Gain-Only Neural Operators for PDE Backstepping (2403.19344v2)
Abstract: In this work we advance the recently-introduced deep learning-powered approach to PDE backstepping control by proposing a method that approximates only the control gain function -- a function of one variable -- instead of the entire kernel function of the backstepping transformation, which depends on two variables. This idea is introduced using several benchmark unstable PDEs, including hyperbolic and parabolic types, and extended to 2X2 hyperbolic systems. By employing a backstepping transformation that utilizes the exact kernel (suitable for gain scheduling) rather than an approximated one (suitable for adaptive control), we alter the quantification of the approximation error. This leads to a significant simplification in the target system, shifting the perturbation due to approximation from the domain to the boundary condition. Despite the notable differences in the Lyapunov analysis, we are able to retain stability guarantees with this simplified approximation approach. Approximating only the control gain function simplifies the operator being approximated and the training of its neural approximation, potentially reducing the neural network size. The trade-off for these simplifications is a more intricate Lyapunov analysis, involving higher Sobolev spaces for some PDEs, and certain restrictions on initial conditions arising from these spaces. It is crucial to carefully consider the specific requirements and constraints of each problem to determine the most suitable approach; indeed, recent works have demonstrated successful applications of both full-kernel and gain-only approaches in adaptive control and gain scheduling contexts.
- G. Andrade, R. Vazquez, and D. Pagano, “Backstepping stabilization of a linearized ODE-PDE Rijke tube model,” Automatica, vol. 96, pp. 98–109, 2018.
- J. Auriol and D. Bresch-Pietri, “Robust state-feedback stabilization of an underactuated network of interconnected n+m𝑛𝑚n+mitalic_n + italic_m hyperbolic PDE systems,” Automatica, vol. 136, p. 110040, 2022.
- L. Bhan, Y. Shi, and M. Krstic, “Neural operators for bypassing gain and control computations in PDE backstepping,” arXiv preprint arXiv:2302.14265, 2023.
- G. Chen, R. Vazquez, and M. Krstic, “Rapid stabilization of Timoshenko beam by PDE backstepping,” IEEE Transactions on Automatic Control, 2023.
- T. Chen and H. Chen, “Universal approximation to nonlinear operators by neural networks with arbitrary activation functions and its application to dynamical systems,” IEEE Transactions on Neural Networks, vol. 6, no. 4, pp. 911–917, 1995.
- J.-M. Coron, R. Vazquez, M. Krstic, and G. Bastin, “Local exponential H2superscript𝐻2{H^{2}}italic_H start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT stabilization of a 2×222{2\times 2}2 × 2 quasilinear hyperbolic system using backstepping,” SIAM J. Control Optim., vol. 51, pp. 2005–2035, 2013.
- G. Cybenko, “Approximation by superpositions of a sigmoidal function,” Mathematics of control, signals and systems, vol. 2, no. 4, pp. 303–314, 1989.
- J. Day, “A Runge-Kutta method for the numerical solution of the Goursat problem in hyperbolic partial differential equations,” The Computer Journal, vol. 9, no. 1, pp. 81–83, 1966.
- B. Deng, Y. Shin, L. Lu, Z. Zhang, and G. E. Karniadakis, “Approximation rates of deeponets for learning operators arising from advection–diffusion equations,” Neural Networks, vol. 153, pp. 411–426, 2022.
- J. Deutscher and J. Gabriel, “A backstepping approach to output regulation for coupled linear wave-ODE systems,” Automatica, vol. 123, p. 109338, 2021.
- F. Di Meglio, R. Vazquez, and M. Krstic, “Stabilization of a system of n+1𝑛1n+1italic_n + 1 coupled first-order hyperbolic linear PDEs with a single boundary input,” IEEE Trans. Aut. Contr., vol. 58, pp. 3097–3111, 2013.
- R. Holten, “Generalized Goursat problem,” Pacific Journal of Mathematics, vol. 12, pp. 207–224, 1962.
