Small-scale Motions Research Articles

Highly stretchable, tough and conductive chitin nanofiber composite hydrogel as a wearable sensor

To meet the requirements of eco-friendly and sustainability in the 21st century, hydrogels based on biopolymer with conductivity and stretchable property have attained increasing attention for strain sensor. However, the as-prepared of hydrogel sensor with excellent mechanical property and high strain sensitivity is still a challenge. In this study, chitin nanofiber (ChNF) reinforced composite hydrogels of PACF are fabricated via a facile one-pot method. The obtained PACF composite hydrogel exhibits good transparency (80.6 % at 800 nm)and excellent mechanical properties (tensile strength, 261.2 kPa; tensile strain as high as 550.3 %). Moreover, the composite hydrogels also demonstrate excellent anti-compression performance. The composite hydrogels own good conductivity (1.20 S/m) and strain sensitivity. Most importantly, the hydrogel can be assembled as a strain/pressure sensor for detecting large-scale and small-scale human motion. Therefore, flexible conductive hydrogel strain sensors will have broad application prospects in artificial intelligence, electronic skin, and personal health.

Properties of the turbulent/non-turbulent layer of a turbulent Boussinesq plume: A study using direct numerical simulation

The turbulent/non-turbulent layer (TNTL) in a turbulent Boussinesq plume is analyzed using direct numerical simulations. The Reynolds number (Re) used in the simulation, based on the scales defined at the bottom hot patch, is 2000, and the Reynolds number (Reλ) based on the Taylor microscale (λ) is 114.4 in the self-similar region. The flow is sufficiently resolved till the Kolmogorov scale. The outer edge of the TNTL is detected using the vorticity magnitude. Conditional statistics of various quantities are calculated with reference to the outer edge of the TNTL, which is referred to as the irrotational boundary (IB). The profiles of conditional vorticity magnitude are used to identify the TNTL and determine its thickness. The presence of viscous superlayer (VSL) and turbulent sublayer (TSL) within the TNTL is revealed by exploring the conditional profiles of the enstrophy transport equation. The baroclinic torque, which is a source of vorticity, has been shown to be inconsequential in determining the width of the VSL. The widths of the TNTL and the VSL are determined to be δTNTL≈15η and δVSL≈3.12η, respectively, where η is the Kolmogorov length scale. This gives the width of the TSL as δTSL=δTNTL−δVSL≈13.88η. The invariants of the velocity gradient tensor have been analyzed across the TNTL. The joint probability density function of the invariants Q and R shows a teardrop shape within the turbulent core. The teardrop shape is not fully developed within the TNTL. The TNTL and the turbulent core have a mixed tendency for irrotational dissipation, vortex sheets, and vortex tubes. When normalized by Kolmogorov length and velocity scales, the conditional statistics of the TNTL of the plume are similar to other turbulent flow types, and hence, this provides additional evidence for the universality of small-scale motion within/around the TNTL of various turbulent flows.

Linear response analysis of supersonic turbulent channel flows with a large parameter space

In this work, the linear responses of turbulent mean flow to both harmonic and stochastic forcing are investigated for supersonic channel flow. Well-established universal relations are utilized to obtain efficiently the mean profiles with a large parameter space, with the bulk Mach number up to 5 and the friction Reynolds number up to$10^4$, so a systematic parameter study is feasible. The most amplified structure takes the form of streamwise velocity and temperature streaks forced optimally by the streamwise vortices. The outer peak of the pre-multiplied energy amplification corresponds to the large-scale motion, whose spanwise wavelength ($\lambda _z^+$) is very insensitive to compressibility effects. In contrast, the classic inner peak representing small-scale near-wall motions disappears for the stochastic response with increasing Mach number. Meanwhile, the small-scale motions become much less coherent. A decomposition of the forcing identifies different effects of the incompressible counterpart and the thermodynamic components. Wall-cooling effects, arising with high Mach number, increase the spacing of the most amplified near-wall streaks; the spacing becomes nearly invariant with Mach number if expressed in semi-local units. Meanwhile, the coherence of stochastic response with$\lambda _z^+>90$is enhanced, but on the other hand, with$\lambda _z^+<90$it is decreased. The geometrical self-similarity of the response in the mid-$\lambda _z$range is still roughly satisfied, insensitive to Mach number. Finally, theoretical analyses of the perturbation equations are presented to help with understanding the scaling of energy amplification.

