Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Global field reconstruction from sparse sensors with Voronoi tessellation-assisted deep learning

Nature Machine Intelligence volume 3, pages 945–951 (2021)Cite this article

Subjects

A preprint version of the article is available at arXiv.

Abstract

Achieving accurate and robust global situational awareness of a complex time-evolving field from a limited number of sensors has been a long-standing challenge. This reconstruction problem is especially difficult when sensors are sparsely positioned in a seemingly random or unorganized manner, which is often encountered in a range of scientific and engineering problems. Moreover, these sensors could be in motion and could become online or offline over time. The key leverage in addressing this scientific issue is the wealth of data accumulated from the sensors. As a solution to this problem, we propose a data-driven spatial field recovery technique founded on a structured grid-based deep-learning approach for arbitrary positioned sensors of any numbers. It should be noted that naive use of machine learning becomes prohibitively expensive for global field reconstruction and is furthermore not adaptable to an arbitrary number of sensors. In this work, we consider the use of Voronoi tessellation to obtain a structured-grid representation from sensor locations, enabling the computationally tractable use of convolutional neural networks. One of the central features of our method is its compatibility with deep learning-based super-resolution reconstruction techniques for structured sensor data that are established for image processing. The proposed reconstruction technique is demonstrated for unsteady wake flow, geophysical data and three-dimensional turbulence. The current framework is able to handle an arbitrary number of moving sensors and thereby overcomes a major limitation with existing reconstruction methods. Our technique opens a new pathway toward the practical use of neural networks for real-time global field estimation.

This is a preview of subscription content, access via your institution

Relevant articles

Open Access articles citing this article.

Access options

Rent or buy this article

Get just this article for as long as you need it

$39.95

Learn more

Prices may be subject to local taxes which are calculated during checkout

Fig. 1: Voronoi tessellation-assisted global data recovery from discrete sensor locations for a two-dimensional cylinder wake.
Fig. 2: Voronoi tessellation-assisted spatial data recovery of a two-dimensional cylinder wake with nsensor = 8 and 16.
Fig. 3: Comparison of spatial data recovery for NOAA sea surface temperature.
Fig. 4: Voronoi-based spatial data recovery of NOAA sea surface temperature.
Fig. 5: Voronoi tessellation-assisted data recovery of turbulent channel flow.
Fig. 6: Robustness of our reconstruction technique for noisy input for the example of turbulent channel flow.

Data availability

Training data used in this study are available on the Open Science Framework (https://doi.org/10.17605/OSF.IO/NKDZP).

Code availability

Sample codes for training our models are available on GitHub (https://github.com/kfukami/Voronoi-CNN and https://zenodo.org/badge/latestdoi/326312928).

References

  1. Manohar, K., Brunton, B. W., Kutz, J. N. & Brunton, S. L. Data-driven sparse sensor placement for reconstruction: demonstrating the benefits of exploiting known patterns. IEEE Control Syst. Mag. 38, 63–86 (2018).

    Article  MathSciNet  Google Scholar 

  2. Akiyama, K. et al. First M87 event horizon telescope results. III. Data processing and calibration. Astrophys. J. Lett. 875, L3 (2019).

    Article  Google Scholar 

  3. Alonso, M. T., López-Dekker, P. & Mallorquí, J. J. A novel strategy for radar imaging based on compressive sensing. IEEE Trans. Geosci. Remote Sens. 48, 4285–4295 (2010).

    Article  Google Scholar 

  4. Mishra, K. V., Kruger, A. & Krajewski, W. F. Compressed sensing applied to weather radar. In 2014 IEEE Geoscience and Remote Sensing Symposium 1832–1835 (IEEE, 2014).

  5. Fukami, K., Fukagata, K. & Taira, K. Assessment of supervised machine learning for fluid flows. Theor. Comp. Fluid Dyn. 34, 497–519 (2020).

    Article  MathSciNet  Google Scholar 

  6. Boisson, J. & Dubrulle, B. Three-dimensional magnetic field reconstruction in the VKS experiment through Galerkin transforms. New J. Phys. 13, 023037 (2011).

    Article  Google Scholar 

  7. Noack, B. R. & Eckelmann, H. A global stability analysis of the steady and periodic cylinder wake. J. Fluid Mech. 270, 297–330 (1994).

    Article  Google Scholar 

  8. Adrian, R. J. & Moin, P. Stochastic estimation of organized turbulent structure: homogeneous shear flow. J. Fluid Mech. 190, 531–559 (1988).

