- Physical Review Journals
All JournalsPhysics Magazine
Access Provided by Xinjiang University

Physical Review Fluids

Export citation

Export citation

Choose format for download:

Download Citation
  • Access by Xinjiang University

Gradient-free optimization of chaotic acoustics with reservoir computing

Francisco Huhn

Luca Magri*

  • Department of Engineering, University of Cambridge, Cambridge CB2 1PZ, United Kingdom

  • Aeronautics Department, Imperial College London, London SW7 2AZ, United Kingdom; Department of Engineering, University of Cambridge, Cambridge CB2 1PZ, United Kingdom; The Alan Turing Institute, London NW1 2DB, United Kingdom; and Institute of Advanced Study, TU Munich, 85748 Munich Garching, Germany (visiting)

  • *lm547@cam.ac.uk
PDF

Phys. Rev. Fluids 7, 014402 – Published 31 January, 2022

DOI: https://doi.org/10.1103/PhysRevFluids.7.014402

Abstract

Gradient-based optimization of chaotic acoustics is challenging for a threefold reason: (i) first-order perturbations grow exponentially in time; (ii) the statistics of the solution may have a slow convergence; and (iii) the time-averaged acoustic energy may physically have discontinuous variations, which means that the gradient does not exist for some design parameters. We develop a versatile optimization method, which finds the design parameters that minimize time-averaged acoustic cost functionals, and overcomes the three aforementioned challenges. The method is gradient-free, model-informed, and data-driven with reservoir computing based on echo state networks. First, we analyze the predictive capabilities of echo state networks in thermoacoustics both in the short- and long-time prediction of the dynamics. We find that both fully data-driven and model-informed architectures are able to learn the chaotic acoustic dynamics, both time-accurately and statistically. Informing the training with a physical reduced-order model with one acoustic mode markedly improves the accuracy and robustness of the echo state networks, while keeping the computational cost low. Echo state networks offer accurate predictions of the long-time dynamics, which would be otherwise expensive by integrating the governing equations to evaluate the time-averaged quantity to optimize. Second, we couple echo state networks with a Bayesian technique to explore the design thermoacoustic parameter space. The computational method is minimally intrusive because it requires only the initialization of the physical and hyperparameter optimizers. Third, we find the set of flame parameters that minimize the time-averaged acoustic energy of chaotic oscillations, which are caused by the positive feedback with a heat source, such as a flame in gas turbines or rocket motors. These oscillations are known as thermoacoustic oscillations. The optimal set of flame parameters is found with the same accuracy as brute-force grid search but with a convergence rate that is more than one order of magnitude faster. This work opens up new possibilities for nonintrusive (“hands-off”) optimization of chaotic systems, in which the cost of generating data, for example, from high-fidelity simulations and experiments, is high.

View figure in article

Physics Subject Headings (PhySH)

Article Text

References (77)

