Time-frequency Methods Research Articles

A promising method for breaking the logjam of time-frequency analysis in astronomy

Abstract Time-frequency analysis could provide detailed dynamic information of celestial bodies and is critical for comprehension of astronomical phenomena. However, it is far from being well-developed in astronomy. Hilbert–Huang transform (HHT) is an advanced time-frequency method but has two problems in analysing astronomical signals. One is that many astronomical signals may be composed of multiple components with various amplitudes and frequencies, while HHT uses assisted noises with the same amplitude to extract all components. The other is that HHT is an empirical method requiring tunable parameters to be optimized using experimental results or known facts, which are challenging to obtain in astronomy and it is therefore hard to determine whether the signal decomposition is right or not. In this study, we adjust the noise amplitude to optimize the decomposition based on the orthogonality of the obtained components and discard the decompositions with non-physical results. Three experiments show that this new extension of HHT is an effective method suitable for high-resolution time-frequency analysis in astronomy. It can be used to dig out valuable pieces of information which are inaccessible with other methods, and thus has the potential to open up new avenues for astronomy research.

Deep Learning-Based Automated Emotion Recognition Using Multimodal Physiological Signals and Time-Frequency Methods

Deep Learning-Based Automated Emotion Recognition Using Multimodal Physiological Signals and Time-Frequency Methods

Determination of duration, threshold and spatiotemporal distribution of extreme continuous precipitation in nine major river basins in China

China is one of the countries most severely affected by extreme precipitation, and it is urgent to carry out relevant research on the spatiotemporal distribution of extreme precipitation in China. Compared with single-day precipitation events, heavy precipitation events often last for multiple days and have greater potential disaster impacts, such as floods and mudslides. Therefore, in this study, we focus on continuous precipitation (CP) events for analysis. In order to determine the most probable duration of continuous heavy precipitation events, we innovatively adopted two time-frequency variation methods, namely Fast Fourier Transform (FFT) and Continuous Wavelet Transform (CWT). Then through Detrended Fluctuation Analysis (DFA) and multifractal DFA (MF-DFA) methods, extreme CP thresholds are extracted according to the fluctuation characteristics of the CP series. The results show that the average duration of heavy precipitation in the Pearl River Basin (PRB) area was the longest (up to 7.7 days), followed by the Continental Basin (CB) area (7.6 days). The Haihe River Basin (HRB) area had the shortest duration of heavy precipitation (5.0 days), followed by the Huaihe River Basin (HURB) area (5.9 days). The extreme CP threshold in the Southeast Basin (SEB) area was the largest (112.17 mm), followed by the PRB area (108.22 mm). The extreme CP threshold in the CB area was the smallest (20.17 mm), followed by the Yellow River Basin (YRB) area (44.31 mm). Through mutation and trend testing to evaluate the changing state of the extreme CP time series, it was found that, with one exception (HRB), the average extreme CP after the mutation in the watershed areas was larger than that before the mutation, and eight watersheds showed an upward trend in extreme CP. This suggests that most watershed areas in China are at risk of extreme CP increases. This study can provide an important reference for the analysis of extreme CP in China.

A subject-independent portable emotion recognition system using synchrosqueezing wavelet transform maps of EEG signals and ResNet-18

ObjectiveDesigning a portable Brain-Computer Interface (aBCI) using EEG signals is challenging due to the numerous channels, though not all are vital for emotional recognition. We aimed to simplify this by creating a two-channel portable aBCI using advanced time-frequency analysis and deep learning. MethodOur approach involved utilizing the time-frequency analysis named synchrosqueezing wavelet transform (SSWT), which provides better frequency resolution for fluctuations of EEG signal than common wavelet transform. Using the ResNet-18 Convolutional Neural Network, we fine-tuned for sadness and happiness classification. The two best channels were identified across four databases: SEED-IV, SEED-V, SEED-GER, and SEED-FRA, using the Leave-One-Subject-Out method. ResultsFinally, we achieved an average accuracy over sad and happy emotions using the SSWT-ResNet18 model of 76.66%, 78.12%, 81.25%, and 75.00% for the SEED-IV, SEED-V, SEED-GER, and SEED-FRA databases, respectively. ConclusionOverall, our study demonstrates the potential for developing a rapid aBCI by utilizing a precise time–frequency method and deep learning technique from the least number of channels. SignificanceOur approach has promising implications for future real-world applications in emotional recognition.

