Abstract
In modern science and engineering disciplines, data-driven discovery methods play a fundamental role in system modeling, as data serve as the external representations of the intrinsic mechanisms within systems. However, empirical data contaminated by process and measurement noise remain a significant obstacle for this type of modeling. In this study, we have developed a data-driven method capable of directly uncovering linear dynamical systems from noisy data. This method combines the Kalman smoothing and sparse Bayesian learning to decouple process and measurement noise under the expectation-maximization framework, presenting an analytical method for alternate state estimation and system identification. Furthermore, the discovered model explicitly characterizes the probability distribution of process and measurement noise, as they are essential for filtering, smoothing, and stochastic control. We have successfully applied the proposed algorithm to several simulation systems. Experimental results demonstrate its potential to enable linear dynamical system discovery in practical applications where noise-free data are intractable to capture.
Similar content being viewed by others
References
Bergen K J, Johnson P A, de Hoop M V, et al. Machine learning for data-driven discovery in solid Earth geoscience. Science, 2019, 363: eaau0323
Xie X, Samaei A, Guo J, et al. Data-driven discovery of dimensionless numbers and governing laws from scarce measurements. Nat Commun, 2022, 13: 7562
Floryan D, Graham M D. Data-driven discovery of intrinsic dynamics. Nat Mach Intell, 2022, 4: 1113–1120
Wang Y, Fang H, Jin J, et al. Data-driven discovery of stochastic differential equations. Engineering, 2022, 17: 244–252
Shen T, Dong Y L, He D X, et al. Online identification of time-varying dynamical systems for industrial robots based on sparse Bayesian learning. Sci China Tech Sci, 2022, 65: 386–395
Shen T, Qiao X, Dong Y, et al. Deep adaptive control with online identification for industrial robots. Sci China Tech Sci, 2022, 65: 2593–2604
Wu J, Li W, Xiong Z. Identification of robot dynamic model and joint frictions using a baseplate force sensor. Sci China Tech Sci, 2022, 65: 30–40
Li B, Liu H, Wang R. Data-driven sensor placement for efficient thermal field reconstruction. Sci China Tech Sci, 2021, 64: 1981–1994
Li H, Wu P, Zeng N, et al. A survey on parameter identification, state estimation and data analytics for lateral flow immunoassay: From systems science perspective. Int J Syst Sci, 2022, 53: 3556–3576
Li X, Feng S, Hou N, et al. Surface microseismic data denoising based on sparse autoencoder and Kalman filter. Syst Sci Control Eng, 2022, 10: 616–628
Gao T T, Yan G. Autonomous inference of complex network dynamics from incomplete and noisy data. Nat Comput Sci, 2022, 2: 160–168
Yuan Y, Tang X, Zhou W, et al. Data driven discovery of cyber physical systems. Nat Commun, 2019, 10: 4894
Aalto A, Viitasaari L, Ilmonen P, et al. Gene regulatory network inference from sparsely sampled noisy data. Nat Commun, 2020, 11: 3493
Schmidt M, Lipson H. Distilling free-form natural laws from experimental data. Science, 2009, 324: 81–85
Brunton S L, Proctor J L, Kutz J N. Discovering governing equations from data by sparse identification of nonlinear dynamical systems. Proc Natl Acad Sci USA, 2016, 113: 3932–3937
Ouala S, Nguyen D, Drumetz L, et al. Learning latent dynamics for partially observed chaotic systems. Chaos-An Interdiscip J Nonlinear Sci, 2020, 30: 103121
Lu P Y, Ariño Bernad J, Soljačiá M. Discovering sparse interpretable dynamics from partial observations. Commun Phys, 2022, 5: 206
Bakarji J, Tartakovsky D M. Data-driven discovery of coarse-grained equations. J Comput Phys, 2021, 434: 110219