  1. L. Rayleigh, The explanation of certain acoustical phenomena, Nature (London) 18, 319 (1878).
  2. T. C. Lieuwen and V. Yang, Combustion Instabilities in Gas Turbine Engines: Operational Experience, Fundamental Mechanisms, and Modeling (American Institute of Aeronautics and Astronautics, Reston, VA, 2005).
  3. F. E. C. Culick, Unsteady Motions in Combustion Chambers for Propulsion Systems (RTO AGARDograph AG-AVT-039, North Atlantic Treaty Organization, 2006).
  4. A. P. Dowling and Y. Mahmoudi, Combustion noise, Proc. Combustion Inst. 35, 65 (2015).
  5. A. P. Dowling, Nonlinear self-excited oscillations of a ducted flame, J. Fluid Mech. 346, 271 (1997).
  6. M. P. Juniper and R. Sujith, Sensitivity and nonlinearity of thermoacoustic oscillations, Annu. Rev. Fluid Mech. 50, 661 (2018).
  7. L. Magri, Adjoint methods as design tools in thermoacoustics, Appl. Mech. Rev. 71, 020801 (2018).
  8. L. Magri and M. P. Juniper, Sensitivity analysis of a time-delayed thermo-acoustic system via an adjoint-based approach, J. Fluid Mech. 719, 183 (2013).
  9. L. Magri and M. Juniper, Global modes, receptivity, and sensitivity analysis of diffusion flames coupled with duct acoustics, J. Fluid Mech. 752, 237 (2014).
  10. A. Orchini and M. P. Juniper, Linear stability and adjoint sensitivity analysis of thermoacoustic networks with premixed flames, Combust. Flame 165, 97 (2015).
  11. G. A. Mensah and J. P. Moeck, Acoustic damper placement and tuning for annular combustors: An adjoint-based optimization study, J. Eng. Gas Turbines Power 139, 061501 (2017).
  12. G. A. Mensah, L. Magri, A. Orchini, and J. P. Moeck, Effects of asymmetry on thermoacoustic modes in annular combustors: A higher-order perturbation study, J. Eng. Gas Turbines Power 141, 041030 (2018).
  13. J. G. Aguilar and M. P. Juniper, Thermoacoustic stabilization of a longitudinal combustor using adjoint methods, Phys. Rev. Fluids 5, 083902 (2020).
  14. J. G. Aguilar and M. P. Juniper, Adjoint methods for elimination of thermoacoustic oscillations in a model annular combustor via small geometry modifications, in Turbo Expo: Power for Land, Sea, and Air, Vol. 51050 (American Society of Mechanical Engineers, New York, NY, 2018), p. V04AT04A054.
  15. F. Schaefer, L. Magri, and W. Polifke, A hybrid adjoint network model for thermoacoustic optimization, in Turbo Expo: Power for Land, Sea, and Air (American Society of Mechanical Engineers, New York, NY, 2021).
  16. A. P. Dowling, A kinematic model of a ducted flame, J. Fluid Mech. 394, 51 (1999).
  17. F. Huhn and L. Magri, Optimisation of chaotically perturbed acoustic limit cycles, Nonlin. Dynam. 100, 1641 (2020).
  18. P. Subramanian, S. Mariappan, R. I. Sujith, and P. Wahi, Bifurcation analysis of thermoacoustic instability in a horizontal Rijke tube, Inte. J. Spray Combust. Dynam. 2, 325 (2011).
  19. L. Kabiraj, R. I. Sujith, and P. Wahi, Bifurcations of self-excited ducted laminar premixed flames, J. Eng. Gas Turbines Power 134, 31502 (2011).
  20. H. Gotoda, H. Nikimoto, T. Miyano, and S. Tachibana, Dynamic properties of combustion instability in a lean premixed gas-turbine combustor, Chaos 21, 013124 (2011).
  21. H. Gotoda, T. Ikawa, K. Maki, and T. Miyano, Short-term prediction of dynamical behavior of flame front instability induced by radiative heat loss, Chaos 22, 033106 (2012).
  22. L. Kabiraj, A. Saurabh, P. Wahi, and R. I. Sujith, Route to chaos for combustion instability in ducted laminar premixed flames, Chaos 22, 023129 (2012).
  23. V. Nair, G. Thampi, and R. I. Sujith, Intermittency route to thermoacoustic instability in turbulent combustors, J. Fluid Mech. 756, 470 (2014).
  24. V. Nair and R. I. Sujith, A reduced-order model for the onset of combustion instability: Physical mechanisms for intermittency and precursors, Proc. Combustion Inst. press 35, 3193 (2015).
  25. K. Kashinath, I. C. Waugh, and M. P. Juniper, Nonlinear self-excited thermoacoustic oscillations of a ducted premixed flame: Bifurcations and routes to chaos, J. Fluid Mech. 761, 399 (2014).
  26. I. C. Waugh, K. Kashinath, and M. P. Juniper, Matrix-free continuation of limit cycles and their bifurcations for a ducted premixed flame, J. Fluid Mech. 759, 1 (2014).