Spectral analysis of turbulence kinetic and internal energy budgets in hypersonic turbulent boundary layers

The spectral transport equations of turbulence kinetic and internal energy (TKE and TIE) are rigorously derived in the compressible turbulent flows, exhibiting a full structural similarity in terms of two-point correlations of velocity and a sound-speed-like variable. The scale dependence of TKE and TIE flow across space and scales, and among components is revealed comprehensively in a similar manner, in hypersonic turbulent boundary layers. Special attention is paid to the effect of wall cooling on the energy evolution, which is observed to exist throughout the boundary layer primarily by means of modifying the small-scale motions.

Study of interscale interactions for turbulence over the obstacle arrays from a machine learning perspective

Investigating interactions between large- and small-scale motions is essential for understanding turbulence over rough boundaries. The present work applies XGBoost models to predict the spatial distribution of ejections and sweeps and quantify their statistical dependence on scale-decomposed velocity fields. Based on large eddy simulation, the models are trained and validated at 20 horizontal planes in turbulence over two types of obstacle arrays. At each height, a default XGBoost model X0 and four comparison models ( X u L , w L , X u L , w S , X u S , w L, and X u S , w S) are trained. The model X0 is trained by the set with four scale-decomposed velocity fields { u L , u S , w L , w S }, where u and w are the streamwise and vertical velocity fluctuations and subscripts L and S refer to above-canyon and sub-canyon scales, while the comparison models are trained by subsets of the scale-decomposed velocity fields. The results indicate that the model X0 predicts the spatial distributions of both ejection and sweep events well, with the structure underestimation being less than 8% within the canopy layer and 3% above it. Along the vertical direction, the relative importance of scale-decomposed velocity fields on the prediction of ejections and sweeps is quantified by the feature importance and prediction errors. The feature importance profiles reveal that both sweeps and ejections are most strongly related to w S within the canopy, but ejections have a stronger dependence on u L well above the canopy. For the comparison models, those trained with w S (namely, X u L , w S and X u S , w S) give better predictions within the canopy layer, whereas those trained with u L (namely, X u L , w S and X u L , w L) perform better above the canopy. This study shows that a machine-learning-based approach can be designed to quantify the relative importance of different scale-decomposed velocity fields on predicting ejections and sweeps and to detect vertical changes of such relative importance.

Resolvent-based analysis of hypersonic turbulent boundary layers with/without wall cooling

The ability of the low-rank approximation of hypersonic turbulent boundary layers with/without wall cooling is examined with the linear resolvent operator in a compressible form. The freestream Mach number of the base flow is 5.86, and the friction Reynolds number is 420. The wall-to-recovery temperature ratio is set as 1.0 and 0.25, respectively, corresponding to an adiabatic wall condition and a cold-wall condition. Different from the resolvent analysis of incompressible turbulent boundary layers, the optimal response mode in the wave-parameter space exhibits a relatively subsonic and a relatively supersonic region [Bae et al., “Resolvent-based study of compressibility effects on supersonic turbulent boundary layers,” J. Fluid Mech. 883, A29 (2020)], divided by the freestream relative Mach number of unity. The features of energy distribution of the optimal response mode in space and scales are examined, and the energy spectra of streamwise velocity and temperature fluctuations, carried by the optimal response mode, are discussed with typical subsets of streamwise and spanwise wavelengths. This reveals the dynamics of the near-wall small-scale and outer larger-scale motions and the distinction in the relatively subsonic/supersonic region. Moreover, the coherent structures, including the velocity and temperature streaks, quasi-streamwise vortices, and large-scale/very-large-scale motions, are identified in the optimal response mode. Special attention is paid to the effects of wall cooling.

Implications of Different Solar Photospheric Flux-transport Models for Global Coronal and Heliospheric Modeling

The concept of surface-flux transport (SFT) is commonly used in evolving models of the large-scale solar surface magnetic field. These photospheric models are used to determine the large-scale structure of the overlying coronal magnetic field, as well as to make predictions about the fields and flows that structure the solar wind. We compare predictions from two SFT models for the solar wind, open magnetic field footpoints, and the presence of coronal magnetic null points throughout various phases of a solar activity cycle, focusing on the months of April in even-numbered years between 2012 and 2020, inclusively. We find that there is a solar-cycle dependence to each of the metrics considered, but there is not a single phase of the cycle in which all the metrics indicate good agreement between the models. The metrics also reveal large, transient differences between the models when a new active region is rotating into the assimilation window. The evolution of the surface flux is governed by a combination of large-scale flows and comparatively small-scale motions associated with convection. Because the latter flows evolve rapidly, there are intervals during which their impact on the surface flux can only be characterized in a statistical sense, thus their impact is modeled by introducing a random evolution that reproduces the typical surface flux evolution. We find that the differences between the predicted properties are dominated by differences in the model assumptions and implementation, rather than the selection of a particular realization of the random evolution.

Dynamic Cross-Linking Network Construction of Carboxymethyl Starch Enabling Temperature and Strain Bimodal Film Sensors

Building stimulus-responsive units in the hydrogel coatings remains challenging for film sensors consisting of alternated layers of inert substrates and hydrogel coatings. An interesting film sensor with a carboxymethyl starch-based hydrogel coating was developed here. The cross-linking networks of carboxymethyl starch play the roles of structure-constructing units and stimulus-controlling units simultaneously, endowing the coatings with thermal sensing and strain sensing capabilities. The dynamic cross-links formed via the boronic ester bonds are temperature-sensitive, releasing or consuming additional acid ions with temperature alteration, and also as primary networks give the hydrogel strength and stretchability with the assistance of semi-penetrated polyacrylamide chains. Therefore, as-prepared flexible film sensors can be used to detect the periodic changes of human temperature and small-scale motion with multiple working modes, discriminating the physical states related to human health. Moreover, this kind of starch-based coating is degradable in a strongly alkaline solution and the inert substrate layer can protect the skin from erosion caused by direct hydrogel-skin contact, and thereby the film sensor is human- and environmentally friendly. This work also proposes a strategy of building temperature-sensitive units in the film sensor via regulating the chemical networks, instead of tuning physical structures.

Large-eddy simulation of separated turbulent flows over a three-dimensional hill using WRF and OpenFOAM

It is not clearly known about the limitations of the large-eddy simulation (LES) mode of an atmospheric model (WRF) in predicting the microscale flows for engineering purpose. This study chooses a typical separated turbulent flow past a three-dimensional axisymmetric hill and investigates the performance of WRF-LES in comparison with the popular CFD solver (OpenFOAM). The numerical models and conditions are set similarly between the two codes. The instantaneous visualization shows that both WRF-LES and OpenFOAM-LES can produce the primary flow features, including hairpin vortices, horseshoes, and surface-shear-induced vortices at different scales, with high similarity. Nevertheless, the turbulent kinetic energy in the near wake produced by WRF-LES is underestimated, in comparison with WRF-LES. The energy spectra suggest that WRF-LES using the high-order advection schemes has a stronger capacity of generating and maintaining small-scale turbulent motions than OpenFOAM-LES. Furthermore, the deviation of numerical dissipation behavior is examined between the two solvers.

Assessment and application of wavelet-based optical flow velocimetry (wOFV) to wall-bounded turbulent flows

The performance of a wavelet-based optical flow velocimetry (wOFV) algorithm in extracting high accuracy and high-resolution velocity fields from tracer particle images in wall-bounded turbulent flows is assessed. wOFV is first evaluated using synthetic particle images generated from a channel flow DNS of a turbulent boundary layer. The sensitivity of wOFV to the regularization parameter (lambda) is quantified and results are compared to cross-correlation-based PIV. Results on synthetic particle images indicated different sensitivity to under-regularization or over-regularization depending on which region of the boundary layer is being analyzed. Nonetheless, tests on synthetic data revealed that wOFV can modestly outperform PIV in vector accuracy across a broad lambda range. wOFV showed clear advantages over PIV in resolving the viscous sublayer and obtaining highly accurate estimates of the wall shear stress and thus normalizing boundary layer variables. wOFV was also applied to experimental data of a developing turbulent boundary layer. Overall, wOFV revealed good agreement with both PIV and a combined PIV + PTV method. However, wOFV was able to successfully resolve the wall shear stress and correctly normalize the boundary layer streamwise velocity to wall units where PIV and PIV + PTV showed larger deviations. Analysis of the turbulent velocity fluctuations revealed spurious results for PIV in close proximity to the wall, leading to significantly exaggerated and non-physical turbulence intensity in the viscous sublayer region. PIV + PTV showed only a minor improvement in this aspect. wOFV did not exhibit this same effect, revealing that it is more accurate in capturing small-scale turbulent motion in the vicinity of boundaries. The enhanced vector resolution of wOFV enabled improved estimation of instantaneous derivative quantities and intricate flow structure both closer to the wall and more accurately than the other velocimetry methods. These aspects show that, within a reasonable lambda range that can be verified using physical principles, wOFV can provide improvements in diagnostics capability in resolving turbulent motion occurring in the vicinity of physical boundaries.Graphical abstract

