Here,an adaptive electrospinning system based on reinforcement learning(E-RL)was developed to produce uniform-thickness nanofiber filters.The E-RL accomplished a real-time thickness measurement of an electrospun nanofiber filter by measuring the transmitted light through the nanofiber filter using a camera placed at the bottom of the collector and converting it into thickness using the Beer–Lambert law.Based on the measured thickness,the E-RL detected the non-uniformity of the nanofiber filter thickness and manipulated the movable collector to alleviate the non-uniformity of the thickness by a pre-trained reinforcement learning(RL)algorithm.For the training of the RL algo-rithm,the nanofiber production simulation software based on the empirical model of the deposition of the nanofiber filter was developed,and the training process of the RL algorithm was repeated until the optimal policy was achieved.After the training process with the simulation software,the trained model was transferred to the adaptive electrospinning system.By the movement of the collector under the optimal strategy of RL algorithm,the non-uniformity of such nanofiber filters was significantly reduced by approximately five times in standard deviation and error for both simulation and experiment.This finding has great potential in improving the reliability of electrospinning process and nanofiber filters used in research and industrial fields such as environment,energy,and biomedicine.

关键词： Electrospinning Nanofiber filter Uniform thickness Reinforcement learning deep q-network

The application of reinforcement learning to NATM tunnel design

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Underground Space 2022年第6期7卷 990-1002页

作者： Enrico Soranzo Carlotta Guardiani Wei Wu University of Natural Resources and Life Sciences Feistmantelstraße 4Vienna 1180Austria

The New Austrian Tunnelling Method(NATM)tunnel design is performed by testing support classes against the geological profile.We propose to replace this manual process with reinforcement learning,a generic framework within the realm of artificial intelligence that solves control tasks.Previous studies have demonstrated this possibility,albeit with methodological simplifications.We coupled the Finite Difference Method with a Python script,used the output of the first to train the machine learning model implemented in the latter and improved the choice of the support classes.Through benchmark tests,we demonstrated that our method was capable of choosing the optimal support classes for various geological sets and showed the relation between its performance and the number of training episodes.

关键词： deep q-network NATM Reinforcement learning Support classes Tunnelling

现货市场环境下基于深度强化学习的光储联合电站储能系统最优运行方法

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

电网技术 2022年第9期46卷 3365-3375页

作者：龚开王旭邓晖蒋传文马骏超房乐电力传输与功率变换控制教育部重点实验室(上海交通大学) 上海市闵行区200240 国网浙江省电力有限公司电力科学研究院浙江省杭州市310014 国网浙江省电力有限公司电力市场仿真实验室浙江省杭州市310014

光伏-储能联合电站不仅能够有效减少光伏实时出力偏差,也是一种能够提供电能量和调频辅助服务的潜在市场主体。为实现上述3类目标,与光伏出力协同的储能电量调度策略至关重要,然而目前大部分光储联合电站的储能电量调度策略无法同时协... 详细信息

光伏-储能联合电站不仅能够有效减少光伏实时出力偏差,也是一种能够提供电能量和调频辅助服务的潜在市场主体。为实现上述3类目标,与光伏出力协同的储能电量调度策略至关重要,然而目前大部分光储联合电站的储能电量调度策略无法同时协调降低光伏实时出力偏差和参与电能量与调频辅助服务市场3种决策;另一方面,电力现货市场价格与调频信号不确定性及储能电量调度策略将光储联合电站储能运行优化问题转化为一个随机动态非凸优化问题,现有相关研究大部分利用随机场景法或智能算法处理非凸优化,所获得的储能运行方案存在一定的局限性,且难以根据实时数据动态制定运行方案。因此,提出一种现货市场环境下基于DqN(deep q-network)的光储联合电站储能系统优化运行方法,该方法克服了非凸优化难题,结合所提储能电量闭环调度策略能够实现光储联合电站在考虑偏差考核成本、电能量收益、调频辅助服务收益下的储能系统小时级动态优化运行,进而最大化光-储联合电站的经济收益。测试算例通过实际市场数据验证了所提方法的可行性和有效性。

关键词：储能 deep q-network 不确定性电力市场最优运行

deep Reinforcement Learning-Based Computation Offloading for 5G Vehicle-Aware Multi-Access Edge Computing network

在线全文

学校读者我要写书评

暂无评论

China Communications 2021年第11期18卷 26-41页

作者： Ziying Wu Danfeng Yan State Key Laboratory of network and Switching Technology Beijing University of Posts and TelecommunicationsBeijing 100876China

Multi-access Edge Computing(MEC)is one of the key technologies of the future 5G network.By deploying edge computing centers at the edge of wireless access network,the computation tasks can be offloaded to edge servers rather than the remote cloud server to meet the requirements of 5G low-latency and high-reliability application scenarios.Meanwhile,with the development of IOV(Internet of Vehicles)technology,various delay-sensitive and compute-intensive in-vehicle applications continue to appear.Compared with traditional Internet business,these computation tasks have higher processing priority and lower delay requirements.In this paper,we design a 5G-based vehicle-aware Multi-access Edge Computing network(VAMECN)and propose a joint optimization problem of minimizing total system cost.In view of the problem,a deep reinforcement learningbased joint computation offloading and task migration optimization(JCOTM)algorithm is proposed,considering the influences of multiple factors such as concurrent multiple computation tasks,system computing resources distribution,and network communication bandwidth.And,the mixed integer nonlinear programming problem is described as a Markov Decision Process.Experiments show that our proposed algorithm can effectively reduce task processing delay and equipment energy consumption,optimize computing offloading and resource allocation schemes,and improve system resource utilization,compared with other computing offloading policies.

关键词： multi-access edge computing computation offloading 5G vehicle-aware deep reinforcement learning deep q-network

在线全文

学校读者我要写书评

暂无评论

基于深度强化学习的高频量化交易策略研究

现代电子技术 2023年第2期46卷 125-131页

作者：文馨贤昆明理工大学信息工程与自动化学院云南昆明650500

当前国内金融市场的投资交易已从基于传统技术分析等方法的主观交易逐渐转向基于程序化的量化策略交易。股票市场已有大量量化策略的研究工作,但针对期货市场的量化交易策略的研究还不足,已有策略在日内高频交易中的投资回报和风险控制... 详细信息

当前国内金融市场的投资交易已从基于传统技术分析等方法的主观交易逐渐转向基于程序化的量化策略交易。股票市场已有大量量化策略的研究工作,但针对期货市场的量化交易策略的研究还不足,已有策略在日内高频交易中的投资回报和风险控制还有待优化。为提升期货高频量化策略的盈利和风控能力,文中设计一种期货交易环境,将1 min时间粒度的高频K线作为环境状态,针对期货交易中持仓状态和交易操作构建相应的动作空间及算法;采用基于LSTM的深度强化学习模型LSTM-Dueling DqN,使其更适用于处理序列输入的状态空间,并显著提升模型的学习速度。对DqN、Double DqN、基于全连接神经网络的Dueling DqN(FF-Dueling DqN)三个基准模型进行实验对比,得到文中构建的交易策略在四个黑色系商品期货交易中累计收益率最高达到43%,年化收益率达到153%,最大回撤控制在10.7%以内。实验结果表明,所提策略在震荡行情和趋势行情中都能实现超出业绩基准的超额收益。

关键词：交易策略深度强化学习 LSTM deep q-network 高频交易期货量化金融