    Article  Google Scholar 

  9. Suzuki, T. & Hasegawa, Y. Estimation of turbulent channel flow at Reτ = 100 based on the wall measurement using a simple sequential approach. J. Fluid Mech. 830, 760–796 (2006).

    Article  MathSciNet  Google Scholar 

  10. Bui-Thanh, T., Damodaran, M. & Willcox, K. Aerodynamic Data Reconstruction and Inverse Design Using Proper Orthogonal Decomposition 42 (AIAA, 2004).

  11. LeCun, Y., Bengio, Y. & Hinton, G. Deep learning. Nature 521, 436–444 (2015).

    Article  Google Scholar 

  12. Brunton, S. L., Noack, B. R. & Koumoutsakos, P. Machine learning for fluid mechanics. Annu. Rev. Fluid Mech. 52, 477–508 (2020).

    Article  Google Scholar 

  13. LeCun, Y., Bottou, L., Bengio, Y. & Haffner, P. Gradient-based learning applied to document recognition. Proc. IEEE 86, 2278–2324 (1998).

    Article  Google Scholar 

  14. Fukami, K., Fukagata, K. & Taira, K. Super-resolution reconstruction of turbulent flows with machine learning. J. Fluid Mech. 870, 106–120 (2019).

    Article  MathSciNet  Google Scholar 

  15. Rumelhart, D. E., Hinton, G. E. & Williams, R. J. Learning representations by back-propagation errors. Nature 322, 533—536 (1986).

    MATH  Google Scholar 

  16. Wu, Z. et al. A comprehensive survey on graph neural networks. IEEE Trans. Neural Netw. Learn. Syst. 32, 4–24 (2020).

    Article  Google Scholar 

  17. Mallick, T., Balaprakash, P., Rask, E. & Macfarlane, J. Transfer learning with graph neural networks for short-term highway traffic forecasting. Preprint at https://arxiv.org/abs/2004.08038 (2020).

  18. Chai, X. et al. Deep learning for irregularly and regularly missing data reconstruction. Sci. Rep. 10, 1–18 (2020).

    Article  Google Scholar 

  19. Machicoane, N. et al. in Flowing Matter 177–209 (Springer, 2019).

  20. Voronoi, G. New applications of continuous parameters to the theory of quadratic forms. First thesis on some properties of perfect positive quadratic forms. J. Reine Angew. Math. 133, 97–178 (1908).

    Article  MathSciNet  Google Scholar 

  21. Aurenhammer, F. Voronoi diagrams-a survey of a fundamental geometric data structure. ACM Comput. Surv. 23, 345–405 (1991).

    Article  Google Scholar 

  22. Fukami, K., Fukagata, K. & Taira, K. Machine-learning-based spatio-temporal super resolution reconstruction of turbulent flows. J. Fluid Mech. 909, A9 (2021).

    Article  MathSciNet  Google Scholar 

  23. Nair, V. & Hinton, G. E. Rectified linear units improve restricted Boltzmann machines. Proc. 27th International Conference on Machine Learning 807–814 (2010).

  24. Kingma, D. P. & Ba, J. Adam: a method for stochastic optimization. Preprint at https://arxiv.org/abs/1412.6980 (2014).

  25. Prechelt, L. Automatic early stopping using cross validation: quantifying the criteria. Neural Netw. 11, 761–767 (1998).

    Article  Google Scholar 

  26. Brunton, S. L. & Kutz, J. N. Data-driven Science and Engineering: Machine Learning, Dynamical Systems, and Control (Cambridge Univ. Press, 2019).

  27. Taira, K. & Colonius, T. The immersed boundary method: a projection approach. J. Comput. Phys. 225, 2118–2137 (2007).

    Article  MathSciNet  Google Scholar 

  28. Colonius, T. & Taira, K. A fast immersed boundary method using a nullspace approach and multi-domain far-field boundary conditions. Comput. Methods Appl. Mech. Eng. 197, 2131–2146 (2008).

    Article  MathSciNet  Google Scholar 

  29. Fukagata, K., Kasagi, N. & Koumoutsakos, P. A theoretical prediction of friction drag reduction in turbulent flow by superhydrophobic surfaces. Phys. Fluids 18, 051703 (2006).