Amplitude modulation in particle-laden atmospheric surface layers

Particle effects on the amplitude modulation are investigated in this study based on observational data with various mass loading acquired from long-term measurements of aeolian sandstorms in high-Reynolds-number ($Re_{\tau }\sim O(10^6)$) near-neutral atmospheric surface layers. In both particle-laden and unladen flows, in addition to the positive top–down modulation behaviour in the logarithmic region, a significant modulation effect that exists for some specific motions is also found for the single-point amplitude modulation. The most energetic turbulent motions exhibit the strongest modulation effect, and the modulating signals do not change with the small-scale motions being modulated. In particle-laden flows, the length of the most energetic structure is almost constant, thus the scales of the modulating signal and carrier signal are hardly affected by particles. However, the addition of particles changes the distribution of energy between multi-scale turbulent motions. The kinetic energy of the large-scale component is less enhanced than the total kinetic energy by particles. This leads to a reduced energy proportion of the large-scale component and an augmented one of the small-scale component. Moreover, the particles produce a large damping in the degree of the amplitude modulation and move down the positions of the modulating signals and carrier signals corresponding to the strongest inter-layer modulation, but the damping is weakened with the wall-normal distance due to the decreased mass loading. This study may provide a more general insight into the modulation mechanism between multi-scale turbulent motions and the effect of particles on turbulence.

Imaging-spectroscopy of a band-split type II solar radio burst with the Murchison Widefield Array

Type II solar radio bursts are caused by magnetohydrodynamic (MHD) shocks driven by solar eruptive events such as coronal mass ejections (CMEs). Often, both fundamental and harmonic bands of type II bursts are split into sub-bands, which are generally believed to be coming from upstream and downstream regions of the shock; however, this explanation remains unconfirmed. Here, we present combined results from imaging analyses of type II radio burst band splitting and other fine structures observed by the Murchison Widefield Array (MWA) and extreme ultraviolet observations from Solar Dynamics Observatory (SDO)/Atmospheric Imaging Assembly (AIA) on 28 September 2014. The MWA provides imaging-spectroscopy in the range 80−300 MHz with a time resolution of 0.5 s and frequency resolution of 40 kHz. Our analysis shows that the burst was caused by a piston-driven shock with a driver speed of ∼112 km s−1 and shock speed of ∼580 km s−1. We provide rare evidence that band splitting is caused by emission from multiple parts of the shock (as opposed to the upstream–downstream hypothesis). We also examine the small-scale motion of type II fine structure radio sources in MWA images, and suggest that this motion may arise because of radio propagation effects from coronal turbulence, and is not due to the physical motion of the shock location. We present a novel technique that uses imaging spectroscopy to directly determine the effective length scale of turbulent density perturbations, which is found to be 1−2 Mm. The study of the systematic and small-scale motion of fine structures may therefore provide a measure of turbulence in different regions of the shock and corona.

Human Activity Recognition Using Smartphones With WiFi Signals

In this article, we present a work using a smartphone with an off-the-shelf WiFi router for human activity recognition with various scales. The router serves as a hotspot for transmitting WiFi packets. The smartphone is configured with customized firmware and developed software for capturing WiFi channel state information (CSI) data. We extract the features from the CSI data associated with specific human activities, and utilize the features to classify the activities using machine learning models. To evaluate the system performance, we test 20 types of human activities with different scales including seven small motions, four medium motions, and nine big motions. We recruit 60 participants and spend 140 hours for data collection at various experimental settings, and have 36 000 data points collected in total. Furthermore, for comparison, we adopt three distinct machine learning models, including convolutional neural networks (CNNs), decision tree, and long short-term memory. The results demonstrate that our system can predict these human activities with an overall accuracy of 97.25%. Specifically, our system achieves a mean accuracy of 97.57% for recognizing small-scale motions that are particularly useful for gesture recognition. We then consider the adaptability of the machine learning algorithms in classifying the motions, where CNN achieves the best predicting accuracy. As a result, our system enables human activity recognition in a more ubiquitous and mobile fashion that can potentially enhance a wide range of applications such as gesture control, sign language recognition, etc.

