Frequency Domain Features Research Articles

Real-time lithology identification from drilling data with self & cross attention model and wavelet transform

Accurate lithology identification during drilling operations is pivotal for optimizing drilling efficiency and informed decision-making. Traditional methods, such as mud logging, Logging While Drilling (LWD), and Measurement While Drilling (MWD), often face challenges such as delays and inaccuracies, lacking consistent utilization of real-time engineering data across all drilling sites. Consequently, these methods frequently overlook the complex interactions of influencing factors, resulting in underutilized geological information that could significantly enhance operational outcomes. This paper introduces the SelfAttention-CrossWT (SACWT) model, an innovative approach that integrates self-attention and cross-attention mechanisms with Wavelet Transform (WT) to utilize real-time drilling parameters, including Weight on Bit (WOB) and Rate of Penetration (ROP). The SACWT model effectively harnesses both temporal and frequency-domain characteristics of drilling data, facilitating comprehensive extraction of latent patterns and dynamics, which substantially improves lithology identification accuracy. The output of the model is the real-time lithology classification at the drill bit's depth. The performance of the SACWT model is compared with Long Short-Term Memory (LSTM) networks and a Self-Attention (SA) model that does not utilize frequency-domain data. When applied to a dataset from the Volve oil field, the SACWT model achieved identification accuracies of 87% and 91% in two separate test wells, outperforming traditional methods. These results underscore the model's efficacy and the significant potential of utilizing real-time drilling data for enhanced lithological characterization. This breakthrough offers a robust framework for future exploration and development operations, providing a more precise, efficient, and cost-effective approach to geological analysis.

Emergency Vehicle Classification Using Combined Temporal and Spectral Audio Features with Machine Learning Algorithms

This study presents a novel approach to emergency vehicle classification that leverages a comprehensive set of informative audio features to distinguish between ambulance sirens, fire truck sirens, and traffic noise. A unique contribution lies in combining time domain features, including root mean square (RMS) and zero-crossing rate, to capture the temporal characteristics, like signal energy changes, with frequency domain features derived from short-time Fourier transform (STFT). These include spectral centroid, spectral bandwidth, and spectral roll-off, providing insights into the sound’s frequency content for differentiating siren patterns from traffic noise. Additionally, Mel-frequency cepstral coefficients (MFCCs) are incorporated to capture the human-like auditory perception of the spectral information. This combination captures both temporal and spectral characteristics of the audio signals, enhancing the model’s ability to discriminate between emergency vehicles and traffic noise compared to using features from a single domain. A significant contribution of this study is the integration of data augmentation techniques that replicate real-world conditions, including the Doppler effect and noise environment considerations. This study further investigates the effectiveness of different machine learning algorithms applied to the extracted features, performing a comparative analysis to determine the most effective classifier for this task. This analysis reveals that the support vector machine (SVM) achieves the highest accuracy of 99.5%, followed by random forest (RF) and k-nearest neighbors (KNNs) at 98.5%, while AdaBoost lags at 96.0% and long short-term memory (LSTM) has an accuracy of 93%. We also demonstrate the effectiveness of a stacked ensemble classifier, and utilizing these base learners achieves an accuracy of 99.5%. Furthermore, this study conducted leave-one-out cross-validation (LOOCV) to validate the results, with SVM and RF achieving accuracies of 98.5%, followed by KNN and AdaBoost, which are 97.0% and 90.5%. These findings indicate the superior performance of advanced ML techniques in emergency vehicle classification.

Application of Machine Learning Techniques for Countering Side-Channel Attacks in Cryptographic Systems

The use of machine learning algorithms in order to, not only, detect the adversarial intend behind side-channel attacks on cryptographic systems, but also to resist Differential Power Analysis (DPA) attacks. In particular, with the help of the DPA Challenge Dataset containing power traces of AES encryption operations, we propose a detailed step-by-step approach that includes data acquisition, preprocessing, feature extraction, and model assessment. The pre-processing includes noise reduction, normalization and segmental processing of the collected data for which basic statistical and frequency domain analysis can be used for extraction of relevant features. Support Vector Machines (SVMs) are then trained and tested in order to classify and in turn predict attack scenarios as per the subsequently derived features. As the outcome of the result pages show, the SVM model successfully classifies attack and non-attack traces at a rate of 88% on the validation set, which underlines the usage of machine learning to boost cryptographic security. Investigation of the feature relevance demonstrates that frequency-domain features, namely FFT coefficients are most impactful. The findings of this research prove that machine learning can be useful in preventing side-channel attacks apart from providing valuable information on enhancing the understanding of different defenses in cryptographic systems as well as future development of this domain.

