Abstract
Proper orthogonal decomposition (POD) has been widely used to extract dominant modes and structures from massive dynamic computational data to improve the understanding and discovery of the phenomena as well as to guide experimental design and control. This paper presents a framework and data mining technique that directly identifies the region of interest (ROI) from the POD modes and determines relevant feature for targeted visualization and learning. Two key elements in the procedure are described, including (1) POD to reduce data dimensions and to decouple the time-averaged and time-varying flow structures in high-fidelity Computational Fluid Dynamics (CFD) data with non-uniform grids, and (2) feature mining, including clustering-based data mining and filtering to detect both mean and unsteady flow features in the ROI. The rationale and benefits of our POD-compatible feature detection for fast scalable feature extraction are discussed. Case studies of vortex extraction are undertaken to validate the present approach. The POD accurately captures the characteristic flow structures and provides useful insight into the underlying flow phenomena. The feature mining module is capable of identifying key features in the ROI (3–10 % of the original data) for focused visualization, discovery, and learning.
Graphical Abstract
Similar content being viewed by others
References
Chong MS, Perry AE, Cantwell BJ (1990) A general classification of three-dimensional flow fields. Phys Fluids 2:408–420
Frederich O, Scouten J, Luchtenburg DM, Thiele F (2010) Large-scale dynamics in the flow around a finite cylinder with a ground plate. Fluid Dyn Res 43(1):015504
Haimes R (1999) Automated feature extraction from transient CFD simulations. In: Proceedings of the 7th Annual Conference of the CFD Society of Canada, Keynote Address. Halifax, NS
Heiberg E, Ebbers T, Wigstrom L, Karlsson M (2003) Three-dimensional flow characterization using vector pattern matching. Visual Com Graph IEEE Trans 9(3):313–319
Heiland RW, Baker MP, Tafti DK (2001) VisBench: a framework for remote data visualization and analysis. In: International Conference on Computational Science pp 718–727
Huera-Huarte FJ, Vernet A (2010) Vortex modes in the wake of an oscillating long flexible cylinder combining POD and fuzzy clustering. Exp Fluids 48(6):999–1013
Hunt JCR, Wray AA, Moin P (1988) Eddies, streams, and convergence zones in turbulent flows. vol 1. Center for Turbulence Research Report CTR-S88
Jeong J, Hussain F (1995) On the identification of a vortex. J Fluid Mech 285(1):69–94
Manhart M, Wengle H (1993) A spatiotemporal decomposition of a fully inhomogeneous turbulent flow field. Theoret Comput Fluid Dyn 5(4–5):223–242
Mitchell AM, Morton SA, Forsythe JR, Cummings RM (2006) Analysis of delta-wing vortical substructures using detached-eddy simulation. AIAA journal 44(5):964–972
Muld TW, Efraimsson G, Henningson DS (2012a) Flow structures around a high-speed train extracted using proper orthogonal decomposition and dynamic mode decomposition. Computers and Fluids
Muld TW, Efraimsson G, Henningson DS (2012b) Mode decomposition on surface-mounted cube. Flow Turbul Combust 88(3):279–310
Paik J, Escauriaza C, Sotiropoulos F (2007) On the bimodal dynamics of the turbulent horseshoe vortex system in a wing-body junction. Phys Fluids 19:045107
Perrin R, Braza M, Cid E, Cazin S, Chassaing P, Mockett C, Reimann T, Thiele F (2008) Coherent and turbulent process analysis in the flow past a circular cylinder at high Reynolds number. J Fluids Struct 24(8):1313–1325
Pobitzer A, Tutkun M, Andreassen O, Fuchs R, Peikert R, Hauser H (2011) Energy-scale aware feature extraction for flow visualization. Computer Graphics Forum 30(3):771–780
Post FH, Vrolijk B, Hauser H, Laramee RS, Doleisch H (2003) The state of the art in flow visualisation: feature extraction and tracking. Computer Graphics Forum 22(4):775–792
Thompson DS, Nair JS, Venkata SSD, Machiraju RK, Jiang M, Craciun G (2002) Physics-based feature mining for large data exploration. Comput Sci Eng 4(4):22–30
Xu C, Schuster E (2011) Model order reduction for high dimensional linear systems based on rank-1 incremental proper orthogonal decomposition. In: American Control Conference (ACC). IEEE pp 2975–2981
Acknowledgments
This research was sponsored by DoD/AFOSR under the contract number FA9550-12-C-0049. This article was cleared for public release by the 88th Air Base Wing at Wright-Patterson AFB, case number 88ABW-2013-0233.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Wang, Y., Qian, J., Song, H. et al. Feature extraction from massive, dynamic computational data based on proper orthogonal decomposition and feature mining. J Vis 17, 363–372 (2014). https://doi.org/10.1007/s12650-014-0214-5
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s12650-014-0214-5