- K. Hornik, M. Stinchcombe, and H. White, “Multilayer feedforward networks are universal approximators,” Neural networks, vol. 2, no. 5, pp. 359–366, 1989.
- L. Jadachowsi, T. Meurer, and A. Kugi, “An efficient implementation of backstepping observers for time-varying parabolic PDEs,” IFAC Proceedings Volumes, vol. 45, pp. 798–803, 2012.
- S. Kerschbaum and J. Deutscher, “Backstepping control of coupled linear parabolic PDEs with space and time dependent coefficients,” IEEE Trans. Aut. Contr., vol. 65, no. 7, pp. 3060–3067, 2019.
- M. Krstic, L. Bhan, and Y. Shi, “Neural operators of backstepping controller and observer gain functions for reaction-diffusion PDEs,” arXiv preprint arXiv:2303.10506, 2023.
- M. Krstic and A. Smyshlyaev, “Backstepping boundary control for first-order hyperbolic PDEs and application to systems with actuator and sensor delays,” Systems & Control Letters, vol. 57, no. 9, pp. 750–758, 2008.
- Z. Li, N. Kovachki, K. Azizzadenesheli, et al., “Neural operator: Graph kernel network for partial differential equations,” arXiv preprint, vol. 2003.03485, 2020.
- Z. Li, N. Kovachki, K. Azizzadenesheli, B. Liu, K. Bhattacharya, A. Stuart, and A. Anandkumar, “Fourier neural operator for parametric partial differential equations,” arXiv preprint, vol. 2010.08895, 2020.
- L. Lu, P. Jin, and G. E. Karniadakis, “Deeponet: Learning nonlinear operators for identifying differential equations based on the universal approximation theorem of operators,” arXiv preprint, vol. 1910.03193, 2019.
- L. Lu, P. Jin, G. Pang, Z. Zhang, and G. E. Karniadakis, “Learning nonlinear operators via Deeponet based on the universal approximation theorem of operators,” Nature Machine Intelligence, vol. 3, no. 3, pp. 218–229, 2021.
- J. Qi, R. Vazquez, and M. Krstic, “Multi-agent deployment in 3-D via PDE control,” IEEE Transactions on Automatic Control, vol. 60, pp. 891–906, 2015.
- J. Qi, J. Zhang, and M. Krstic, “Neural operators for delay-compensating control of hyperbolic PIDEs,” arXiv preprint arXiv:2307.11436, 2023.
- A. Smyshlyaev, E. Cerpa, and M. Krstic, “Boundary stabilization of a 1-D wave equation with in-domain antidamping,” SIAM J. Control Optim., vol. 48, pp. 4014–4031, 2010.
- ——, “Closed-form boundary state feedbacks for a class of 1-D partial integro-differential equations,” IEEE Transactions on Automatic control, vol. 49, no. 12, pp. 2185–2202, 2004.
- R. Vazquez, G. Chen, J. Qiao, and M. Krstic, “The power series method to compute backstepping kernel gains: theory and practice,” in 2023 Conference on Decision and Control, 2023.
- R. Vazquez and M. Krstic, “Control of 1-D parabolic PDEs with Volterra nonlinearities — Part I: Design,” Automatica, vol. 44, pp. 2778–2790, 2008.
- ——, “Boundary observer for output-feedback stabilization of thermal convection loop,” IEEE Trans. Control Syst. Technol., vol. 18, pp. 789–797, 2010.
- ——, “Boundary control of reaction-diffusion PDEs on balls in spaces of arbitrary dimensions,” ESAIM:Control Optim. Calc. Var., vol. 22, no. 4, pp. 1078–1096, 2016.
- R. Vazquez, E. Trelat, and J.-M. Coron, “Control for fast and stable laminar-to-high-Reynolds-numbers transfer in a 2-D Navier-Stokes channel flow,” Disc. Cont. Dyn. Syst. Ser. B, vol. 10, pp. 925–956, 2008.
- S. Wang, M. Diagne, and M. Krstić, “Deep learning of delay-compensated backstepping for reaction-diffusion PDEs,” arXiv preprint arXiv:2308.10501, 2023.