Depression detection based on the temporal-spatial-frequency feature fusion of EEG

Depression is a prevalent affective psychiatric disorder projected to be the leading contributor to the world’s disease burden by 2030. Due to its high prevalence and low recognition rate, an objective and effective detection method is urgently needed. Deep learning methods based on electroencephalography (EEG) have shown significant potential in depression detection. However, excessive channels can increase redundancy and computational complexity in EEG, while irrelevant channels may reduce accuracy. Additionally, existing models often overlook the complementarity between the temporal-, spatial-, and frequency-domain features of EEG, limiting their detection capabilities. To address these issues, we propose a method that fuses the temporal, spatial, and frequency domain features of EEG to enhance the detection accuracy while eliminating redundant channels. We introduce an EEG channel selection method based on frequency domain weighting that automatically adjusts the channel weights to select the EEG channels that best capture spatial information across the delta, theta, alpha, beta, and gamma bands, thereby optimizing the extraction of spatial-frequency features. In addition, we designed a multiscale spatiotemporal convolutional attention network to extract the spatiotemporal features of EEG. In this network, the multiscale convolutional attention module enhanced the model’s ability to capture spatial features, whereas the temporal trend-aware self-attention module extracted long-term temporal features by analyzing global correlations across different time points. Experimental results on the MODMA dataset show that our method achieved a 97.24% detection accuracy, surpassing current state-of-the-art models. This study offers a novel approach for constructing depression detection models, providing a foundation for future research and application.

Post-Stroke Dysarthria Voice Recognition based on Fusion Feature MSA and 1D

Post-stroke Dysarthria (PSD) is one of the common sequelae of stroke. PSD can harm patients’ quality of life and, in severe cases, be life-threatening. Most of the existing methods use frequency domain features to recognize the pathological voice, which makes it hard to completely represent the characteristics of pathological voice. Although some results have been achieved, there is still a long way to go for practical applications. Therefore, an improved deep learning-based model is proposed to classify between the pathological voice and the normal voice, using a novel fusion feature (MSA) and an improved 1D ResNet network hybrid bi-directional LSTM with dilated convolution (named 1D DRN-biLSTM). The experimental results show that our fusion features bring greater improvement in pathological speech recognition than the method that only analyzes the MFCC features, and can better synthesize the hidden features that characterize pathological speech. In terms of model structure, the introduction of dilated convolution and LSTM can further improve the performance of the 1D Resnet network, compared to ordinary networks such as CNN and LSTM. The accuracy of this method reaches 82.41% and 100% at the syllable level and speaker level, respectively. Our scheme outperforms other existing methods in terms of feature learning capability and recognition rate, and will help to play an important role in the assessment and diagnosis of PSD in China.

DF-YOLO: An attempt on enhancing generalization in fabric defect detection based on YOLO network

This article introduces Domain-fusion YOLO (DF-YOLO), a novel object detection network enhancing YOLOv5 series networks’ generalizability, particularly in fabric defect detection. DF-YOLO incorporates a unique feature extractor in its backbone, enhancing frequency domain feature perception alongside spatial information. Utilizing Fourier transform, the network better discerns fabric textures and anomalies, addressing traditional convolutional neural networks’ limited receptive field issue. Feature fusion and dimension reduction are applied for capturing latent features. In its neck, DF-YOLO integrates an improved BiFPN-based feature-fusion structure and CBAM attention mechanism, optimizing feature selection and fusion across scales. This mitigates information loss due to varying input resolutions and architecture depth. DF-YOLO marks a significant advancement in textile anomaly detection, improving generalization while maintaining efficiency. Its innovative approach benefits automated quality control, offering a sophisticated tool for industries requiring precise fabric inspection.