    Article  Google Scholar 

  30. Fukami, K., Nabae, Y., Kawai, K. & Fukagata, K. Synthetic turbulent inflow generator using machine learning. Phys. Rev. Fluids 4, 064603 (2019).

    Article  Google Scholar 

  31. Raissi, M., Perdikaris, P. & Karniadakis, G. E. Physics-informed neural networks: a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations. J. Comput. Phys. 378, 686–707 (2019).

    Article  MathSciNet  Google Scholar 

  32. Raissi, M., Yazdani, A. & Karniadakis, G. E. Hidden fluid mechanics: learning velocity and pressure fields from flow visualizations. Science 367, 1026–1030 (2020).

    Article  MathSciNet  Google Scholar 

  33. Hasegawa, K., Fukami, K., Murata, T. & Fukagata, K. Machine-learning-based reduced-order modeling for unsteady flows around bluff bodies of various shapes. Theor. Comput. Fluid Dyn. 34, 367–388 (2020).

    Article  MathSciNet  Google Scholar 

  34. Lee, S. & You, D. Data-driven prediction of unsteady flow fields over a circular cylinder using deep learning. J. Fluid Mech. 879, 217–254 (2019).

    Article  MathSciNet  Google Scholar 

  35. Du, Y. & Zaki, T. A. Evolutional deep neural network. Preprint at https://arxiv.org/abs/2103.09959 (2021).

  36. Morimoto, M., Fukami, K., Zhang, K. & Fukagata, K. Generalization techniques of neural networks for fluid flow estimation. Preprint at https://arxiv.org/abs/2011.11911 (2020).

  37. Hasegawa, K., Fukami, K., Murata, T. & Fukagata, K. CNN-LSTM based reduced order modeling of two-dimensional unsteady flows around a circular cylinder at different Reynolds numbers. Fluid Dyn. Res. 52, 065501 (2020).

    Article  MathSciNet  Google Scholar 

  38. Kim, H., Kim, J., Won, S. & Lee, C. Unsupervised deep learning for super-resolution reconstruction of turbulence. J. Fluid Mech. 910, A29 (2021).

    Article  MathSciNet  Google Scholar 

Download references

Acknowledgements

R.M. and N.R. were supported by the U.S. Department of Energy, Office of Science, Office of Advanced Scientific Computing Research, under contract DE-AC02-06CH11357. This research was funded in part and used resources of the Argonne Leadership Computing Facility, which is a DOE Office of Science User Facility supported under contract DE-AC02-06CH11357. Koji.F. thanks support from the Japan Society for the Promotion of Science (grant nos. 18H03758, 21H05007). K.T. acknowledges support from the US Air Force Office of Scientific Research (grant nos. FA9550-16-1-0650 and FA9550-21-1-0178) and the US Army Research Office (grant no. W911NF-19-1-0032).

Author information

Authors and Affiliations

  1. Department of Mechanical and Aerospace Engineering, University of California, Los Angeles, CA, USA

    Kai Fukami & Kunihiko Taira

  2. Department of Mechanical Engineering, Keio University, Minato, Japan

    Kai Fukami & Koji Fukagata

  3. Argonne Leadership Computing Facility, Argonne National Laboratory, Lemont, IL, USA

    Romit Maulik

  4. High Energy Physics Division, Argonne National Laboratory, Lemont, IL, USA

    Nesar Ramachandra

  5. CPS Division, Argonne National Laboratory, Lemont, IL, USA

    Nesar Ramachandra

Authors
  1. Kai Fukami
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Romit Maulik
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Nesar Ramachandra
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Koji Fukagata
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Kunihiko Taira
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

Kai.F., R.M and N.R. designed research. Kai.F. performed research. Kai.F. analysed data. Kai.F. and K.T. wrote the paper. Koji.F. and K.T. supervised. All authors reviewed the manuscript.

Corresponding author

Correspondence to Kai Fukami.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Peer review information Nature Machine Intelligence thanks Mickaël Bourgoin and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data

Extended Data Fig. 1 Velocity RMS of turbulent channel flow.

Root mean squared value of streamwise velocity fluctuation in turbulent channel flow.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Fukami, K., Maulik, R., Ramachandra, N. et al. Global field reconstruction from sparse sensors with Voronoi tessellation-assisted deep learning. Nat Mach Intell 3, 945–951 (2021). https://doi.org/10.1038/s42256-021-00402-2

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/s42256-021-00402-2

This article is cited by

Search

Advanced search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing