Turbulent Signals Research Articles

Noise reduction method for mine wind speed sensor data based on CEEMDAN-wavelet threshold

The mine wind speed sensor is an important intelligent sensing equipment in the mine intelligent ventilation system that can provide accurate and key wind speed parameters for the intelligent ventilation system. The turbulent pulsation characteristics of the airflow in the underground tunnel are a major factor for the inaccurate measurement of mine wind speed. Therefore, according to the random non-stationary characteristics of a turbulent pulsation signal, a denoising method based on adaptive complete ensemble empirical mode decomposition (CEEMDAN) combined with the wavelet threshold is proposed for suppressing the turbulent pulsation noise in the wind speed signal. First, the CEEMDAN algorithm is used for decomposing the wind speed signal into a series of IMF components. Second, the continuous mean square error criterion is used for determining the high-frequency IMF components with more noise. The wavelet threshold denoising method is used for denoising the high-frequency IMF components with more noise. Finally, the denoised IMF components and remaining low-frequency IMF components are reconstructed for obtaining the denoised signal. The results of the denoising analysis of measured turbulent pulsation signals, comparative analysis of denoising of simulated turbulent pulsation signals by different joint denoising methods, and denoising analysis of actual mine wind speed sensor data indicate that the joint denoising method proposed in this study has a higher signal-to-noise ratio and lower root mean square error of the wind speed signal after denoising. Compared with the EMD-wavelet threshold and EEMD-wavelet threshold denoising methods, the denoising method proposed in this study is better and has higher denoising accuracy, which provides a new method for processing actual mine wind speed sensor data.

Voice Type Component Profile Model of Glottal Gap Voice in Ex Vivo Canine Larynges

This paper aimed to create a voice type component profile (VTCP) to model a glottal gap during phonation. VTCP was generated from phonation from eight excised canine larynges with induced posterior glottal gaps. Glottal gaps were induced using metal shims with widths ranging from 0.5 to 3.5mm, in 0.5 mm increments. Acoustic data was collected under each gap condition, and a custom MATLAB software was used to analyze the data for VTCP. The trends of each VTC were analyzed with changes in glottal gap size. Increasing glottal gap led to a decrease in VTC1 and an increase in VTC4. Post-Hoc Tukey tests found that for both VTC4 and VTC1, The 1.5mm group differed from the 2, 2.5, and 3.5mm group. Additionally, in both VTC1 and VTC4 there was a significant difference between both the control and 0.5mm shim groups and the 2, 2.5, 3, and 3.5mm shim groups. VTC2 followed a similar trend as VTC1. Average VTC3 was highest at control, and the 1, 2, 2.5, 3, and 3.5mm shim groups significantly differed from the control group. Kendall correlation tests showed that there was a significant weak inverse relationship with glottal gap size and VTC1, VTC2, and VTC3 and a significant weak positive correlational relationship with glottal gap size and VTC4. As glottal gap width increased, phonation exhibited increased turbulence associated with increased breathiness. Periodicity decreased as turbulence increased. Increased turbulence was shown to be derived from the glottal level. The size of the induced glottal gap was the cause of the turbulent signals. The study highlights the potential of VTCP analysis in detecting subtle changes in glottal closure and suggests avenues for further research in diagnosing voice disorders.

Flow-Field Inference for Turbulent Exhale Flow Measurement.