  27. A. Orchini, S. J. Illingworth, and M. P. Juniper, Frequency domain and time domain analysis of thermoacoustic oscillations with wave-based acoustics, J. Fluid Mech. 775, 387 (2015).
  28. F. Huhn and L. Magri, Stability, sensitivity and optimisation of chaotic acoustic oscillations, J. Fluid Mech. 882, A24 (2020).
  29. E. N. Lorenz, Deterministic nonperiodic flow, J. Atmos. Sci. 20, 130 (1963).
  30. D. J. Lea, M. R. Allen, and T. W. Haine, Sensitivity analysis of the climate of a chaotic system, Tellus A: Dynam. Meteorol. Oceanogr. 52, 523 (2000).
  31. D. J. Lea, T. W. N. Haine, M. R. Allen, and J. A. Hansen, Sensitivity analysis of the climate of a chaotic ocean circulation model, Quart. J. R. Meteorol. Soc. 128, 2587 (2002).
  32. G. L. Eyink, T. W. Haine, and D. J. Lea, Ruelle's linear response formula, ensemble adjoint schemes and Lévy flights, Nonlinearity 17, 1867 (2004).
  33. J. Thuburn, Climate sensitivities via a Fokker-Planck adjoint approach, Quarterly J. R. Meteorological Soc. 131, 73 (2005).
  34. P. J. Blonigan and Q. Wang, Probability density adjoint for sensitivity analysis of the mean of chaos, J. Comput. Phys. 270, 660 (2014).
  35. D. Lasagna, Sensitivity analysis of chaotic systems using unstable periodic orbits, SIAM J. Appl. Dynamical Systems 17, 547 (2018).
  36. C. E. Leith, Climate response and fluctuation dissipation, J. Atmospheric Sci. 32, 2022 (1975).
  37. R. V. Abramov and A. J. Majda, Blended response algorithms for linear fluctuation-dissipation for complex nonlinear dynamical systems, Nonlinearity 20, 2793 (2007).
  38. R. V. Abramov and A. J. Majda, New approximations and tests of linear fluctuation-response for chaotic nonlinear forced-dissipative dynamical systems, J. Nonlin. Sci. 18, 303 (2008).
  39. Q. Wang, Forward and adjoint sensitivity computation of chaotic dynamical systems, J. Comput. Phys. 235, 1 (2013).
  40. Q. Wang, R. Hu, and P. Blonigan, Least squares shadowing sensitivity analysis of chaotic limit cycle oscillations, J. Comput. Phys. 267, 210 (2014).
  41. A. Ni and Q. Wang, Sensitivity analysis on chaotic dynamical systems by nonintrusive least squares shadowing (NILSS), J. Comput. Phys. 347, 56 (2017).
  42. P. J. Blonigan and Q. Wang, Multiple shooting shadowing for sensitivity analysis of chaotic dynamical systems, J. Comput. Phys. 354, 447 (2018).
  43. A. Ni and C. Talnikar, Adjoint sensitivity analysis on chaotic dynamical systems by nonintrusive least squares adjoint shadowing (NILSAS), J. Comput. Phys. 395, 690 (2019).
  44. N. Chandramoorthy and Q. Wang, A computable realization of Ruelle's formula for linear response of statistics in chaotic systems, arXiv:2002.04117.
  45. A. Ni, Fast linear response algorithm for differentiating stationary measures of chaos, arXiv:2009.00595.
  46. P. J. Blonigan, Adjoint sensitivity analysis of chaotic dynamical systems with nonintrusive least squares shadowing, J. Comput. Phys. 348, 803 (2017).
  47. A. Ni, Hyperbolicity, shadowing directions and sensitivity analysis of a turbulent three-dimensional flow, J. Fluid Mech. 863, 644 (2019).
  48. D. Ruelle, A review of linear response theory for general differentiable dynamical systems, Nonlinearity 22, 855 (2009).
  49. I. Goodfellow, Y. Bengio, and A. Courville, Deep Learning (MIT Press, Cambridge, MA, 2016), http://www.deeplearningbook.org.
  50. S. Hochreiter and J. Schmidhuber, Long short-term memory, Neural Comput. 9, 1735 (1997).
  51. K. Cho, B. van Merrienboer, C. Gulcehre, D. Bahdanau, F. Bougares, H. Schwenk, and Y. Bengio, Learning phrase representations using RNN encoder-decoder for statistical machine translation, arXiv:1406.1078.
  52. H. Jaeger and H. Haas, Harnessing nonlinearity: Predicting chaotic systems and saving energy in wireless communication, Science 304, 78 (2004).
  53. M. Lukoševičius, A Practical Guide to Applying Echo State Networks (Springer, Berlin, 2012), pp. 659–686.
  54. J. Pathak, A. Wikner, R. Fussell, S. Chandra, B. R. Hunt, M. Girvan, and E. Ott, Hybrid forecasting of chaotic processes: Using machine learning in conjunction with a knowledge-based model, Chaos: An Inte. J. Nonlin. Sci. 28, 041101 (2018).