Monitoring the gas metal arc additive manufacturing process using unsupervised machine learning

The study aimed to assess the performance of several unsupervised machine learning (ML) techniques in online anomaly (The term “anomaly” is used here to indicate a departure from expected process behavior which may indicate a quality issue which requires further investigation. The term “defect detection” has often been used previously but the specific imperfection is often indirectly inferred.) detection during surface tension transfer (STT)-based wire arc additive manufacturing. Recent advancements in quality monitoring for wire arc manufacturing were reviewed, followed by a comparison of unsupervised ML techniques using welding current and welding voltage data collected during a defect-free deposition process. Both time domain and frequency domain feature extraction techniques were applied and compared. Three analysis methodologies were adopted: ML algorithms such as isolation forest, local outlier factor, and one-class support vector machine. The results highlight that incorporating frequency analysis, such as fast Fourier transform (FFT) and discrete wavelet transform (DWT), for feature extraction based on general frequency response and defined bandwidth frequency response, significantly improves performance, reflected in a 14% increase in F2 score, compared with time-domain features extraction. Additionally, a deep learning approach employing a convolutional autoencoder (CAE) demonstrated superior performance by processing time-frequency domain data stored as spectrograms obtained through short-time Fourier transform (STFT) analysis. The CAE method outperformed frequency domain analysis and traditional ML approaches, achieving an additional 5% improvement in F2-score. Notably, the F2-score (The F2 score is the weighted harmonic mean of the precision and recall (given a threshold value). Unlike the F1 score, which gives equal weight to precision and recall, the F2 score gives more weight to recall than to precision.) increased significantly from 0.78 in time domain analysis to 0.895 in time-frequency analysis. The study emphasizes the potential of utilizing low-cost sensors to develop anomaly detection modules with enhanced accuracy. These findings underscore the importance of incorporating advanced data processing techniques in wire arc additive manufacturing for improved quality control and process optimization.

Web tension AI modeling and reconstruction for digital twin of roll-to-roll system

AbstractDigital twins (DT) are gaining attention as an emerging technology in Smart manufacturing systems. These DTs comprise various units that enable simulation, monitoring, and prediction of the manufacturing process. This study introduces a predictive model for web tension and a tension reconstruction algorithm for the DT of the roll-to-roll (R2R) system. The observed web tension signals from tension sensors decomposed into a mean component, a sinusoidal wave, and a random noise. Utilizing deep neural networks, the predictive model integrated various sub-models to forecast statistical (mean, standard deviation) and frequency domain (main frequency, signal-to-noise ratio) features of the web tension signal. Through fivefold cross-validation, 23 model architectures were optimized, with selected architectures ranging from 16-32-32-1 to 16-32-64-32-1 nodes per layer. Overall, R2 scores on the test set ranged from approximately 52 to 100%. The proposed reconstruction algorithm generated tension signals from the model’s predictions that closely resemble the original tension signals, indicating credible reconstructions. The proposed predictive model and reconstruction algorithm were integrated into the DT of the R2R system, offering a valuable tool for monitoring and optimizing the R2R process.

Instrumented timed up and go test and machine learning-based levodopa response evaluation: a pilot study

BackgroundThe acute levodopa challenge test (ALCT) is a universal method for evaluating levodopa response (LR). Assessment of Movement Disorder Society’s Unified Parkinson’s Disease Rating Scale part III (MDS-UPDRS III) is a key step in ALCT, which is some extent subjective and inconvenience.MethodsThis study developed a machine learning method based on instrumented Timed Up and Go (iTUG) test to evaluate the patients’ response to levodopa and compared it with classic ALCT. Forty-two patients with parkinsonism were recruited and administered with levodopa. MDS-UPDRS III and the iTUG were conducted in both OFF-and ON-medication state. Kinematic parameters, signal time and frequency domain features were extracted from sensor data. Two XGBoost models, levodopa response regression (LRR) model and motor symptom evaluation (MSE) model, were trained to predict the levodopa response (LR) of the patients using leave-one-subject-out cross-validation.ResultsThe LR predicted by the LRR model agreed with that calculated by the classic ALCT (ICC = 0.95). When the LRR model was used to detect patients with a positive LR, the positive predictive value was 0.94.ConclusionsMachine learning based on wearable sensor data and the iTUG test may be effective and comprehensive for evaluating LR and predicting the benefit of dopaminergic therapy.

