Model selection for hybrid dynamical systems via sparse regression

Niall M Mangan*, T. Askham, S. L. Brunton, J. N. Kutz, J. L. Proctor

*Corresponding author for this work

Research output: Contribution to journalArticle

3 Citations (Scopus)

Abstract

Hybrid systems are traditionally difficult to identify and analyse using classical dynamical systems theory. Moreover, recently developed model identification methodologies largely focus on identifying a single set of governing equations solely from measurement data. In this article, we develop a new methodology, Hybrid-Sparse Identification of Nonlinear Dynamics, which identifies separate nonlinear dynamical regimes, employs information theory to manage uncertainty and characterizes switching behaviour. Specifically, we use the nonlinear geometry of data collected from a complex system to construct a set of coordinates based on measurement data and augmented variables. Clustering the data in these measurement-based coordinates enables the identification of nonlinear hybrid systems. This methodology broadly empowers nonlinear system identification without constraining the data locally in time and has direct connections to hybrid systems theory. We demonstrate the success of this method on numerical examples including a mass-spring hopping model and an infectious disease model. Characterizing complex systems that switch between dynamic behaviours is integral to overcoming modern challenges such as eradication of infectious diseases, the design of efficient legged robots and the protection of cyber infrastructures.

Original languageEnglish (US)
Article number20180534
JournalProceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences
Volume475
Issue number2223
DOIs
StatePublished - Mar 1 2019

Fingerprint

Hybrid Dynamical Systems
Hybrid systems
Model Selection
dynamical systems
regression analysis
Dynamical systems
Regression
System theory
infectious diseases
methodology
complex systems
Hybrid Systems
Large scale systems
Identification (control systems)
Infectious Diseases
Systems Theory
information theory
system identification
Information theory
Methodology

Keywords

  • Data-driven discovery
  • Hybrid systems
  • Information criteria
  • Model selection
  • Nonlinear dynamics
  • Sparse regression

ASJC Scopus subject areas

  • Mathematics(all)
  • Engineering(all)
  • Physics and Astronomy(all)

Cite this

@article{84886e240cf949e7a7e4c3e4deea93d3,
title = "Model selection for hybrid dynamical systems via sparse regression",
abstract = "Hybrid systems are traditionally difficult to identify and analyse using classical dynamical systems theory. Moreover, recently developed model identification methodologies largely focus on identifying a single set of governing equations solely from measurement data. In this article, we develop a new methodology, Hybrid-Sparse Identification of Nonlinear Dynamics, which identifies separate nonlinear dynamical regimes, employs information theory to manage uncertainty and characterizes switching behaviour. Specifically, we use the nonlinear geometry of data collected from a complex system to construct a set of coordinates based on measurement data and augmented variables. Clustering the data in these measurement-based coordinates enables the identification of nonlinear hybrid systems. This methodology broadly empowers nonlinear system identification without constraining the data locally in time and has direct connections to hybrid systems theory. We demonstrate the success of this method on numerical examples including a mass-spring hopping model and an infectious disease model. Characterizing complex systems that switch between dynamic behaviours is integral to overcoming modern challenges such as eradication of infectious diseases, the design of efficient legged robots and the protection of cyber infrastructures.",
keywords = "Data-driven discovery, Hybrid systems, Information criteria, Model selection, Nonlinear dynamics, Sparse regression",
author = "Mangan, {Niall M} and T. Askham and Brunton, {S. L.} and Kutz, {J. N.} and Proctor, {J. L.}",
year = "2019",
month = "3",
day = "1",
doi = "10.1098/rspa.2018.0534",
language = "English (US)",
volume = "475",
journal = "Proceedings of The Royal Society of London, Series A: Mathematical and Physical Sciences",
issn = "0080-4630",
publisher = "Royal Society of London",
number = "2223",

}

Model selection for hybrid dynamical systems via sparse regression. / Mangan, Niall M; Askham, T.; Brunton, S. L.; Kutz, J. N.; Proctor, J. L.

In: Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, Vol. 475, No. 2223, 20180534, 01.03.2019.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Model selection for hybrid dynamical systems via sparse regression

AU - Mangan, Niall M

AU - Askham, T.

AU - Brunton, S. L.

AU - Kutz, J. N.

AU - Proctor, J. L.

PY - 2019/3/1

Y1 - 2019/3/1

N2 - Hybrid systems are traditionally difficult to identify and analyse using classical dynamical systems theory. Moreover, recently developed model identification methodologies largely focus on identifying a single set of governing equations solely from measurement data. In this article, we develop a new methodology, Hybrid-Sparse Identification of Nonlinear Dynamics, which identifies separate nonlinear dynamical regimes, employs information theory to manage uncertainty and characterizes switching behaviour. Specifically, we use the nonlinear geometry of data collected from a complex system to construct a set of coordinates based on measurement data and augmented variables. Clustering the data in these measurement-based coordinates enables the identification of nonlinear hybrid systems. This methodology broadly empowers nonlinear system identification without constraining the data locally in time and has direct connections to hybrid systems theory. We demonstrate the success of this method on numerical examples including a mass-spring hopping model and an infectious disease model. Characterizing complex systems that switch between dynamic behaviours is integral to overcoming modern challenges such as eradication of infectious diseases, the design of efficient legged robots and the protection of cyber infrastructures.

AB - Hybrid systems are traditionally difficult to identify and analyse using classical dynamical systems theory. Moreover, recently developed model identification methodologies largely focus on identifying a single set of governing equations solely from measurement data. In this article, we develop a new methodology, Hybrid-Sparse Identification of Nonlinear Dynamics, which identifies separate nonlinear dynamical regimes, employs information theory to manage uncertainty and characterizes switching behaviour. Specifically, we use the nonlinear geometry of data collected from a complex system to construct a set of coordinates based on measurement data and augmented variables. Clustering the data in these measurement-based coordinates enables the identification of nonlinear hybrid systems. This methodology broadly empowers nonlinear system identification without constraining the data locally in time and has direct connections to hybrid systems theory. We demonstrate the success of this method on numerical examples including a mass-spring hopping model and an infectious disease model. Characterizing complex systems that switch between dynamic behaviours is integral to overcoming modern challenges such as eradication of infectious diseases, the design of efficient legged robots and the protection of cyber infrastructures.

KW - Data-driven discovery

KW - Hybrid systems

KW - Information criteria

KW - Model selection

KW - Nonlinear dynamics

KW - Sparse regression

UR - http://www.scopus.com/inward/record.url?scp=85064262230&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85064262230&partnerID=8YFLogxK

U2 - 10.1098/rspa.2018.0534

DO - 10.1098/rspa.2018.0534

M3 - Article

VL - 475

JO - Proceedings of The Royal Society of London, Series A: Mathematical and Physical Sciences

JF - Proceedings of The Royal Society of London, Series A: Mathematical and Physical Sciences

SN - 0080-4630

IS - 2223

M1 - 20180534

ER -