Naive Bayesian classifier target recognition based on ridge features

Traditional time-frequency algorithms do not have a high feature recognition rate for targets in a low signal-to-noise background, and the number of operations for classification using plain Bayes is large and slow. To solve the above problems and achieve recognition of different ship echo signals in water, this paper proposes a target recognition method based on time-frequency ridge features by using collected audio data of different ship types, combined with Fourier Simultaneous Compression Transform (abbreviation: FSST), starting from time-frequency features, by extracting the FSST time-frequency ridge and finally using the Hopper Parsimonious Bayes classifier to obtain respectively. The recognition rate of ships under FSST and the recognition rate of ships based on ridge line features are obtained respectively. The final comparison shows that the correct recognition rate of the plain Bayesian classifier using the spine features of the time-frequency map is much higher than that of the conventional FSST time-frequency method, and it also outperforms the FSST in terms of computational power and speed.

The application of time-frequency methods of acoustic signal processing in the diagnostics of tram drive components

The application of time-frequency methods of acoustic signal processing in the diagnostics of tram drive components

Investigating bridge vibrational modes under operational conditions using time-frequency analysis

This study aims to accurately and effectively investigate the instantaneous frequency (IF) of the monitored acceleration of the bridge, as well as the individual components of the monitored signal, using advanced time-frequency techniques. To achieve this, first, synthetic signals resembling the monitored acceleration are constructed and processed using four time-frequency techniques. The accuracy of IF tracking, signal reconstruction, and representation resolution are obtained and compared between these different time-frequency techniques. Subsequently, the monitored vertical and torsional accelerations from the Humber Bridge are analysed using the Fourier synchrosqueezed transform and the improved multisynchrosqueezing transform. This allows for the reconstruction of individual acceleration components along with the corresponding variation of the natural frequency. The magnitude of each component of the monitored acceleration is then obtained. The relative importance of modes in the monitored acceleration is evaluated based on the acceleration root mean square. This study presents a novel approach that involves reconstructing the monitored acceleration and assessing the magnitude of individual components of the monitored data of the bridge under operational conditions using the advanced time-frequency method. The method is particularly suitable for addressing multicomponent monitored data and is expected to benefit future practical research in the bridge health monitoring field.

Three-component high-resolution seismic time–frequency polarization filter

SUMMARY The analysis of earthquake recordings from three-component instruments can be challenging due to overlapping events. Time–frequency (TF) polarization methods are efficient tools for this purpose, which can discriminate these events. Previous polarization methods did not consider all three components simultaneously while transferring data to TF domain, which can cause inaccuracies in the reconstruction of wave amplitudes. Therefore, the three-component sparse adaptive S transform (3C-SAST) algorithm is preferred to other TF decompositions since it is mainly developed for polarization analysis purposes, and outperforms other TF methods. In this paper, we developed the 3C-SAST by adding a parameter to adjust the sparseness of the solution and make the resolution flexible. The developed TF decomposition is then used to extend the Morozov & Smithson method to TF domain, and devise a new TF polarization filter whose invertibility and resolution flexibility make it a promising tool for wavefield separation. This filter can eliminate the out-of-plane arrival energies and extract the Rayleigh waves for multicomponent data, which has application in Rayleigh wave tomography and seismological studies. We demonstrated the efficiency of the proposed method for seismic surface waves separation using synthetic signals and three-component teleseismic earthquake recording.

Nonlinear and chaos features over EMD/VMD decomposition methods for ictal EEG signals detection

The detection and identification of epileptic seizures attracted considerable relevance for the neurophysiologists. In order to accomplish the detection of epileptic seizures or equivalently ictal EEG states, this paper proposes the use of nonlinear and chaos features not computed over the raw EEG signals as it was commonly experienced, but instead over intrinsic mode functions (IMFs) extracted subsequently to the application of newly time-frequency signal decomposition methods on the basis of empirical mode decomposition (EMD) and variational mode decomposition (VMD) methods. The first step within the proposed methodology is to excerpt the various components of the IMFs by EMD and VMD decomposition methods on time EEG segments. The Hjorth parameters, the Hurst exponent, the Recurrence Quantification Analysis (RQA), the detrended fluctuation analysis (DFA), the Largest Lyapunov Exponent (LLE), The Higuchi and Katz fractal dimensions (HFD and KFD), seven nonlinear and chaos features computed over the IMFs were investigated and their classification performances evaluated using the k-nearest neighbor (KNN) and the multilayer perceptron neural network (MLPNN) classifiers. Furthermore, the combination of the best nonlinear features has also been examined in terms of sensitivity, specificity and overall classification accuracy. The publicly available Bonn EEG dataset has been has been employed to validate the efficiency of the proposed method for detecting ictal EEG signals from normal or interictal EEG segments. Among the several experiments involved in the current study, the ultimate results establish that the overall classification accuracy can achieve 100%, 99.45%, 99.8%, 99.8%, 98.6% and 99.1% for six different epileptic seizure detection case problems studied, confirming the ability of the proposed methodology in helping the clinic practitioners in the epilepsy detection care units to classify seizure events with a great confidence.

