A relaxed model selection method for Duffing oscillator identification

This study presents a relaxed model selection procedure based on the sparse regression system identification method for Duffing oscillator identification. A two-stage relaxation procedure is presented. In a first stage an elastic net optimizer underpins the sparse regression that enables to explore the possible models. The subsequent stage employs the Akaike information criteria that employs the one standard error rule (1-SE) to select the appropriate model. We study the effect of relaxation in both stages, i.e. providing more exploration capabilities in the regression and the selection, by applying the two-stage methodology on experimental data collected on a mechanical Duffing oscillator setup. Our analysis shows that relaxation is advantageous when dealing with noisy experimental data and allows to find model structures and associated parameter values in the mechatronic Duffing oscillator. Results show that relaxation is pivotal when dealing with noisy experimental data and that the methodology possesses the capability to find model structures and associated parameter values of a Duffing oscillator. The presented results can be of benefit when identifying the system behavior of other mechatronic systems that exhibit complex, chaotic behavior that is difficult to model starting from first principles.

[1]  Steven L Brunton,et al.  Sparse identification of nonlinear dynamics for rapid model recovery. , 2018, Chaos.

[2]  R. Tibshirani,et al.  On the “degrees of freedom” of the lasso , 2007, 0712.0881.

[3]  Cristian R. Rojas,et al.  Sparse Iterative Learning Control with Application to a Wafer Stage: Achieving Performance, Resource Efficiency, and Task Flexibility , 2017, ArXiv.

[4]  David Zhang,et al.  A Survey of Sparse Representation: Algorithms and Applications , 2015, IEEE Access.

[5]  Sophie M. Fosson,et al.  Centralized and Distributed Online Learning for Sparse Time-Varying Optimization , 2020, IEEE Transactions on Automatic Control.

[6]  Tor Arne Johansen,et al.  On Tikhonov regularization, bias and variance in nonlinear system identification , 1997, Autom..

[7]  Lennart Ljung,et al.  Perspectives on system identification , 2010, Annu. Rev. Control..

[8]  David Akopian,et al.  A Joint Indoor WLAN Localization and Outlier Detection Scheme Using LASSO and Elastic-Net Optimization Techniques , 2016, IEEE Transactions on Mobile Computing.

[9]  H. Zou,et al.  Regularization and variable selection via the elastic net , 2005 .

[10]  David M. Blei,et al.  Exploiting Covariate Similarity in Sparse Regression via the Pairwise Elastic Net , 2010, AISTATS.

[11]  R. Tibshirani,et al.  Least angle regression , 2004, math/0406456.

[12]  R. Tibshirani.  Regression Shrinkage and Selection via the Lasso , 1996 .

[13]  Gavin C. Cawley,et al.  On Over-fitting in Model Selection and Subsequent Selection Bias in Performance Evaluation , 2010, J. Mach. Learn. Res..

[14]  Steven L. Brunton,et al.  Data-driven discovery of partial differential equations , 2016, Science Advances.

[15]  Wei-Yin Loh,et al.  Classification and regression trees , 2011, WIREs Data Mining Knowl. Discov..

[16]  Rik Pintelon,et al.  Real-time integration and differentiation of analog signals by means of digital filtering , 1990 .

[17]  J N Kutz,et al.  Model selection for dynamical systems via sparse regression and information criteria , 2017, Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences.

[18]  Rasmus Larsen,et al.  SpaSM: A MATLAB Toolbox for Sparse Statistical Modeling , 2018 .

[19]  Luc Dupre,et al.  Sparse Identification of Nonlinear Duffing Oscillator From Measurement Data , 2018 .

[20]  H. Akaike,et al.  Information Theory and an Extension of the Maximum Likelihood Principle , 1973 .

[21]  S. Brunton,et al.  Discovering governing equations from data by sparse identification of nonlinear dynamical systems , 2015, Proceedings of the National Academy of Sciences.