Vision-based pulmonary diagnostics present a unique approach for tracking and measuring natural breathing behaviors through remote imaging. While many existing methods correlate chest and diaphragm movements to respiratory behavior, we look at how the direct visualization of thermal CO2 exhale flow patterns can be tracked to directly measure expiratory flow. In this work, we present a novel method for isolating and extracting turbulent exhale flow signals from thermal image sequences through flow-field prediction and optical flow measurement. The objective of this work is to introduce a respiratory diagnostic tool that can be used to capture and quantify natural breathing, to identify and measure respiratory metrics such as breathing rate, flow, and volume. One of the primary contributions of this work is a method for capturing and measuring natural exhale behaviors that describe individualized pulmonary traits. By monitoring subtle individualized respiratory traits, we can perform secondary analysis to identify unique personalized signatures and abnormalities to gain insight into pulmonary function. In our study, we perform data acquisition within a clinical setting to train an inference model (FieldNet) that predicts flow-fields to quantify observed exhale behaviors over time. Expiratory flow measurements capturing individualized flow signatures from our initial cohort demonstrate how the proposed flow field model can be used to isolate and analyze turbulent exhale behaviors and measure anomalous behavior. Our results illustrate that detailed spatial flow analysis can contribute to unique signatures for identifying patient specific natural breathing behaviors and abnormality detection. This provides the first-step towards a non-contact respiratory technology that directly captures effort-independent behaviors based on the direct measurement of imaged CO2 exhaled airflow patterns.

Self-supervised transformers for turbulent flow time series

There has been a rapid advancement in deep learning models for diverse research fields and, more recently, in fluid dynamics. This study presents self-supervised transformers' deep learning for complex turbulent flow signals across various test problems. Self-supervision aims to leverage the ability to extract meaningful representations from sparse flow time-series data to improve the transformer model accuracy and computational efficiency. Two high-speed flow cases are considered: a supersonic compression ramp and shock-boundary layer interaction over a statically deformed surface. Several training scenarios are investigated across the two different supersonic configurations. The training data concern wall pressure fluctuations due to their importance in aerodynamics, aeroelasticity, noise, and acoustic fatigue. The results provide insight into transformers, self-supervision, and deep learning with application to complex time series. The architecture is extendable to other research domains where time series data are essential.

A multiscale and multicriteria generative adversarial network to synthesize 1-dimensional turbulent fields

This article introduces a new neural network stochastic model to generate a 1-dimensional stochastic field with turbulent velocity statistics. Both the model architecture and training procedure ground on the Kolmogorov and Obukhov statistical theories of fully developed turbulence, so guaranteeing descriptions of (1) energy distribution, (2) energy cascade and (3) intermittency across scales in agreement with experimental observations. The model is a generative adversarial network (GAN) with multiple multiscale optimization criteria. First, we use three physics-based criteria: the variance, skewness and flatness of the increments of the generated field, that retrieve respectively the turbulent energy distribution, energy cascade and intermittency across scales. Second, the GAN criterion, based on reproducing statistical distributions, is used on segments of different length of the generated field. Furthermore, to mimic multiscale decompositions frequently used in turbulence’s studies, the model architecture is fully convolutional with kernel sizes varying along the multiple layers of the model. To train our model, we use turbulent velocity signals from grid turbulence at Modane wind tunnel.

Generating Corcos-coherent signal series using phase perturbation

A method for producing signals with coherence that decays exponentially as a function of the product of frequency and distance in the manner described by Corcos is proposed. This type of coherence constraint is used for simulating loads resulting from turbulent propagation or pressure signals resulting from wind noise. The present method is executed by stochastically perturbing the phase of the signal as a Gaussian random variable scaled by a simple function of frequency, distance, and a velocity-like decay parameter. While the present focus of this work is on producing signals that exhibit Corcos-type coherence decay, other varieties of coherence relationships including those associated with a diffuse field can also be produced with this approach. [SNL is managed and operated by NTESS under DOE NNSA contract DE-NA0003525.]

Generalizability of transformer-based deep learning for multidimensional turbulent flow data

Deep learning has been going through rapid advancement and becoming useful in scientific computation, with many opportunities to be applied to various fields, including but not limited to fluid flows and fluid–structure interactions. High-resolution numerical simulations are computationally expensive, while experiments are equally demanding and encompass instrumentation constraints for obtaining flow, acoustics and structural data, particularly at high flow speeds. This paper presents a Transformer-based deep learning method for turbulent flow time series data. Turbulent signals across spatiotemporal and geometrical variations are investigated. The pressure signals are coarsely-grained, and the Transformer creates a fine-grained pressure signal. The training includes data across spatial locations of compliant panels with static deformations arising from the aeroelastic effects of shock-boundary layer interaction. Different training approaches using the Transformer were investigated. Evaluations were carried out using the predicted pressure signal and their power spectra. The Transformer's predicted signals show promising performance. The proposed method is not limited to pressure fluctuations and can be extended to other turbulent or turbulent-like signals.

LSTM Reconstruction of Turbulent Pressure Fluctuation Signals

This paper concerns the application of a long short-term memory model (LSTM) for high-resolution reconstruction of turbulent pressure fluctuation signals from sparse (reduced) data. The model’s training was performed using data from high-resolution computational fluid dynamics (CFD) simulations of high-speed turbulent boundary layers over a flat panel. During the preprocessing stage, we employed cubic spline functions to increase the fidelity of the sparse signals and subsequently fed them to the LSTM model for a precise reconstruction. We evaluated our reconstruction method with the root mean squared error (RMSE) metric and via inspection of power spectrum plots. Our study reveals that the model achieved a precise high-resolution reconstruction of the training signal and could be transferred to new unseen signals of a similar nature with extremely high success. The numerical simulations show promising results for complex turbulent signals, which may be experimentally or computationally produced.

Physical consistency and invariance in machine learning of turbulent signals

This paper concerns an investigation of the invariance and consistency of deep learning of turbulent pressure fluctuations. The long-short-memory model is employed to predict wall pressure fluctuations across physical regimes featuring turbulence, shock–boundary layer interaction, and separation. The model's sensitivity to the data inputs is examined using different input data sets. Training the deep learning model based on the raw signals from different flow regions leads to large inaccuracies. It is shown that the data must be appropriately pre-processed before training for the deep learning model predictions to become consistent. Removing the mean and using the normalized fluctuating component of the signal, the deep learning predictions not only greatly improved in accuracy but, most importantly, converged and became consistent, provided that the signal sparsity remains within the inertial sub-range of the turbulence energy spectrum cascade. The power spectra of the surface pressure fluctuations reveal that the model provides high accuracy up to a certain frequency for the fully turbulent flow. The deep learning model's consistency is evidenced by being transferable across the various probe positions on the wall despite the significant differences in the turbulent flow properties in the training data set, i.e., signals obtained before, after, and inside the shock–boundary layer interaction regions. The model's prediction consistency and invariance to the turbulent signal training location(s) are promising for applying deep learning models to various turbulent flows.

Resonant duet modeling approach for the spectral analysis of strongly nonlinear wave turbulent signal

Resonant duet modeling approach for the spectral analysis of strongly nonlinear wave turbulent signal

Interacting Linear Modes in the Turbulent Flow of an Industrial Swirled Combustor

Swirled flows often feature global modes that become manifest as skew-symmetric helical vortices, also known as precessing vortex cores, which dominate the flow dynamics. This study uses linear stability analysis (LSA) and bispectral mode decomposition (BMD) to elucidate the interactions of such modes in the turbulent, nonreacting, swirled flow of an industrial, three-passage, aeroengine fuel injector obtained via large-eddy simulation. A discrete Fourier transform in time retrieves the modes as narrow-banded, evenly spaced peaks in the spectral amplitude of the turbulent signal, where the second mode is dominant. Similar arrangements are known to appear as a consequence of a saturated global mode, which amplifies its higher harmonic frequencies due to nonlinear effects. We show that each mode appears in the spectrum of LSA eigenvalues, indicating that the subdominant peaks are caused not only due to nonlinear interactions but that they have an underlying linear mechanism. A structural sensitivity analysis based on the adjoint LSA shows that the observed helical modes originate close to the exit plane of the fuel injector. Finally, the BMD reveals significant nonlinear interaction between the individual modes. It is hypothesized that this interaction amplifies the modes, which are linearly stable, leading to the strong dynamics in the flow.

Propagation of random excitations in the polar representation of strongly nonlinear four-wave systems of turbulence.

The present work is concerned with the uncertainty propagation of the wave turbulent system. In particular, we study the temporal development and long-term behavior of the probability with respect to the amplitude and phase of complex-valued waves constituting the generic four-wave system of turbulence. Our approach to approximating the target distribution function is via the three steps: (i) to grasp the physical process described by the true turbulence model as random process, (ii) to determine the stochastic differential equation whose solution exhibits statistically similar behavior with the underlying turbulent signal, and (iii) to solve the corresponding Kolmogorov forward equation. Our implementation of the methodology is distinguished by employing a number of simplified stochastic models and applying one of them in the adaptive fashion which varies subject to the different parameter regime of the true dynamical system model. Accordingly, we become able to demonstrate the effectiveness of this reduced-order modeling framework for the analysis of the turbulent system characterized by not only weak but strong interactions among the nonlinear waves. We numerically corroborate our theoretical predictions in the context of the generalized Majda-Mclaughlin-Tabak wave turbulence prototype.

On the potential of transfer entropy in turbulent dynamical systems

Information theory (IT) provides tools to estimate causality between events, in various scientific domains. Here, we explore the potential of IT-based causality estimation in turbulent (i.e. chaotic) dynamical systems and investigate the impact of various hyperparameters on the outcomes. The influence of Markovian orders, i.e. the time lags, on the computation of the transfer entropy (TE) has been mostly overlooked in the literature. We show that the history effect remarkably affects the TE estimation, especially for turbulent signals. In a turbulent channel flow, we compare the TE with standard measures such as auto- and cross-correlation, showing that the TE has a dominant direction, i.e. from the walls towards the core of the flow. In addition, we found that, in generic low-order vector auto-regressive models (VAR), the causality time scale is determined from the order of the VAR, rather than the integral time scale. Eventually, we propose a novel application of TE as a sensitivity measure for controlling computational errors in numerical simulations with adaptive mesh refinement. The introduced indicator is fully data-driven, no solution of adjoint equations is required, with an improved convergence to the accurate function of interest. In summary, we demonstrate the potential of TE for turbulence, where other measures may only provide partial information.

Effects of subgrid‐scale turbulence parametrization on the representation of clear‐air turbulence using kilometre‐ to hectometre‐scale numerical simulations

A turbulence event arising in a jet exit region above the Belgium–Luxembourg area, determined from airliner in‐situ measurements, is reproduced using the meteorological models Application de la Recherche à l'Opérationnel à Méso‐Echelle (AROME) and Meso‐NH at horizontal resolutions of 1.3 km and 260 m. The behaviour of the subgrid turbulence scheme at 1.3 km and its sensitivity to various parameters are analysed, with results being evaluated using measurements. An increase of the vertical resolution around the tropopause levels with m is shown to greatly enhance the turbulence representation. The use of a non‐local formulation of the mixing length in the current parametrization at 1.3 km allows reproduction of a turbulence signal in agreement with the observations. On the contrary, the use of a fully three‐dimensional formulation has no impact on the simulation at this resolution (1.3 km). Using the 260 m runs, this turbulence event is linked to hydrodynamical wind shear instabilities characterized by horizontal wavelength of 4.5 km, sub‐resolved at the operational resolution. At these small grid‐size scales, turbulence evolution and equation budgets reflect an equilibrium between dynamical production and turbulence dissipation, and highlight the importance of horizontal gradients. Subgrid turbulence intensities are assessed to be underestimated by the current parametrization at 1.3 km when compared with this high‐resolution reference simulation. Finally, different tests on the turbulence parametrization illustrate a transfer between resolved and subgrid kinetic energy in the model. This transfer stresses the importance of a trade‐off between mixing intensity and the representation of wind at resolved scales for the upper troposphere.

Spatial coherence length and wave structure function for plane waves transmitted in anisotropic turbulent oceans.

The spatial coherence length and wave phase structure function are two important factors in describing turbulence's effect on light propagation in seawater. This paper derives the wave phase structure function and spatial coherence length of plane waves in moderate to strong turbulent channels by deriving a "modification seawater turbulence power spectrum" and an oceanic-modified Rytov approximation. The evolutions in wave structure function, coherence length with the temperature dissipation rate, energy dissipation rate, anisotropy turbulence factor, signal wavelength, and propagation distance are analyzed by numerical calculation. In the moderate and strong turbulence regions, the phase structure function and spatial coherence length increase and decrease with increasing transmission distance and turbulence strength, respectively, and there is a saturation tendency for both. The fluctuation of seawater salinity has a greater effect on the phase structure function and coherence length than the temperature fluctuation. In addition, the wave structure function decreases with increasing signal wavelength and degree of turbulent anisotropy, but the trend of spatial coherence length is reversed.

A case of life-threatening hematuria.

A 66-year-old male with a past medical history of hypertension and benign prostatic hypertrophy, not on anticoagulation, presented with difficulty voiding and hematuria. He was seen at an outside hospital earlier in the day and received a straight catheterization for urinary retention, he was prescribed antibiotics for a urinary infection, and he was discharged home. Subsequently, after discharge, he developed hematuria with inability to void, prompting a visit to our emergency department (ED). On arrival, the patient was uncomfortable appearing, tachycardic with a heart rate of 116 beats per minute and a blood pressure of 114/67 mm Hg. His lower abdomen was tender to palpation. A point-of-care ultrasound (POCUS) of the bladder was performed. The ultrasound demonstrated a distended-appearing bladder containing a large area of layering mixed echogenicity with intrinsic turbulent Doppler signal (Video 1). The patient had a three-way Foley catheter placed in the ED for continuous bladder irrigation. He had a computerized tomography (CT) angiography of the abdomen and pelvis that did not reveal any active hemorrhage. He also had a CT urogram. Initially, the hematuria improved, however, following the abnormal ultrasound Doppler images, the patient was admitted to the urology service. Although he was admitted, his hemoglobin declined from 12 to 7.9 g/dL, and he suffered a cardiac arrest. The reason for the arrest was presumed to be from hemorrhagic shock. There was return of spontaneous circulation after initiation of massive transfusion protocol. He underwent cystoscopy showing an enlarged median prostate lobe with active extravasation, requiring embolization of the prostatic artery by interventional radiology (IR). Hematuria and urinary obstruction are common causes of ED visits with multiple underlying etiologies.1 Emergency physicians are tasked with the challenge of determining which patients can be managed as outpatients. In this case, gross hematuria from an actively bleeding prostate led to the hemorrhagic shock requiring multiple transfusions and need for IR embolization. The most common etiologies of prostatic hematuria are benign prostatic hyperplasia, iatrogenic urologic trauma, prostate cancer, and radiation therapy.2 In this patient, POCUS showed turbulent Doppler flow within the bladder (Figure 1A), which was concerning for active bleeding, and ultimately led to the patient's admission and emergent urologic consultation. Additionally, the POCUS also showed layering mixed echogenicities seen within the distended bladder, indicating a clot at various stages (Figure 1B). This case illustrates the importance of considering POCUS in patients with urinary retention and hematuria, especially in the setting of recent catheter placement.

