Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
125 tokens/sec
GPT-4o
47 tokens/sec
Gemini 2.5 Pro Pro
43 tokens/sec
o3 Pro
4 tokens/sec
GPT-4.1 Pro
47 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

A neural network-based approach to hybrid systems identification for control (2404.01814v2)

Published 2 Apr 2024 in eess.SY, cs.LG, cs.SY, and math.OC

Abstract: We consider the problem of designing a machine learning-based model of an unknown dynamical system from a finite number of (state-input)-successor state data points, such that the model obtained is also suitable for optimal control design. We adopt a neural network (NN) architecture that, once suitably trained, yields a hybrid system with continuous piecewise-affine (PWA) dynamics that is differentiable with respect to the network's parameters, thereby enabling the use of derivative-based training procedures. We show that a careful choice of our NN's weights produces a hybrid system model with structural properties that are highly favorable when used as part of a finite horizon optimal control problem (OCP). Specifically, we rely on available results to establish that optimal solutions with strong local optimality guarantees can be computed via nonlinear programming (NLP), in contrast to classical OCPs for general hybrid systems which typically require mixed-integer optimization. Besides being well-suited for optimal control design, numerical simulations illustrate that our NN-based technique enjoys very similar performance to state-of-the-art system identification methods for hybrid systems and it is competitive on nonlinear benchmarks.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (63)
  1. Differentiable MPC for end-to-end planning and control. Advances in Neural Information Processing Systems, 31, 2018.
  2. B. Amos and J. Z. Kolter. OptNet: Differentiable optimization as a layer in neural networks. In International Conference on Machine Learning, pages 136–145. PMLR, 2017.
  3. Deep convolutional networks in system identification. In 2019 IEEE 58th Conference on Decision and Control (CDC), pages 3670–3676, 2019.
  4. Control for societal-scale challenges: Road map 2030, 2023. [Online]. Available: https://ieeecss.org/control-societal-scale-challenges-roadmap-2030.
  5. R. Batruni. A multilayer neural network with piecewise-linear structure and back-propagation learning. IEEE Transactions on Neural Networks, 2(3):395–403, 1991.
  6. A. Bemporad. A piecewise linear regression and classification algorithm with application to learning and model predictive control of hybrid systems. IEEE Transactions on Automatic Control, 68(6):3194–3209, 2023.
  7. A. Bemporad and M. Morari. Control of systems integrating logic, dynamics, and constraints. Automatica, 35(3):407–427, 1999.
  8. Convex analysis and optimization, volume 1. Athena Scientific, 2003.
  9. Efficient and modular implicit differentiation. In Advances in Neural Information Processing Systems, volume 35, pages 5230–5242, 2022.
  10. JAX: composable transformations of Python+NumPy programs, 2018.
  11. Identification of hybrid and linear parameter varying models via recursive piecewise affine regression and discrimination. In 2016 European Control Conference (ECC), pages 2632–2637. IEEE, 2016.
  12. V. Breschi and M. Mejari. Shrinkage strategies for structure selection and identification of piecewise affine models. In 2020 59th IEEE Conference on Decision and Control (CDC), pages 1626–1631. IEEE, 2020.
  13. Piecewise affine regression via recursive multiple least squares and multicategory discrimination. Automatica, 73:155–162, 2016.
  14. Y. Chen and M. Florian. The nonlinear bilevel programming problem: Formulations, regularity and optimality conditions. Optimization, 32(3):193–209, 1995.
  15. Constructive neural networks with piecewise interpolation capabilities for function approximations. IEEE Transactions on Neural Networks, 5(6):936–944, 1994.
  16. Canonical piecewise-linear representation. IEEE Transactions on Circuits and Systems, 35(1):101–111, 1988.
  17. Switched and piecewise affine systems, page 87–138. Cambridge University Press, 2009.
  18. End-to-end differentiable physics for learning and control. Advances in Neural Information Processing Systems, 31, 2018.
  19. F. Fabiani and P. J. Goulart. Reliably-stabilizing piecewise-affine neural network controllers. IEEE Transactions on Automatic Control, 68(9):5201–5215, 2023.
  20. F. Fabiani and P. J. Goulart. Robust stabilization of polytopic systems via fast and reliable neural network-based approximations. International Journal of Robust and Nonlinear Control, 2024. (In press).
  21. M. Fält and P. Giselsson. System identification for hybrid systems using neural networks. arXiv preprint arXiv:1911.12663, 2019.
  22. On the adaptation of recurrent neural networks for system identification. Automatica, 155:111092, 2023.
  23. M. Forgione and D. Piga. Model structures and fitting criteria for system identification with neural networks. In 2020 IEEE 14th International Conference on Application of Information and Communication Technologies (AICT), pages 1–6, 2020.
  24. M. Forgione and D. Piga. Continuous-time system identification with neural networks: Model structures and fitting criteria. European Journal of Control, 59:69–81, 2021.
