Physics-informed deep Koopman operator for Lagrangian dynamic systems

被引:0
|
作者
Xuefeng WANG [1 ]
Yang CAO [1 ]
Shaofeng CHEN [1 ]
Yu KANG [1 ,2 ,3 ]
机构
[1] Department of Automation, University of Science and Technology of China
[2] State Key Laboratory of Fire Science, University of Science and Technology of China
[3] Institute of Advanced Technology, University of Science and Technology of
关键词
D O I
暂无
中图分类号
O19 [动力系统理论]; TP13 [自动控制理论];
学科分类号
070104 ; 0711 ; 071101 ;
摘要
Accurate mechanical system models are crucial for safe and stable control. Unlike linear systems,Lagrangian systems are highly nonlinear and difficult to optimize because of their unknown system model.Recent research thus used deep neural networks to generate linear models of original systems by mapping nonlinear dynamic systems into a linear space with a Koopman observable function encoder. The controller then relies on the Koopman linear model. However, without physical information constraints, ensuring control consistency between the original nonlinear system and the Koopman system is tough, as the learning process of the Koopman observation function is unsupervised. This paper thus proposes a two-stage learning algorithm that uses structural subnetworks to build a physics-informed network topology to simultaneously learn the Koopman observable functions and the system energy representation. In the Koopman matrix learning session, a quadratic-constrained optimization problem is solved to ensure that the Koopman representation satisfies the energy difference matching hard constraint. The proposed energy-preserving deep Lagrangian Koopman(EPDLK) framework effectively represents the dynamics of the Lagrangian system while ensuring control consistency. The effectiveness of EPDLK is compared with those of various Koopman observable function construction methods in multistep prediction and trajectory tracking tasks. EPDLK achieves better control consistency by guaranteeing energy difference matching, which facilitates the application of the control law generated on the Koopman system directly to the original nonlinear Lagrangian system.
引用
收藏
页码:146 / 163
页数:18
相关论文
共 50 条
  • [21] Physics-Informed Dynamic Mode Decomposition for Power Systems Dynamic State Estimation with Uncertainties
    Yi, Wenting
    Bu, Siqi
    18TH INTERNATIONAL CONFERENCE ON PROBABILISTIC METHODS APPLIED TO POWER SYSTEMS, PMAPS 2024, 2024, : 323 - 328
  • [22] Surrogate modeling for radiative heat transfer using physics-informed deep neural operator networks
    Lu, Xiaoyi
    Wang, Yi
    PROCEEDINGS OF THE COMBUSTION INSTITUTE, 2024, 40 (1-4)
  • [23] Efficient Uncertainty Quantification in Electromagnetic Modeling Using Physics-Informed Deep Operator Neural Networks
    Qi, Shutong
    Sarris, Costas D.
    2024 IEEE INC-USNC-URSI RADIO SCIENCE MEETING (JOINT WITH AP-S SYMPOSIUM), 2024, : 159 - 159
  • [24] STACKED NETWORKS IMPROVE PHYSICS-INFORMED TRAINING: APPLICATIONS TO NEURAL NETWORKS AND DEEP OPERATOR NETWORKS
    Howard, Amanda a.
    Murphy, Sarah h.
    Ahmed, Shady e.
    Stinis, Panos
    FOUNDATIONS OF DATA SCIENCE, 2024,
  • [25] Physics-Informed deep Autoencoder for fault detection in New-Design systems
    Lai, Chenyang
    Baraldi, Piero
    Zio, Enrico
    MECHANICAL SYSTEMS AND SIGNAL PROCESSING, 2024, 215
  • [26] Deep Koopman Operator With Control for Nonlinear Systems
    Shi, Haojie
    Meng, Max Q-H
    IEEE ROBOTICS AND AUTOMATION LETTERS, 2022, 7 (03) : 7700 - 7707
  • [27] Generic bounds on the approximation error for physics-informed (and) operator learning
    De Ryck, Tim
    Mishra, Siddhartha
    ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 35 (NEURIPS 2022), 2022,
  • [28] A physics-informed kernel approach to learning the operator for parametric PDEs
    J. Kurz
    B. Bowman
    M. Seman
    C. Oian
    T. Khan
    Neural Computing and Applications, 2024, 36 (36) : 22773 - 22787
  • [29] Physics-informed neural networks for multi-stage Koopman modeling of microbial fermentation
    Li, Quan
    Zhang, Jingran
    Wan, Haiying
    Zhao, Zhonggai
    Liu, Fei
    JOURNAL OF PROCESS CONTROL, 2024, 143
  • [30] Lagrangian large eddy simulations via physics-informed machine learning
    Tian, Yifeng
    Woodward, Michael
    Stepanov, Mikhail
    Fryer, Chris
    Hyett, Criston
    Livescu, Daniel
    Chertkov, Michael
    PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA, 2023, 120 (34)