Leveraging Koopman operator and Deep Neural Net-works for Parameter Estimation and Future Prediction of Duffing oscillators

پذیرفته شده برای ارائه شفاهی ، صفحه 1-8 (8) اصل مقاله (1023.25 K)

کد مقاله : 1056-ISAV2023 (R1)

Yassin Riyazi؛ Navidreza Ghanbari؛ Arash Bahrami

School of Mechanical Engineering, College of Engineering, University of Tehran

The exploration of nonlinear dynamical systems holds significant importance across scientific and engineering disciplines, primarily for its applications in modeling real-world phenomena. Traditional methods employed for the analysis and prediction of the behavior of these systems typically involve intricate mathematical techniques and numerical simulations. This pa-per presents an innovative approach that combine the capabilities of the Koopman operator and deep neural networks to establish a linear representation of the Duffing oscillator. This newly developed methodology facilitates effective parameter estimation and the accurate prediction of the oscillator's future behavior. Moreover, the paper proposes a modified training procedure aimed at confining the Koopman operator to a linear layer within the neural network, as opposed to its application across the entire network. This synergy between the Koopman operator and deep neural networks not only simplifies the analysis of nonlinear systems but also paves the way for significant advancements in predictive modeling across diverse fields.

Koopman operator؛ Parameter estimation؛ Duffing oscillator؛ Nonlinear dynamical systems

Identification of Mechanical Systems

[1] Hamel, “Georg Duffing, Ingenieur: Erzwungene Schwingungen bei veränderlicher Eigenfrequenz und ihre 
technische Bedeutung. Sammlung Vieweg. Heft 41/42, Braunschweig 1918. VI+134 S,” ZAMM - J. Appl. Math. 
Mech. / Zeitschrift für Angew. Math. und Mech., vol. 1, no. 1, 1921.
[2] B. O. Koopman, “Hamiltonian Systems and Transformation in Hilbert Space,” Proc. Natl. Acad. Sci., vol. 17, no. 
5, 1931.
[3] G. E. Hinton and R. R. Salakhutdinov, “Reducing the dimensionality of data with neural networks,” Science (80-
. )., vol. 313, no. 5786, 2006.
[4] B. Lusch, J. N. Kutz, and S. L. Brunton, “Deep learning for universal linear embeddings of nonlinear dynamics,” 
Nat. Commun., vol. 9, no. 1, 2018.
[5] K. Champion, B. Lusch, J. Nathan Kutz, and S. L. Brunton, “Data-driven discovery of coordinates and governing 
equations,” Proc. Natl. Acad. Sci. U. S. A., vol. 116, no. 45, 2019.
[6] E. Kaiser, J. N. Kutz, and S. L. Brunton, “Sparse identification of nonlinear dynamics for model predictive control 
in the low-data limit,” Proc. R. Soc. A Math. Phys. Eng. Sci., vol. 474, no. 2219, 2018.
[7] A. Krizhevsky, I. Sutskever, and G. E. Hinton, “2012 AlexNet,” Adv. Neural Inf. Process. Syst., 2012.
[8] K. Simonyan and A. Zisserman, “Very deep convolutional networks for large-scale image recognition,” in 3rd 
International Conference on Learning Representations, ICLR 2015 - Conference Track Proceedings, 2015.
[9] M. Shafiq and Z. Gu, “Deep Residual Learning for Image Recognition: A Survey,” Applied Sciences 
(Switzerland), vol. 12, no. 18. 2022.
[10] S. L. Brunton, M. Budišić, E. Kaiser, and J. N. Kutz, “Modern Koopman Theory for Dynamical Systems,” SIAM 
Rev., vol. 64, no. 2, 2022.
[11] P. J. Schmid, “Dynamic mode decomposition of numerical and experimental data,” J. Fluid Mech., vol. 656, 
2010.
[12] C. Runge, “Ueber die numerische Auflösung von Differentialgleichungen,” Math. Ann., vol. 46, no. 2, 1895.
[13] C. Szegedy et al., “Going deeper with convolutions,” in Proceedings of the IEEE Computer Society Conference 
on Computer Vision and Pattern Recognition, 2015.
[14] Y. LeCun, L. Bottou, Y. Bengio, and P. Haffner, “Gradient-based learning applied to document recognition,” 
Proc. IEEE, vol. 86, no. 11, 1998.
[15] S. Hochreiter and J. Schmidhuber, “Long Short-Term Memory,” Neural Comput., vol. 9, no. 8, 1997.
[16] M. O. Williams, I. G. Kevrekidis, and C. W. Rowley, “A Data–Driven Approximation of the Koopman operator: 
Extending Dynamic Mode Decomposition,” J. Nonlinear Sci., vol. 25, no. 6, 2015.
[17] Q. Li, F. Dietrich, E. M. Bollt, and I. G. Kevrekidis, “Extended dynamic mode decomposition with dictionary 
learning: A data-driven adaptive spectral decomposition of the Koopman operator,” Chaos, vol. 27, no. 10, 2017