Energy loss analysis of transition simulation for a prototype reversible pump turbine during load rejection process

The dynamic behavior of a reversible pump turbine (RPT) during the load rejection process was investigated through numerical simulation and experimental testing. The study focused on various parameters, such as the operation of guide vanes, torque, flow, head, and runner forces. The pressure fluctuations at different monitoring points were analyzed using the time-frequency method, specifically the continuous wavelet transforms (CWT). The results revealed significant pressure pulsation amplitudes in the vaneless region. Additionally, the entropy production rate was employed to quantify energy loss in different domains, including direct dissipation, turbulent dissipation, and wall shear stress. Moreover, the study utilized the vorticity transport equation to examine vortex characteristics and transport in the stay vanes, guide vanes, and runner flow channels. The vortex characteristics were classified into relative vortex stretching (RVS), Coriolis force (CORF), and viscous diffusion (VISD). The findings highlight significant energy loss in flow patterns associated with high RVS and CORF regions. This study provides valuable insights into the transition process and energy loss of RPTs during load rejection, serving as a basis for future research aiming to enhance the safety and reliability of RPTs under load rejection conditions.

A time-frequency method for induction motor fault feature extraction using instantaneous power signals

To solve the issue that the characteristic frequencies of the broken rotor bar fault (BRBF) are always submerged by the fundamental component during the stator current spectrum analysis in light loads, this paper proposes an effective approach based on the instantaneous power signal and the local maximum synchrosqueezing transform (LMSST). By multiplying the current and voltage, the instantaneous power signal can provide more fault features and enhance fault component amplitudes. The LMSST method is used to obtain a time-frequency analysis with more concentrated energy, and more fault characteristics are extracted from the reconstructed signal, which helps to accurately identify the motor BRBF.

Classification of Epileptic and Psychogenic Nonepileptic Seizures via Time-Frequency Features of EEG Data.

The majority of psychogenic nonepileptic seizures (PNESs) are brought on by psychogenic causes, but because their symptoms resemble those of epilepsy, they are frequently misdiagnosed. Although EEG signals are normal in PNES cases, electroencephalography (EEG) recordings alone are not sufficient to identify the illness. Hence, accurate diagnosis and effective treatment depend on long-term video EEG data and a complete patient history. Video EEG setup, however, is more expensive than using standard EEG equipment. To distinguish PNES signals from conventional epileptic seizure (ES) signals, it is crucial to develop methods solely based on EEG recordings. The proposed study presents a technique utilizing short-term EEG data for the classification of inter-PNES, PNES, and ES segments using time-frequency methods such as the Continuous Wavelet transform (CWT), Short-Time Fourier transform (STFT), CWT-based synchrosqueezed transform (WSST), and STFT-based SST (FSST), which provide high-resolution time-frequency representations (TFRs). TFRs of EEG segments are utilized to generate 13 joint TF (J-TF)-based features, four gray-level co-occurrence matrix (GLCM)-based features, and 16 higher-order joint TF moment (HOJ-Mom)-based features. These features are then employed in the classification procedure. Both three-class (inter-PNES versus PNES versus ES: ACC: 80.9%, SEN: 81.8%, and PRE: 84.7%) and two-class (Inter-PNES versus PNES: ACC: 88.2%, SEN: 87.2%, and PRE: 86.1%; PNES versus ES: ACC: 98.5%, SEN: 99.3%, and PRE: 98.9%) classification algorithms performed well, according to the experimental results. The STFT and FSST strategies surpass the CWT and WSST strategies in terms of classification accuracy, sensitivity, and precision. Moreover, the J-TF-based feature sets often perform better than the other two.

Wavelet-Based Output-Only Damage Detection of Composite Structures.

Health monitoring of structures operating in ambient environments is performed through operational modal analysis, where the identified modal parameters, such as resonant frequencies, damping ratios and operation deflection shapes, characterize the state of structural integrity. The current study shows that, first, time-frequency methods, such as continuous wavelet transform, can be used to identify these parameters and may even provide a large amount of such data, increasing the reliability of structural health monitoring systems. Second, the identified resonant frequencies and damping ratios are used as features in a damage-detection scheme, utilizing the kernel density estimate (KDE) of an underlying probability distribution of features. The Euclidean distance between the centroids of the KDEs, at reference and in various other cases of structural integrity, is used as an indicator of deviation from reference. Validation of the algorithm was carried out in a vast experimental campaign on glass fibre-reinforced polymer samples with a cylindrical shell structure subjected to varying degrees of damage. The proposed damage indicator, when compared with the well-known Mahalanobis distance metric, yielded comparable damage detection accuracy, while at the same time being not only simpler to calculate but also able to capture the severity of damage.