Wavelet analysis of mixing layers with emphasis on intermittency and cavitation events

Many previous studies show that shear cavitation dominates the cavitation characteristics of inducers under the design point. However, understanding this cavitation type is quite difficult due to some complex coupled factors in inducers, such as the rotation and wall effects between the rotator and stator. Therefore, as a first step, a fundamental investigation concerning two-dimensional cavitating mixing layers is carried out in this work. The present study uses a large eddy simulation coupled with a homogeneous cavitation model to clarify the linkage between turbulence and cavitation in fully developed cavitating mixing layers. Special emphasis is put on the analysis of cavitation events at a single point in cavitating mixing layers. With the help of potent wavelet analysis, nonstationary turbulent cavitating flow signals are analyzed in the time-frequency domain. The results indicate that cavitation in mixing layers is more closely associated with the small-scale intermittency features that correspond to the violent fluctuation of the energy distribution in small-scale structures rather than with the large-scale coherent structures, providing a new perspective from which to understand the cavitation process in mixing layers. The close relationship between cavitation events and turbulent intermittency may also explain the discrepancies in the energy spectrum between cavitating and conventional mono-phase flows in the inertial subrange. Ensemble averaging of the velocity signal allows for the reconstruction of the time signatures of small-scale intermittency, whose results indicate that the three cavitating cases do not reveal any evident coherent structures.

Direct wavefront sensing with a plenoptic sensor based on deep learning.

Traditional plenoptic wavefront sensors (PWS) suffer from the obvious step change of the slope response which leads to the poor performance of phase retrieval. In this paper, a neural network model combining the transformer architecture with the U-Net model is utilized to restore wavefront directly from the plenoptic image of PWS. The simulation results show that the averaged root mean square error (RMSE) of residual wavefront is less than 1/14λ (Marechal criterion), proving the proposed method successfully breaks through the non-linear problem existed in PWS wavefront sensing. In addition, our model performs better than the recently developed deep learning models and traditional modal approach. Furthermore, the robustness of our model to turbulence strength and signal level is also tested, proving the good generalizability of our model. To the best of our knowledge, it is the first time to perform direct wavefront detection with a deep-learning-based method in PWS-based applications and achieve the state-of-the-art performance.

Convolutional-neural-network-based versus vision-transformer-based SNR estimation for visible light communication networks.

Visible light communication (VLC) has emerged as a promising technology for future sixth-generation (6 G) communications. Estimating and predicting the impairments, such as turbulence and free space signal scattering, can help to construct flexible and adaptive VLC networks. However, the monitoring of impairments of VLC is still in its infancy. In this Letter, we experimentally demonstrate a deep-neural-network-based signal-to-noise ratio (SNR) estimation scheme for VLC networks. A vision transformer (ViT) is first utilized and compared with the conventional scheme based on a convolutional neural network (CNN). Experimental results show that the ViT-based scheme exhibits robust performance in SNR estimation for VLC networks compared to the CNN-based scheme. Specifically, the ViT-based scheme can achieve accuracies of 76%, 63.33%, 45.33%, and 37.67% for 2-quadrature amplitude modulation (2QAM), 4QAM, 8QAM, and 16QAM, respectively, against 65%, 57.67%, 41.67%, and 34.33% for the CNN-based scheme. Additionally, data augmentation has been employed for achieving enhanced SNR estimation accuracies of 95%, 79.67%, 58.33%, and 50.33% for 2QAM, 4QAM, 8QAM, and 16QAM, respectively. The effect of the SNR step size of a contour stellar image dataset on the SNR estimation accuracy is also studied.

Airflow-based odometry for MAVs using thermal anemometers

This article concerns airflow-based odometry for estimating MAV flight speed from airflow measurements provided by a set of thermal anemometers. Our approach relies on a Gated Recurrent Unit (GRU) based deep learning approach to extract deep features from noisy and turbulent measurement signals of triaxial thermal anemometers, in order to establish the underlying mapping between the airflow measurement and the flight speed. The proposed solution is validated on a multi-rotor MAV. The results show that the GRU-based model can effectively extract noise features and perform denoising, and compensate for induced velocity effects along the propellers’ rotation axis. As a consequence, robust prediction of the flight speed is performed, including during takeoff and landing that induce ground effects and strong variations of vertical airflow.