Fourier-Feature-Embedded Complex-Valued Physics-Informed Convolutional Residual Networks for Sparse and Gappy Noisy Flow-Field Reconstruction
Abstract
Reconstructing flow fields from sparse and noisy measurements becomes particularly challenging when data are completely absent from dynamically critical regions, such as the vortex-formation region of a wake. Conventional interpolation and numerical reconstruction methods generally require sufficient spatial or boundary information, while physics-informed neural networks (PINNs) can also face difficulties when sparse and noisy observations provide little or no supervision in the region of interest. To address this gap, we propose a Fourier-feature-embedded complex-valued physics-informed convolutional residual network (FCV-PICRN) for sparse, gappy, and noisy flow-field reconstruction. The framework integrates Fourier features for oscillatory and multiscale structures, complex-valued representations for coupled amplitude--phase learning, and convolutional residual blocks constrained by the Navier--Stokes equations. The method is first verified using the Taylor--Green vortex at Re=100 and then evaluated for the cylinder wake at Re=100. Three severe missing-data configurations are imposed in the dynamically sensitive wake: a large hole, a large truncation, and a plus-shaped gap extending in both streamwise and transverse directions. These gaps are further combined with additive Gaussian and multiplicative speckle noise. Despite having no supervised velocity data inside the excluded regions, FCV-PICRN reconstructs velocity, pressure, and vorticity with high structural similarity index (SSIM) and peak signal-to-noise ratio (PSNR), and remains robust under Gaussian noise up to 20%. Finally, for forced two-dimensional Kolmogorov turbulence with 5% Gaussian noise, FCV-PICRN yields lower reconstruction errors than a conventional PINN. These results demonstrate robust physics-informed reconstruction of severely incomplete and noisy flows across laminar and turbulent regimes.
Related
- Extending convolutional neural networks to finite-volume meshes for fine-scale reconstruction of wall-bounded turbulence
- Non-linear mechanical field reconstruction coupling recurrent neural networks with physics-informed graph neural networks
- Reconstruction of Molten Pool Flow Fields from High-Speed Video Using Physics-Informed Neural Networks
- Physics-informed Kolmogorov-Arnold networks (PIKANs) for noise-robust reconstruction of boiling heat flux from infrared measurements
- Learning Memory Operators with Physics-Informed Neural Networks: State Recovery, Kernel Reconstruction, and Parameter Identifiability
BibTeX
@misc{barman2026fourierfeatureembedded,
title = {{Fourier-Feature-Embedded Complex-Valued Physics-Informed Convolutional Residual Networks for Sparse and Gappy Noisy Flow-Field Reconstruction}},
author = {Biswanath Barman and Rajendra K. Ray},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.22743},
eprint = {2610.22743},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.22743},
note = {CC BY-NC-SA 4.0},
}
RIS
TY - UNPB
TI - Fourier-Feature-Embedded Complex-Valued Physics-Informed Convolutional Residual Networks for Sparse and Gappy Noisy Flow-Field Reconstruction
AU - Biswanath Barman
AU - Rajendra K. Ray
PY - 2026
DA - 2026/10/15
PB - arXiv
DO - 10.48550/arxiv.2610.22743
UR - https://arxiv.org/abs/2610.22743
AB - Reconstructing flow fields from sparse and noisy measurements becomes particularly challenging when data are completely absent from dynamically critical regions, such as the vortex-formation region of a wake. Conventional interpolation and numerical reconstruction methods generally require sufficient spatial or boundary information, while physics-informed neural networks (PINNs) can also face difficulties when sparse and noisy observations provide little or no supervision in the region of interest. To address this gap, we propose a Fourier-feature-embedded complex-valued physics-informed convolutional residual network (FCV-PICRN) for sparse, gappy, and noisy flow-field reconstruction. The framework integrates Fourier features for oscillatory and multiscale structures, complex-valued representations for coupled amplitude--phase learning, and convolutional residual blocks constrained by the Navier--Stokes equations. The method is first verified using the Taylor--Green vortex at Re=100 and then evaluated for the cylinder wake at Re=100. Three severe missing-data configurations are imposed in the dynamically sensitive wake: a large hole, a large truncation, and a plus-shaped gap extending in both streamwise and transverse directions. These gaps are further combined with additive Gaussian and multiplicative speckle noise. Despite having no supervised velocity data inside the excluded regions, FCV-PICRN reconstructs velocity, pressure, and vorticity with high structural similarity index (SSIM) and peak signal-to-noise ratio (PSNR), and remains robust under Gaussian noise up to 20%. Finally, for forced two-dimensional Kolmogorov turbulence with 5% Gaussian noise, FCV-PICRN yields lower reconstruction errors than a conventional PINN. These results demonstrate robust physics-informed reconstruction of severely incomplete and noisy flows across laminar and turbulent regimes.
ER -