  55. N. A. K. Doan, W. Polifke, and L. Magri, Physics-informed echo state networks for chaotic systems forecasting, in Proceedings of the International Conference on Computational Science (ICCS) (Springer International Publishing, Cham, 2019), pp. 192–198.
  56. Z. Lu, J. Pathak, B. Hunt, M. Girvan, R. Brockett, and E. Ott, Reservoir observers: Model-free inference of unmeasured variables in chaotic systems, Chaos 27, 041102 (2017).
  57. N. A. K. Doan, W. Polifke, and L. Magri, Learning hidden states in a chaotic system: A physics-informed echo state network approach, in Proceedings of the International Conference on Computational Science (ICCS) (Springer International Publishing, Cham, 2020), pp. 117–123.
  58. A. Racca and L. Magri, Automatic-differentiated physics-informed echo state network (API-ESN), Computational Science ICCS (Springer International Publishing, Cham, 2021).
  59. F. Huhn and L. Magri, Learning ergodic averages in chaotic systems, in Proceedings of the International Conference on Computational Science (ICCS) (Springer International Publishing, Cham, 2020), pp. 124–132.
  60. A. G. Hart, J. L. Hook, and J. H. P. Dawes, Echo state networks trained by Tikhonov least squares are l2(μ) approximators of ergodic dynamical systems, Physica D 421, 132882 (2021).
  61. M. P. Juniper, Triggering in the horizontal Rijke tube: Nonnormality, transient growth, and bypass transition, J. Fluid Mech. 667, 272 (2011).
  62. L. V. King, On the convection of heat from small cyclinders in a stream of fluid: Determination of the convection constants of small platinum wires with applications to hot-wire anemometry, Proc. R. Soc. 214, 373 (1914).
  63. M. A. Heckl, Active control of the noise from a Rijke tube, J. Sound Vib. 124, 117 (1988).
  64. W. Polifke, A. Poncet, C. O. Paschereit, and K. Döbbeling, Reconstruction of acoustic transfer matrices by instationary computational fluid dynamics, J. Sound Vib. 245, 483 (2001).
  65. A. Orchini, G. Rigas, and M. P. Juniper, Weakly nonlinear analysis of thermoacoustic bifurcations in the Rijke tube, J. Fluid Mech. 805, 523 (2016).
  66. K. Balasubramanian and R. I. Sujith, Nonnormality and nonlinearity in combustion-acoustic interaction in diffusion flames, J. Fluid Mech. 594, 29 (2008).
  67. L. D. Landau and E. M. Lifshitz, Fluid Mechanics, 2nd ed. (Pergamon Press, Oxford, UK, 1987).
  68. L. N. Trefethen, Spectral Methods in MATLAB, Vol. 10 (SIAM, Philadelphia, PA, 2000).
  69. C. A. Kennedy, M. H. Carpenter, and R. Lewis, Low-storage, explicit Runge-Kutta schemes for the compressible Navier-Stokes equations, Appl. Numer. Math. 35, 177 (2000).
  70. C. E. Rasmussen and C. K. I. Williams, Gaussian Processes for Machine Learning (MIT Press, Cambridge, MA, 2006).
  71. A. Racca and L. Magri, Robust optimization and validation of echo state networks for learning chaotic dynamics, Neural Networks 142, 252 (2021).
  72. M. Hoffman, E. Brochu, and N. de Freitas, Portfolio allocation for bayesian optimization, in Proceedings of the 27th Conference on Uncertainty in Artificial Intelligence (UAI) (AUAI Press, Arlington, VA, 2011), pp. 327–336.
  73. J. Yperman and T. Becker, Bayesian optimization of hyper-parameters in reservoir computing, arXiv:1611.05193.
  74. J. Reinier Maat, N. Gianniotis, and P. Protopapas, Efficient optimization of echo state networks for time series datasets, Proceedings of the International Joint Conference on Neural Networks (IJCNN) (IEEE, 2018), pp. 1–7.
  75. S. Kullback and R. A. Leibler, On information and sufficiency, Ann. Math. Stat. 22, 79 (1951).
  76. D. Scott, Multivariate Density Estimation: Theory, Practice, and Visualization (John Wiley & Sons, New York, NY, 1992).
  77. P. Virtanen, R. Gommers, T. E. Oliphant, M. Haberland, T. Reddy, D. Cournapeau, E. Burovski, P. Peterson, W. Weckesser, J. Bright, S. J. van der Walt, M. Brett, J. Wilson, K. J. Millman, N. Mayorov, A. R. J. Nelson, E. Jones, R. Kern, E. Larson, C. J. Carey et al., SciPy 1.0: Fundamental algorithms for scientific computing in Python, Nat. Methods 17, 261 (2020).

Outline

Information

Phys. Rev. Fluids 7, 014402– Published 31 January, 2022
  • Received 17 June 2021
  • Accepted 11 January 2022
Crossmark - Check for updates button

©2022 American Physical Society

Sign In to Your Journals Account

Filter

Filter

Article Lookup

Enter a citation