Self-supervised time-frequency representation based on generative adversarial networks

Time-frequency (TF) transforms are commonly used to analyze local features of nonstationary seismic data and to help uncover structural or geologic information. Traditional TF transforms, such as short-time Fourier transform, continuous wavelet transform, and S-transform, suffer from the Heisenberg uncertainty principle, and their TF resolution is limited. The sparse TF (STF) transform has been proposed to address this disadvantage; however, expensive calculation and parameter selection present difficulties. We have developed a self-supervised TF representation based on a generative adversarial networks (STFR-GANs) model to map a 1D seismic signal into a 2D STF image. This model includes three components: a generator, a discriminator, and a reconstruction module. The generator is used to generate the STF spectrum of the input seismic trace, whereas the discriminator distinguishes if this generated STF spectrum is optimal. The reconstruction module serves as a physical constraint to ensure the accuracy of the generated STF spectrum. When implementing model training, the discriminator learns to identify the ideal STF and guides the generator to produce a TF spectrum closer to the ideal one. After model training, we applied the model to synthetic and field data to demonstrate its effectiveness and stability in characterizing the TF features of seismic data. Our results find that STFR-GAN can effectively provide TF representations with higher readability than those of traditional TF methods, further benefiting fluvial channel delineation.

Autonomic and Vascular Responses during Reactive Hyperemia in Healthy Individuals and Patients with Sickle Cell Anemia.

Background and Objectives: To compare autonomic and vascular responses during reactive hyperemia (RH) between healthy individuals and patients with sickle cell anemia (SCA). Materials and Methods: Eighteen healthy subjects and 24 SCA patients were subjected to arterial occlusion for 3 min at the lower right limb level. The pulse rate variability (PRV) and pulse wave amplitude were measured through photoplethysmography using the Angiodin® PD 3000 device, which was placed on the first finger of the lower right limb 2 min before (Basal) and 2 min after the occlusion. Pulse peak intervals were analyzed using time-frequency (wavelet transform) methods for high-frequency (HF: 0.15-0.4) and low-frequency (LF: 0.04-0.15) bands, and the LF/HF ratio was calculated. Results: The pulse wave amplitude was higher in healthy subjects compared to SCA patients, at both baseline and post-occlusion (p < 0.05). Time-frequency analysis showed that the LF/HF peak in response to the post-occlusion RH test was reached earlier in healthy subjects compared to SCA patients. Conclusions: Vasodilatory function, as measured by PPG, was lower in SCA patients compared to healthy subjects. Moreover, a cardiovascular autonomic imbalance was present in SCA patients with high sympathetic and low parasympathetic activity in the basal state and a poor response of the sympathetic nervous system to RH. Early cardiovascular sympathetic activation (10 s) and vasodilatory function in response to RH were impaired in SCA patients.

Experimental investigation on vortex-induced vibration characteristics of a segmented free-hanging flexible riser

Different from the traditional marine riser, the vertical lifting pipeline in the deep-sea mining system can be regarded as a free-hanging flexible riser with an unconstrained bottom end. Likewise, problems in terms of vortex-induced vibration (VIV) and flexible deformations can arise during operation. This paper presents experimental tests on a segmented free-hanging marine riser used for deep-sea mining, focusing on the vortex-induced vibration. The riser was formed by connecting pipe sections using flanges, with the top fixed and a lumped mass block added at the bottom to simulate the lifting pump and buffer station, resulting in weakly constrained boundary conditions. By employing the modal superposition principle and the wavelet time-frequency method, we comprehensively investigated the response characteristics of the typical pipe sections and the coupled oscillation of the riser's vibration modes. During the experiment, two important phenomena that differed from previous VIV tests were observed. Firstly, at low reduced velocities (Ur ≤ 7.77), the dominant frequency no longer conformed to the doubling principle in both directions but remained consistent. Secondly, the weak constraint effect of the concentration of mass at the bottom disrupts the formation of vortex shedding, thereby reducing the bending response of the CF direction of the bottom pipe sections. Meanwhile, the drag force enhances the bending response of the IL direction. Additionally, the middle and bottom pipe sections of the riser exhibited obvious mode competition, with the dominant mode often followed by several subordinate modes with considerable vibration energy.