Simple model for the bottleneck effect in isotropic turbulence based on Kolmogorov's hypotheses

We propose a simple model for the bottleneck effect in isotropic turbulence based on Kolmogorov's hypotheses. The model of the longitudinal structure function consists of two quadratic functions representing large- and small-scale motions, with parameters derived from asymptotic behaviors of the structure function. The model shows a decaying power law for the bump height in the compensated spectrum, which agrees with the result of direct numerical simulations at a range of Reynolds numbers.

Influence of Atmospheric Flow Structure on Optical Turbulence Characteristics

This article discusses the quality of astronomical images under conditions of moderate small-scale turbulence and varying meso-scale airflows above the Baikal Astrophysical Observatory (BAO). We applied a Weather Research and Forecasting (WRF) Model, as well as statistical estimations of the Fried parameter from the differential motion of the solar images. The simulations were performed with a fairly high horizontal resolution within a large area of 1600 × 1600 km. A high horizontal resolution provides representative estimations of atmospheric characteristics and correct accounting of large-scale air advection. We considered the influence of atmospheric motions over the cold water area of Lake Baikal, as well as meso-scale vortex structures over rough terrain on solar image quality. A better understanding of structured turbulent small-scale motions and optical turbulence over rough terrain may help to develop advanced methods for diagnostics and prediction of image quality. For the first time, we have shown that the BAO is located at the periphery of a meso-scale atmospheric vortex structure with an anticyclonic direction of airflows in the daytime. An increase in image quality was associated with weakening airflows over Lake Baikal and a decrease in the intensity of wind speed fluctuations. Calculated spectra of atmospheric turbulence in the daytime were close to the classical form. At night and in the morning, the spectra had a steeper slope on small scales. Deformations of the spectra were due to the suppression of turbulence under stable stratification of the atmosphere. The characteristic horizontal scales of the transition from “−5/3” to ∼“−3” spectral slope were 2–2.5 km. The results obtained using the WRF model and analysis of optical turbulence strength (namely, the Fried parameter) indicated that the parameterization schemes used in the WRF model were accurate.

Shear-driven magnetic buoyancy in the solar tachocline: the mean electromotive force due to rotation

ABSTRACT The leading theoretical paradigm for the Sun’s magnetic cycle is an αω-dynamo process, in which a combination of differential rotation and turbulent, helical flows produces a large-scale magnetic field that reverses every 11 yr. Most αω solar dynamo models rely on differential rotation in the solar tachocline to generate a strong toroidal field. The most problematic part of such models is then the production of the large-scale poloidal field, via a process known as the α-effect. Whilst this is usually attributed to small-scale convective motions under the influence of rotation, the efficiency of this regenerative process has been called into question by some numerical simulations. Motivated by likely conditions within the tachocline, the aim of this paper is to investigate an alternative mechanism for the poloidal field regeneration, namely the magnetic buoyancy instability in a shear-generated, rotating magnetic layer. We use a local, fully compressible model in which an imposed vertical shear winds up an initially vertical magnetic field. The field ultimately becomes buoyantly unstable, and we measure the resulting mean electromotive force (EMF). For sufficiently rapid rotation, we find that a significant component of the mean EMF is aligned with the direction of the mean magnetic field, which is the characteristic feature of the classical αω-dynamo model. Our results therefore suggest that magnetic buoyancy could contribute directly to the generation of large-scale poloidal field in the Sun.

A high-resolution large-eddy simulation framework for wildland fire predictions using TensorFlow

Background Wildfires are becoming more severe, so we need improved tools to predict them over a wide range of conditions and scales. One approach towards this goal entails the use of coupled fire/atmosphere modelling tools. Although significant progress has been made in advancing their physical fidelity, existing tools have not taken full advantage of emerging programming paradigms and computing architectures to enable high-resolution wildfire simulations. Aims The aim of this study was to present a new framework that enables landscape-scale wildfire simulations with physical representation of combustion at an affordable cost. Methods We developed a coupled fire/atmosphere simulation framework using TensorFlow, which enables efficient and scalable computations on Tensor Processing Units. Key results Simulation results for a prescribed fire were compared with experimental data. Predicted fire behavior and statistical analysis for fire spread rate, scar area, and intermittency showed overall reasonable agreement. Scalability analysis was performed, showing close to linear scaling. Conclusions While mesh refinement was shown to have less impact on global quantities, such as fire scar area and spread rate, it benefits predictions of intermittent fire behavior, buoyancy-driven dynamics, and small-scale turbulent motion. Implications This new simulation framework is efficient in capturing both global quantities and unsteady dynamics of wildfires at high spatial resolutions.