Rudy S H, Brunton S L, Proctor J L, et al. Data-driven discovery of partial differential equations. Sci Adv, 2017, 3: e1602614
Raissi M, Karniadakis G E. Hidden physics models: Machine learning of nonlinear partial differential equations. J Comput Phys, 2018, 357: 125–141
Li X, Li L, Yue Z, et al. Sparse learning of partial differential equations with structured dictionary matrix. Chaos-An Interdiscip J Nonlinear Sci, 2019, 29: 043130
Jin J, Yuan Y, Goncalves J. High precision variational Bayesian inference of sparse linear networks. Automatica, 2020, 118: 109017
Jin J, Yuan Y, Gonçalves J. A full Bayesian approach to sparse network inference using heterogeneous datasets. IEEE Trans Automat Contr, 2020, 66: 3282–3288
Liu H, Shang Z, Ren Z, et al. Recovering unknown topology in a two-layer multiplex network: One layer infers the other layer. Sci China Tech Sci, 2022, 65: 1493–1505
Reinbold P A K, Grigoriev R O. Data-driven discovery of partial differential equation models with latent variables. Phys Rev E, 2019, 100: 022219
Kaheman K, Brunton S L, Kutz J N. Automatic differentiation to simultaneously identify nonlinear dynamics and extract noise probability distributions from data. Mach Learn-Sci Technol, 2022, 3: 015031
Chen Z, Liu Y, Sun H. Physics-informed learning of governing equations from scarce data. Nat Commun, 2021, 12: 6136
Sun F, Liu Y, Wang Q, et al. PiSL: Physics-informed Spline Learning for data-driven identification of nonlinear dynamical systems. Mech Syst Signal Process, 2023, 191: 110165
Wei B. Sparse dynamical system identification with simultaneous structural parameters and initial condition estimation. Chaos Solitons Fractals, 2022, 165: 112866
Schaeffer H. Learning partial differential equations via data discovery and sparse optimization. Proc R Soc A, 2017, 473: 20160446
Särkkä S, Svensson L. Bayesian Filtering and Smoothing. Cambridge: Cambridge University Press, 2023. 17
Shumway R H, Stoffer D S, Stoffer D S. Time Series Analysis and Its Applications. New York: Springer, 2000. 3
Barfoot T D. State Estimation for Robotics. Cambridge: Cambridge University Press, 2017
Song Y, Xu W, Niu L. Multiplicative Levy noise-induced transitions in gene expression. Sci China Tech Sci, 2022, 65: 1700–1709
Gibson S, Ninness B. Robust maximum-likelihood estimation of multivariable dynamic systems. Automatica, 2005, 41: 1667–1682
Einicke G A, Malos J T, Reid D C, et al. Riccati equation and EM algorithm convergence for inertial navigation alignment. IEEE Trans Signal Process, 2008, 57: 370–375
Einicke G A, Falco G, Malos J T. EM algorithm state matrix estimation for navigation. IEEE Signal Process Lett, 2010, 17: 437–440
Schön T B, Wills A, Ninness B. System identification of nonlinear state-space models. Automatica, 2011, 47: 39–49
Wang Y, Cheng C, Sun H, et al. Data augmentation-based statistical inference of diffusion processes. Chaos-An Interdiscip J Nonlinear Sci, 2023, 33: 033115
Wu C F J. On the convergence properties of the EM algorithm. Ann Statist, 1983, 11: 95–103
Mauroy A, Goncalves J. Koopman-based lifting techniques for nonlinear systems identification. IEEE Trans Automat Contr, 2019, 65: 2550–2565
Lusch B, Kutz J N, Brunton S L. Deep learning for universal linear embeddings of nonlinear dynamics. Nat Commun, 2018, 9: 4950
Author information
Authors and Affiliations
Corresponding author
Additional information
This work was supported by the National Natural Science Foundation of China (Grant No. 92167201).
Rights and permissions
About this article
Cite this article
Wang, Y., Yuan, Y., Fang, H. et al. Data-driven discovery of linear dynamical systems from noisy data. Sci. China Technol. Sci. 67, 121–129 (2024). https://doi.org/10.1007/s11431-023-2520-6
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11431-023-2520-6