Data-driven order reduction and velocity field reconstruction using neural networks: The case of a turbulent boundary layer

Giannopoulos, Antonios; Aider, Jean-Luc

doi:10.1063/5.0015870

Cited by 28 publications

(6 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…OF showed clear advantages to extract additional biophysical / chemical informations such as local vorticity or net polymerisation rates from speckle microscopy. Last but not least, OF algorithms have been also applied in experimental machine learning flow control experiments [14] as well as System Identification studies with quite satisfying results [17,16,13].…”

Section: Optical Flow For Pivmentioning

confidence: 99%

“…3(c) and Fig. 3(d) computed for different window sizes (16,32, and 64 pixels) for square interrogation windows with aspect ratios of 1.…”

Section: Steady Particle Images: Noise Response Analysismentioning

confidence: 99%

“…In the case of CC, the standard deviation decreases to a value of 0.3 for X = Y = 64 as the interrogation window increases. The case of OF is more interesting since the standard deviation peaks at 0.27 for KR = 8 and decreases to 0.2 for KR = [16,32]. Therefore, the numerical diffusion induced by OF decreases the sensitivity to the step response, which can turn out to be particularly important for turbulent flows analyses.…”

Section: Step-response Analysismentioning

confidence: 99%

See 2 more Smart Citations

On the optimal window size in optical flow and cross-correlation in particle image velocimetry: application to turbulent flows

et al. 2022

Self Cite

View full text Add to dashboard Cite

Two Particle Image Velocimetry (PIV) softwares applied to turbulent flows are compared. One is based on a standard Cross-Correlation (CC) algorithm and the other is based on an iterative multi-pyramid Optical Flow (OF) algorithm. First, still particle images are used to evaluate the cut-off frequency induced by each method. Then a step response analysis highlights the capabilities of each method to minimise the effect of unresolved velocity gradients. Two different benchmarks with various turbulent length-scales, down to the Taylor micro-scale, are then used to analyse the velocity spectra and the turbulent kinetic energy dissipation estimation. First, a synthetic PIV dataset of homogeneous isotropic turbulence is processed and compared with Direct Numerical Simulation (DNS) results. Then a grid turbulence wind tunnel experimental dataset is used to calculate velocity spectra and second-order structure functions, which are compared to Laser Doppler Velocimetry spectra.All these results point to the fact that, although OF is more diffusive and up to 5% less accurate than cross correlation, the numerical diffusion improves the calculation of sub-window unresolved gradients and allows for direct and more robust measurement of the onset of the viscous subrange in experimental turbulent flows.

show abstract

Section: Optical Flow For Pivmentioning

confidence: 99%

“…3(c) and Fig. 3(d) computed for different window sizes (16,32, and 64 pixels) for square interrogation windows with aspect ratios of 1.…”

Section: Steady Particle Images: Noise Response Analysismentioning

confidence: 99%

Section: Step-response Analysismentioning

confidence: 99%

See 1 more Smart Citation

On the optimal window size in optical flow and cross-correlation in particle image velocimetry: application to turbulent flows

et al. 2022

Self Cite

View full text Add to dashboard Cite

show abstract

“…Gautier et al 2015;Parezanović et al 2016;Li et al 2017), artificial neural networks (e.g. Ling, Kurzawski & Templeton 2016;Giannopoulos & Aider 2020;Ren, Hu & Tang 2020) and the explorative gradient method or EGM (e.g. Fan et al 2020a;Li et al 2022).…”

Section: Introductionmentioning

confidence: 99%

Artificial intelligence control of a low-drag Ahmed body using distributed jet arrays

2023

View full text Add to dashboard Cite

This work proposes a machine-learning or artificial intelligence (AI) control of a low-drag Ahmed body with a rear slant angle φ = 35° with a view to finding strategies for efficient drag reduction (DR). The Reynolds number Re investigated is 1.7 × 105 based on the square root of the body cross-sectional area. The control system comprises of five independently operated arrays of steady microjets blowing along the edges of the rear window and vertical base, twenty-six pressure taps on the rear end of the body and a controller based on an ant colony algorithm for unsupervised learning of a near-optimal control law. The cost function is designed such that both DR and control power input are considered. The learning process of the AI control discovers forcing that produces a DR up to 18 %, corresponding to a drag coefficient reduction of 0.06, greatly exceeding any previously reported DR for this body. Furthermore, the discovered forcings may provide alternative solutions, i.e. a tremendously increased control efficiency given a small sacrifice in DR. Extensive flow measurements performed with and without control indicate significant alterations in the flow structure around the body, such as flow separation over the rear window, recirculation bubbles and C-pillar vortices, which are linked to the pressure rise on the window and base. The physical mechanism for DR is unveiled, along with a conceptual model for the altered flow structure under the optimum control or biggest DR. This mechanism is further compared with that under the highest control efficiency.

show abstract

“…For the model reduction-based methods, the reconstruction task is explicitly decomposed into building reduced order model (ROM) and estimating low-dimensional coefficients. Earlier studies adopts traditional ROMs includes POD [5,6] and DMD [7] to generate reduced basis, the linear combination [8] of which is used to estimate the global field. Then, the combination coefficient is usually obtained by solving an optimization problem [9,10,11] or estimated by regression model.…”

Section: Introductionmentioning

confidence: 99%

RecFNO: a resolution-invariant flow and heat field reconstruction method from sparse observations via Fourier neural operator

Zhao¹,

Chen²,

Gong³

et al. 2023

Preprint

View full text Add to dashboard Cite

Perception of the full state is an essential technology to support the monitoring, analysis, and design of physical systems, one of whose challenges is to recover global field from sparse observations. Well-known for brilliant approximation ability, deep neural networks have been attractive to data-driven flow and heat field reconstruction studies. However, limited by network structure, existing researches mostly learn the reconstruction mapping in finite-dimensional space and has poor transferability to variable resolution of outputs. In this paper, we extend the new paradigm of neural operator and propose an end-to-end physical field reconstruction method with both excellent performance and mesh transferability named RecFNO. The proposed method aims to learn the mapping from sparse observations to flow and heat field in infinite-dimensional space, contributing to a more powerful nonlinear fitting capacity and resolution-invariant characteristic. Firstly, according to different usage scenarios, we develop three types of embeddings to model the sparse observation inputs: MLP, mask, and Voronoi embedding. The MLP embedding is propitious to more sparse input, while the others benefit from spatial information preservation and perform better with the increase of observation data. Then, we adopt stacked Fourier layers to reconstruct physical field in Fourier space that regularizes the overall recovered field by Fourier modes superposition. Benefiting from the operator in infinite-dimensional space, the proposed method obtains remarkable accuracy and better resolution transferability among meshes. The experiments conducted on fluid mechanics and thermology problems show that the proposed method outperforms existing POD-based and CNN-based methods in most cases and has the capacity to achieve zero-shot super-resolution.

show abstract

Data-driven order reduction and velocity field reconstruction using neural networks: The case of a turbulent boundary layer

Cited by 28 publications

References 39 publications

On the optimal window size in optical flow and cross-correlation in particle image velocimetry: application to turbulent flows

On the optimal window size in optical flow and cross-correlation in particle image velocimetry: application to turbulent flows

Artificial intelligence control of a low-drag Ahmed body using distributed jet arrays

RecFNO: a resolution-invariant flow and heat field reconstruction method from sparse observations via Fourier neural operator

Contact Info

Product

Resources

About