Adaptive Q-Learning for Data-Based Optimal Output Regulation With Experience Replay

doi:10.1109/TCYB.2018.2821369

CASIA OpenIR > 多模态人工智能系统全国重点实验室 > 复杂系统智能机理与平行控制团队

	Adaptive Q-Learning for Data-Based Optimal Output Regulation With Experience Replay
	Luo, Biao1 ; Yang, Yin 2; Liu, Derong 3
发表期刊	IEEE TRANSACTIONS ON CYBERNETICS
ISSN	2168-2267
	2018-12-01
卷号	48 期号:12 页码:3337-3348
通讯作者	Luo, Biao(biao.luo@hotmail.com)
摘要	In this paper, the data-based optimal output regulation problem of discrete-time systems is investigated. An off-policy adaptive Q-learning (QL) method is developed by using real system data without requiring the knowledge of system dynamics and the mathematical model of utility function. By introducing the Q-function, an off-policy adaptive QI, algorithm is developed to learn the optimal Q-function. An adaptive parameter alpha(i) in the policy evaluation is used to achieve tradeoff between the current and future Q-functions. The convergence of adaptive QI, algorithm is proved and the influence of the adaptive parameter is analyzed. To realize the adaptive QL algorithm with real system data, the actor-critic neural network (NN) structure is developed. The least-squares scheme and the batch gradient descent method are developed to update the critic and actor NN weights, respectively. The experience replay technique is employed in the learning process, which leads to simple and convenient implementation of the adaptive QL method. Finally, the effectiveness of the developed adaptive QL method is verified through numerical simulations.
关键词	Data-based experience replay neural networks (NNs) off-policy optimal control Q-learning (QL)
DOI	10.1109/TCYB.2018.2821369
关键词[WOS]	DISCRETE-TIME-SYSTEMS ; H-INFINITY CONTROL ; SPATIALLY DISTRIBUTED PROCESSES ; UNCERTAIN NONLINEAR-SYSTEMS ; BARRIER LYAPUNOV FUNCTIONS ; POLICY ITERATION ; CONTROL DESIGN ; CONTROLLER-DESIGN ; UNKNOWN DYNAMICS ; TRACKING CONTROL
收录类别	SCI
语种	英语
资助项目	Qatar National Research Fund under National Priority Research Project[NPRP9-466-1-103] ; National Natural Science Foundation of China[U1501251] ; National Natural Science Foundation of China[61533017] ; National Natural Science Foundation of China[61503377] ; National Natural Science Foundation of China[61503377] ; National Natural Science Foundation of China[61533017] ; National Natural Science Foundation of China[U1501251] ; Qatar National Research Fund under National Priority Research Project[NPRP9-466-1-103]
项目资助者	National Natural Science Foundation of China ; Qatar National Research Fund under National Priority Research Project
WOS研究方向	Automation & Control Systems ; Computer Science
WOS类目	Automation & Control Systems ; Computer Science, Artificial Intelligence ; Computer Science, Cybernetics
WOS记录号	WOS:000450613100007
出版者	IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
引用统计	被引频次：99[WOS] [WOS记录] [WOS相关记录]
文献类型	期刊论文
条目标识符	http://ir.ia.ac.cn/handle/173211/22603
专题	多模态人工智能系统全国重点实验室_复杂系统智能机理与平行控制团队
通讯作者	Luo, Biao
作者单位	1.Chinese Acad Sci, Inst Automat, State Key Lab Management & Control Complex Syst, Beijing 100190, Peoples R China 2.Hamad Bin Khalifa Univ, Coll Sci & Engn, Doha, Qatar 3.Guangdong Univ Technol, Sch Automat, Guangzhou 510006, Guangdong, Peoples R China
第一作者单位	中国科学院自动化研究所
通讯作者单位	中国科学院自动化研究所
推荐引用方式 GB/T 7714	Luo, Biao,Yang, Yin,Liu, Derong. Adaptive Q-Learning for Data-Based Optimal Output Regulation With Experience Replay[J]. IEEE TRANSACTIONS ON CYBERNETICS,2018,48(12):3337-3348.
APA	Luo, Biao,Yang, Yin,&Liu, Derong.(2018).Adaptive Q-Learning for Data-Based Optimal Output Regulation With Experience Replay.IEEE TRANSACTIONS ON CYBERNETICS,48(12),3337-3348.
MLA	Luo, Biao,et al."Adaptive Q-Learning for Data-Based Optimal Output Regulation With Experience Replay".IEEE TRANSACTIONS ON CYBERNETICS 48.12(2018):3337-3348.