Tail risk transmission in technology-driven markets

The soaring popularity of blockchain investing and cryptocurrencies has captured the attention of policymakers and investors alike. However, as cryptocurrencies remain the most volatile and high-risk investment options, they have displayed extreme asymmetric patterns over time. Considering these concerns, we conducted a comprehensive analysis of the tail risk transmission of these technology-driven markets using the Conditional autoregressive Value at risk (CAViaR) model, which sheds valuable light on the market's tail characteristics. We combined the CAViaR approach with the time-frequency methods proposed by Diebold and Yilmaz (2012) and Barunik and Krehlik (2018) to further enhance our analysis. Our results revealed a range of asymmetric economic and financial patterns across markets and highlighted the varying exposure of these markets to different circumstances over time. Finally, we investigated the impact of global factors on the tail risk transmission of technology-driven markets both in the long and short term. This study offers a wealth of insights for policymakers, investors, financial market participants, and scholars of digital finance to help navigate these rapidly evolving markets.

Dynamic linkages between shipping and commodity markets: Evidence from a novel asymmetric time-frequency method

This study extends the existing literature in this area by examining the connectedness and shock spillover between commodity and shipping markets using a new novel time-varying frequency and quantile connectedness method developed by Chatziantoniou et al (2022) based on B&K (2018) and Ando et al (2018). Connectedness and shock transmission between the markets were analysed with daily data covering July 4, 2012 to July 20, 2022. A major value added of this study to the existing literature is the examination of the asymmetric effect of commodity price changes (or return) on the connectedness of the markets. Mean-frequency total connectedness analysis indicates that, the overall shipping market (BDI) is both the transmitter (to) and receiver of the highest shock from the entire market connectedness system. In the short-term, the agricultural markets dominate as both the transmitters and receivers of the major shocks to and from the entire market system, while in the medium-term, the shipping markets dominate as both the transmitters and receivers of the largest shocks to the entire market system. However, in the long-term, connectedness and shock propagation were very low. The time-varying quantile analysis reveals that, connectedness was very strong before, during and after COVID-19 at the bearish and bullish market conditions. Further, the time-varying frequency connectedness analysis shows that, although total connectedness is relatively high overtime, it was propelled by short-term dynamics. Metal markets are connected among themselves, and with both agricultural and shipping markets. Agricultural markets are connected among themselves, and with shipping markets, which are only connected among themselves. There is evidence of the asymmetric effect of commodity return dynamics on the connectedness of the markets. Some important policy recommendations were drawn from the findings.

Non-negative tensor factorization for vibration-based local damage detection

In this study, a novel non-negative tensor factorization (NTF)-based method for vibration-based local damage detection in rolling element bearings is proposed. As the diagnostic signal registered from a faulty machine is non-stationary, the time–frequency method is frequently used as a primary decomposition technique. It is proposed here to extract multi-linear NTF-based components from a 3D array of time–frequency representations of an observed signal partitioned into blocks. As a result, frequency and temporal informative components can be efficiently separated from non-informative ones. The experiments performed on synthetic and real signals demonstrate the high efficiency of the proposed method with respect to the already known non-negative matrix factorization approach.

A novel approach for modeling of the ultrasonic signal backscattered in immersed multilayer structures at normal incidence: Time, Frequency, and Velocity dispersion representation

This study proposes a theoretical procedure for modeling ultrasonic signal backscattered in normal incidence by a multilayer structure (MS) immersed in water. The propagation of a longitudinal ultrasonic wave in a MS is studied analytically by using the transfer matrix method (TMM), in which each layer is established as a quadrupole formalism combining the stresses and velocities. The transfer matrices allow us to determine the reflection coefficient of the multilayer structure. The MS solid/liquid/solid case is studied in both insonation configurations (direct and reverse). The modeled ultrasonic response is represented in both time and frequency domains and validated by comparison with experimental data published previously. The time–frequency methods such as spectrogram (SP) and the smoothed pseudo-Wigner–Ville (SPWV) are also used to investigate the character and the value of the non-dispersive of the longitudinal wave. Moreover, the TMM method is used under different conditions (different layer thicknesses, non-uniform distribution of layers, and any frequency range), which are among the causes of MS dispersion. TMM method is able to show this behavior and is exploited to produce (Distance versus Time) and (Distance versus Frequency) representations of wave propagation for each studied MS. The perfect agreement of the comparison of TMM dispersion results with the results is achieved by using time–frequency methods in the case of a 5 MHz central frequency.