Reinforcement learning-based adaptive PID controller for DPS

被引：39

作者：

Lee, Daesoo ^{[1
]}

Lee, Seung Jae ^{[1
]}

Yim, Solomon C. ^{[2
]}

机构：

[1] Korea Maritime & Ocean Univ, Div Naval Architecture & Ocean Syst Engn, Busan, South Korea

[2] Oregon State Univ, Sch Civil Construct & Environm Engn, Corvallis, OR 97331 USA

来源：

OCEAN ENGINEERING | 2020年 / 216卷

关键词：

Dynamic positioning system (DPS); Proportional-integral-derivative (PID); Fine-tuning; Deep reinforcement learning (DRL); Deep deterministic policy gradient (DDPG); DESIGN;

D O I：

10.1016/j.oceaneng.2020.108053

中图分类号：

U6 [水路运输]; P75 [海洋工程];

学科分类号：

0814 ; 081505 ; 0824 ; 082401 ;

摘要：

A conventional PID controller for the DPS has limitations due to fixed gains and dependence on manual adjustment for its gains. Therefore, several previous studies developed a fuzzy-based adaptive PID controller for the DPS which tunes the gains based on the fuzzy logic. However, the fuzzy logic has its disadvantages due to a manual definition of fuzzy rules and fuzzy variables. To overcome those limitations, a deep reinforcement learning algorithm is adopted to learn the efficient adaptive gain-tuning strategy without human intuition behind since it does not require any prior knowledge about the dynamics of a ship or DPS. Finally, it is shown that the proposed system can result in better station-keeping performance without deterioration in its control efficiency.

引用

页数：12

共 50 条

[1] A Proposal of Adaptive PID Controller Based on Reinforcement Learning
WANG Xue-song
[J]. International Journal of Mining Science and Technology, 2007, (01) : 40 - 44
[2] A Proposal of Adaptive PID Controller Based on Reinforcement Learning
WANG, Xue-song
CHENG, Yu-hu
SUN, Wei
[J]. Journal of China University of Mining and Technology, 2007, 17 (01): : 40 - 44
[3] Asymptotic tracking by a reinforcement learning-based adaptive critic controller
Shubhendu BHASIN
Nitin SHARMA
Parag PATRE
Warren DIXON
[J]. Control Theory and Technology, 2011, 9 (03) : 400 - 409
[4] Asymptotic tracking by a reinforcement learning-based adaptive critic controller
Bhasin S.
Sharma N.
Patre P.
Dixon W.
[J]. Journal of Control Theory and Applications, 2011, 9 (3): : 400 - 409
[5] Adaptive PID Controller based on Reinforcement Learning for Wind Turbine Control
Sedighizadeh, M.
Rezazadeh, A.
[J]. PROCEEDINGS OF WORLD ACADEMY OF SCIENCE, ENGINEERING AND TECHNOLOGY, VOL 27, 2008, 27 : 257 - 262
[6] Application of Reinforcement Learning-Based Adaptive PID Controller for Automatic Generation Control of Multi-Area Power System
Muduli, Rasananda
Jena, Debashisha
Moger, Tukaram
[J]. IEEE TRANSACTIONS ON AUTOMATION SCIENCE AND ENGINEERING, 2024, : 1 - 12
[7] A Reinforcement Learning-Based Adaptive Learning System
Shawky, Doaa
Badawi, Ashraf
[J]. INTERNATIONAL CONFERENCE ON ADVANCED MACHINE LEARNING TECHNOLOGIES AND APPLICATIONS (AMLTA2018), 2018, 723 : 221 - 231
[8] Intelligent PID Controller Based on Deep Reinforcement Learning
Zhai, Yinhe
Zhao, Qiang
Han, Yinghua
Wang, Jinkuan
Zeng, Wenying
[J]. 2024 8TH INTERNATIONAL CONFERENCE ON ROBOTICS, CONTROL AND AUTOMATION, ICRCA 2024, 2024, : 343 - 348
[9] Fuzzy PID Controller for UAV Based on Reinforcement Learning
Zhang, Benyi
Zhang, Weiping
Mou, Jiawang
Yang, Runmin
Zhang, Yichen
[J]. PROCEEDINGS OF 2022 INTERNATIONAL CONFERENCE ON AUTONOMOUS UNMANNED SYSTEMS, ICAUS 2022, 2023, 1010 : 1724 - 1732
[10] Reinforcement learning based adaptive PID controller design for control of linear/nonlinear unstable processes
Shuprajhaa, T.
Sujit, Shiva Kanth
Srinivasan, K.
[J]. APPLIED SOFT COMPUTING, 2022, 128

← 1 2 3 4 5 →