Sparse identification of nonlinear dynamics for model predictive control in the low-data limit

Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences - Tập 474 Số 2219 - Trang 20180335 - 2018
Eurika Kaiser1, J. Nathan Kutz2, Steven L. Brunton1
1Department of Mechanical Engineering, University of Washington, Seattle, WA, 98195
2Department of Applied Mathematics, University of Washington, Seattle, WA 98195

Tóm tắt

Data-driven discovery of dynamics via machine learning is pushing the frontiers of modelling and control efforts, providing a tremendous opportunity to extend the reach of model predictive control (MPC). However, many leading methods in machine learning, such as neural networks (NN), require large volumes of training data, may not be interpretable, do not easily include known constraints and symmetries, and may not generalize beyond the attractor where models are trained. These factors limit their use for the online identification of a model in the low-data limit, for example following an abrupt change to the system dynamics. In this work, we extend the recent sparse identification of nonlinear dynamics (SINDY) modelling procedure to include the effects of actuation and demonstrate the ability of these models to enhance the performance of MPC, based on limited, noisy data. SINDY models are parsimonious, identifying the fewest terms in the model needed to explain the data, making them interpretable and generalizable. We show that the resulting SINDY-MPC framework has higher performance, requires significantly less data, and is more computationally efficient and robust to noise than NN models, making it viable for online training and execution in response to rapid system changes. SINDY-MPC also shows improved performance over linear data-driven models, although linear models may provide a stopgap until enough data is available for SINDY. SINDY-MPC is demonstrated on a variety of dynamical systems with different challenges, including the chaotic Lorenz system, a simple model for flight control of an F8 aircraft, and an HIV model incorporating drug treatment.

Từ khóa


Tài liệu tham khảo

10.1137/1.9781611974508

10.1073/pnas.1517384113

Allgöwer F Badgwell TA Qin JS Rawlings JB Wright SJ. 1999 Nonlinear predictive control and moving horizon estimation: an introductory overview. In Advances in control: Highlights of ECC'99 (ed. PM Frank) pp. 391–449. London UK: Springer.

Camacho EF, 2013, Model predictive control

Skogestad S, 2005, Multivariable feedback control: analysis and design, 2

10.1007/978-1-4757-3290-0

10.1016/0005-1098(89)90002-2

10.1016/S0098-1354(98)00301-9

10.1007/s12555-011-0300-6

10.1016/j.automatica.2014.10.128

10.2514/1.G002507

10.1115/1.4031175

10.2514/3.20031

10.1016/0005-1098(76)90080-7

10.1093/imamci/1.3.243

Maner BR Doyle FJ Ogunnaike BA Pearson RK. 1994 A nonlinear model predictive control scheme using second order Volterra models. In American Control Conference Baltimore MD 29 June–1 July vol. 3 pp. 3253–3257. Piscataway NJ: IEEE.

10.1007/BF02532251

10.1002/9781118535561

10.1109/MASSP.1987.1165576

10.1109/37.466261

10.1109/TNNLS.2015.2411671

10.1016/j.compchemeng.2007.05.002

10.1007/s00332-015-9258-5

Korda M Mezić I. 2016 Linear predictors for nonlinear dynamical systems: Koopman operator meets model predictive control. (http://arxiv.org/abs/1611.03537)

Peitz S Schäfer K Ober-Blöbaum S Eckstein J Köhler U Dellnitz M. 2016 A multiobjective MPC approach for autonomously driven electric vehicles. (http://arxiv.org/abs/1610.08777)

10.1016/j.conengprac.2008.05.005

Zhang T Kahn G Levine S Abbeel P. 2016 Learning deep control policies for autonomous aerial vehicles with MPC-guided policy search. In IEEE International Conference on Robotics and Automation Stockholm Sweden 16–21 May pp. 528–535. Piscataway NJ: IEEE.

10.1073/pnas.0609476104

10.1126/science.1165893

10.1063/1.5027470

10.1017/jfm.2017.823

10.1137/15M1013857

10.1109/TAC.2005.852557

10.1109/TAC.2014.2351851

10.1016/j.automatica.2014.10.017

Pan W Yuan Y Gonçalves J Stan GB. 2012 Reconstruction of arbitrary biochemical reaction networks: a compressive sensing approach. In 2012 IEEE 51st Annual Conference on Decision and Control (CDC) Maui HI 10–13 December pp. 2334–2339. Piscataway NJ: IEEE.

10.1016/j.automatica.2015.06.003

10.1109/TAC.2015.2426291

Nelles O, 2013, Nonlinear system identification: from classical approaches to neural networks and fuzzy models

10.1016/j.automatica.2014.01.001

10.1016/0167-2789(92)90242-F

10.5402/2011/164564

10.1111/j.2517-6161.1996.tb02080.x

10.1109/TIT.2005.864420

Su W Bogdan M Candès EJ. 2016 False discoveries occur early on the Lasso path. (http://arxiv.org/abs/1511.01957)

Zhang L Schaeffer H. 2018 On the convergence of the SINDy algorithm. (http://arxiv.org/abs/1805.06445)

Zheng P Askham T Brunton SL Kutz JN Aravkin AY. 2018 A unified framework for sparse relaxed regularized regression: SR3. (http://arxiv.org/abs/1807.05411)

Gevers M Bazanella AS Coutinho DF Dasgupta S. 2013 Identifiability and excitation of polynomial systems. In 52nd IEEE Conference Decision and Control (CDC) Florence Italy 10–13 December pp. 4278–4283. IEEE.

10.1016/j.automatica.2017.01.029

10.1098/rspa.2017.0009

10.1126/sciadv.1602614

10.1098/rspa.2016.0446

10.1038/s41467-017-00030-8

Tran G Ward R. 2016 Exact recovery of chaotic systems from highly corrupted data. (http://arxiv.org/abs/1607.01067)

10.1103/PhysRevE.96.023302

10.1017/S0022112009992059

10.1017/S0022112010001217

10.3934/jcd.2014.1.391

10.1017/CBO9780511919701

Johansen TA, 2011, Introduction to nonlinear model predictive control and moving horizon estimation, Selected topics on constrained and nonlinear control

10.1016/j.techfore.2002.11.001

10.1216/rmjm/1181072471

10.1016/0893-6080(89)90020-8

10.1109/TIT.1970.1054411

10.1175/1520-0469(1963)020<0130:DNF>2.0.CO;2

10.1016/0005-1098(77)90070-X

10.1016/j.sysconle.2004.05.008

10.1109/TII.2012.2205582

10.1137/S0036144598335107

10.1016/j.jtbi.2005.05.004

10.1006/jtbi.2001.2426

10.1137/130949282

10.1103/PhysRevE.92.033304

10.1017/jfm.2018.147

10.1007/BFb0091924

10.1007/s11071-005-2824-x

10.1063/1.4772195

Kaiser E Kutz JN Brunton SL. 2017 Data-driven discovery of Koopman eigenfunctions for control. (http://arxiv.org/abs/1707.01146).

10.1109/TMBMC.2016.2633265

Thông tin
Thông tin xuất bản

Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences

Tập 474 Số 2219

20180335

Thông tin tác giả