Optical Flow Computation for Video Under the Dynamic Illumination

Optical flow computation for video under the dynamic illumination is a challenging issue in video multimedia applications. In this paper, we solve this issue by introducing an illumination-invariant framework for variational optical flow estimation. It consists of an illumination-invariance model that handles complex illumination changes and a data enhancement model that guarantees highly accurate optical flow estimation. In this framework, we design a log-correlation descriptor for the data term, which handles complex illumination changes by eliminating the common parameters shared by the neighboring pixels in the corresponding illumination model while improving the accuracy of optical flow estimation by enhancing the discriminability of the data term matching. We also introduce a novel optical flow model with <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$L_{0}$</tex-math></inline-formula> regularizer, which reconstructs optical flow field by a sparse flow gradient counting scheme. Different from other edge-preserving regularizers, it does not depend on local motion features, but locates important flow edges globally. Therefore, it will not cause edge blurriness due to avoiding local filtering or average operation. It is particularly effective for enhancing major flow edges while eliminating a manageable degree of low-amplitude motion structures to control smoothing and reduce oversegmentation artifacts. Even small-scale motion structures with high contrast can be preserved remarkably well. The experimental results show our method significantly outperforms previous illumination-robust optical flow methods in handling complex illumination changes, and achieves competitive evaluation results on the challenging MPI-Sintel and Kitti datasets.

CNN-Based Fluid Motion Estimation Using Correlation Coefficient and Multiscale Cost Volume

Motion estimation for complex fluid flows via their image sequences is a challenging issue in computer vision. It plays a significant role in scientific research and engineering applications related to meteorology, oceanography, and fluid mechanics. In this paper, we introduce a novel convolutional neural network (CNN)-based motion estimator for complex fluid flows using multiscale cost volume. It uses correlation coefficients as the matching costs, which can improve the accuracy of motion estimation by enhancing the discrimination of the feature matching and overcoming the feature distortions caused by the changes of fluid shapes and illuminations. Specifically, it first generates sparse seeds by a feature extraction network. A correlation pyramid is then constructed for all pairs of sparse seeds, and the predicted matches are iteratively updated through a recurrent neural network, which lookups a multi-scale cost volume from a correlation pyramid via a multi-scale search scheme. Then it uses the searched multi-scale cost volume, the current matches, and the context features as the input features to correlate the predicted matches. Since the multi-scale cost volume contains motion information for both large and small displacements, it can recover small-scale motion structures. However, the predicted matches are sparse, so the final flow field is computed by performing a CNN-based interpolation for these sparse matches. The experimental results show that our method significantly outperforms the current motion estimators in capturing different motion patterns in complex fluid flows, especially in recovering some small-scale vortices. It also achieves state-of-the-art evaluation results on the public fluid datasets and successfully captures the storms in Jupiter’s White Ovals from the remote sensing images.

Deep learning method for the super-resolution reconstruction of small-scale motions in large-eddy simulation

A super-resolution reconstruction model for the subgrid scale (SGS) turbulent flow field in large-eddy simulation (LES) is proposed, and it is called the meta-learning deep convolutional neural network (MLDCNN). Direct numerical simulation (DNS) data of isotropic turbulence are used as the dataset of the model. The MLDCNN is an unsupervised learning model, which only includes high-resolution DNS data without manually inputting preprocessed low-resolution data. In this model, the training process adopts the meta-learning method. First, in the a priori test, the SGS turbulent flow motions in the filtered DNS (FDNS) flow field are reconstructed, and the energy spectrum and probability density function of the velocity gradient of the DNS flow field are reconstructed with high accuracy. Then, in the a posteriori test, the super-resolution reconstruction of the LES flow field is carried out. The difficulty of LES flow field reconstruction is that it contains filtering loss and subgrid model errors relative to the DNS flow field. The super-resolution reconstruction of the LES flow field achieves good results through this unsupervised learning model. The proposed model makes a good prediction of small-scale motions in the LES flow field. This work improves the prediction accuracy of LES, which is crucial for the phenomena dominated by small-scale motions, such as relative motions of particles suspended in turbulent flows.