  25. A note on the group lasso and a sparse group lasso. arXiv preprint arXiv:1001.0736, 2010.
  26. A locally-biased form of the DIRECT algorithm. Journal of Global Optimization, 21:27–37, 2001.
  27. A new algorithm for learning in piecewise-linear neural networks. Neural Networks, 13(4):485–505, 2000.
  28. Learning queuing networks by recurrent neural networks. In Proceedings of the ACM/SPEC International Conference on Performance Engineering, pages 56–66, 2020.
  29. A survey on switched and piecewise affine system identification. IFAC Proceedings Volumes, 45(16):344–355, 2012. 16th IFAC Symposium on System Identification.
  30. Deep Learning. MIT Press, 2016.
  31. Neural network design. PWS Publishing Co., 1997.
  32. A sequential convex programming approach to solving quadratic programs and optimal control problems with linear complementarity constraints. IEEE Control Systems Letters, 6:536–541, 2021.
  33. M. Heemels and B. Brogliato. The complementarity class of hybrid dynamical systems. European Journal of Control, 9(2-3):322–360, 2003.
  34. Modelling, well-posedness, and stability of switched electrical networks. In International Workshop on Hybrid Systems: Computation and Control, pages 249–266. Springer, 2003.
  35. Equivalence of hybrid dynamical models. Automatica, 37(7):1085–1091, 2001.
  36. Every continuous piecewise affine function can be obtained by solving a parametric linear program. In 2013 European Control Conference (ECC), pages 2657–2662. IEEE, 2013.
  37. Inverse parametric optimization with an application to hybrid system control. IEEE Transactions on Automatic Control, 60(4):1064–1069, 2014.
  38. Strong stationarity conditions for optimal control of hybrid systems. IEEE Transactions on Automatic Control, 62(9):4512–4526, 2017.
  39. A. C. Hindmarsh. ODEPACK, a systemized collection of ODE solvers. Scientific computing, 1983.
  40. Computationally efficient predictive control based on ANN state-space models. In 2023 62nd IEEE Conference on Decision and Control (CDC), pages 6336–6341. IEEE, 2023.
  41. Multilayer feedforward networks are universal approximators. Neural Networks, 2(5):359–366, 1989.
  42. D. R. Hush and B. Horne. Efficient algorithms for function approximation with piecewise linear sigmoidal networks. IEEE Transactions on Neural Networks, 9(6):1129–1141, 1998.
  43. MathWorks Inc. Control of a Two-Tank System. [Online]. Available: https://it.mathworks.com/help/robust/ug/control-of-a-two-tank-system.html.
  44. Learning linear complementarity systems. In Learning for Dynamics and Control Conference, pages 1137–1149. PMLR, 2022.
  45. D. Kraft. A software package for sequential quadratic programming. Forschungsbericht-Deutsche Forschungs-und Versuchsanstalt fur Luft-und Raumfahrt, 1988.
  46. F. Lauer. On the complexity of piecewise affine system identification. Automatica, 62:148–153, 2015.
  47. Learning particle dynamics for manipulating rigid bodies, deformable objects, and fluids. In International Conference on Learning Representations, 2018.
  48. G. P. Liu and V. Kadirkamanathan. Predictive control for non-linear systems using neural networks. International Journal of Control, 71(6):1119–1132, 1998.
  49. Mathematical programs with equilibrium constraints. Cambridge University Press, 1996.
  50. A. Magnani and S. P. Boyd. Convex piecewise-linear fitting. Optimization and Engineering, 10:1–17, 2009.
  51. D. Masti and A. Bemporad. Learning nonlinear state–space models using autoencoders. Automatica, 129:109666, 2021.
  52. Integrated neural networks for nonlinear continuous-time system identification. IEEE Control Systems Letters, 4(4):851–856, 2020.
  53. M. D. Mejari. Towards automated data-driven modeling of linear parameter-varying systems. PhD thesis, IMT School for Advanced Studies Lucca, 2018.
  54. Identification of hybrid systems: A tutorial. European Journal of Control, 13(2):242–260, 2007.
  55. Kernel methods in system identification, machine learning and function estimation: A survey. Automatica, 50(3):657–682, 2014.
  56. D. Ralph. Mathematical programs with complementarity constraints in traffic and telecommunications networks. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 366(1872):1973–1987, 2008.
  57. Identification of piecewise affine systems via mixed-integer programming. Automatica, 40(1):37–50, 2004.
  58. J. Schoukens and L. Ljung. Nonlinear system identification: A user-oriented road map. IEEE Control Systems Magazine, 39(6):28–99, 2019.
  59. M. Sugiyama. Introduction to statistical machine learning. Morgan Kaufmann, 2015.
  60. SciPy 1.0: Fundamental Algorithms for Scientific Computing in Python. Nature Methods, 17:261–272, 2020.
  61. A. Wächter and L. T. Biegler. On the implementation of an interior-point filter line-search algorithm for large-scale nonlinear programming. Mathematical Programming, 106:25–57, 2006.
  62. P. J. Werbos. Neural networks for control and system identification. In Proceedings of the 28th IEEE Conference on Decision and Control,, pages 260–265, 1989.
  63. Modeling design and control problems involving neural network surrogates. Computational Optimization and Applications, pages 1–42, 2022.

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now