Effects of platelet-rich plasma injection on electrical activity and biomechanics of the erector spinae muscles in lumbar myofascial pain syndrome

Low back pain (LBP) is a highly prevalent disease. Among the various causes of LBP, one of the most frequent is myofascial pain syndrome (MPS) which affects the spinal stabilizer muscles. The aims of this study were to compare the differences in muscular electrical activity and biomechanical properties between the painful and non-painful sides in patients with unilateral MPS and to verify the feasibility of surface electromyography (sEMG) and MyotonPRO for assisting in MPS assessment. Forty patients with unilateral lumbar MPS were recruited via the Department of Rehabilitation Medicine Center of West China Hospital Sichuan University from October 2022 to October 2023. The electrical properties of the bilateral erector spinae muscles were characterized by sEMG signals during a trunk extension task. The following four time-domain features of sEMG were extracted: root mean square (RMS), mean absolute value (MAV), integrated EMG (iEMG), and waveform length (WL). And two frequency domain features were extracted: the median frequency (MDF) and mean power frequency (MPF). The mechanical properties of the muscles were assessed by MyotonPRO at rest. The following biomechanical parameters were acquired: oscillation frequency [Hz], dynamic stiffness [N/m], logarithmic decrement, relaxation time [ms], and Creep. The visual analog scale (VAS) was used to evaluate the pain severity, and the Oswestry Disability Index (ODI) was used to evaluate the severity of disability and disruption to lifestyle activities caused by LBP pain. The outcome measures were obtained prior to the Platelet-rich plasma (PRP) treatment and repeated two weeks after treatment. (1) Prior to the PRP treatment, all sEMG time-domain features on the painful side were significantly higher than those on the non-painful side (RMS, p < 0.001; MAV, p < 0.001; iEMG, p < 0.001; WL, p = 0.001). However, there was no significant difference in the sEMG frequency-domain features (MPF, p = 0.478; MDF, p = 0.758). On the mechanical side, there were significant differences in oscillation frequency (p = 0.041) and logarithmic decrement (p = 0.022) between the painful side and non-painful side, but no significant differences in dynamic stiffness, relaxation time, and creep (both p > 0.05). (2) Two weeks after the PRP treatment, statistically significant decreases were observed in both post-treatment VAS (p < 0.001) and ODI scales (p < 0.001), indicating the PRP treatment clinically significantly reduced the level of. MPS. This change coincided with all sEMG time-domain features, in which the values at the painful side decreased significantly (RMS, p = 0.001; MAV, p = 0.001; iEMG, p = 0.001; WL, p = 0.001). However, no significant difference in the sEMG frequency-domain features (MPF, p = 0.620; MDF, p = 0.850) was found. On the mechanical side, only logarithmic decrement on the painful side increased significantly (p < 0.001). Our combined MyotonPRO and sEMG results indicated that MPS likely leads to increased muscle tone and decreased muscle elasticity, manifested by abnormal time-domain features of sEMG and biomechanical properties. The changes in these objective measurements were agreed with the changes in subjective outcome measures of pain and function currently assessed in the patients with MPS. A single PRP treatment may alleviate muscle dysfunction caused by MPS. These preliminary results demonstrated the potential feasibility of using sEMG and MyotonPRO as tools for assessing the neuromuscular function of MPS.

Identification of coating layer pipeline defects based on the GA-SENet-ResNet18 model

Detecting damage in coated pipelines is a challenging and costly task. This study proposes a method for pipeline defect identification based on VMD-DWT noise reduction and GA-SENet-ResNet18. Combining wavelet transform to convert denoised defect signals into time-frequency representations enhances the model's ability to capture both time-domain and frequency-domain features of defect signals, thereby improving its recognition capability for different types of defects. The study analyzed the feature extraction capabilities of ALexNet, GooleNet, VGG16, ResNet18, SENet-ResNet18, and GA-SENet-ResNet18 models in pipeline defect recognition. Experimental results show that SENet-ResNet18 achieved an accuracy of 0.9591 on the training set in 9m38s, significantly outperforming the first four models. GA-SENet-ResNet18 achieved 96.83 % accuracy, 96.67 % precision, 96.73 % recall, and 96.68 % F1 score in pipeline defect signal recognition. Compared to ResNet18, it improved accuracy by 2.06 %, precision by 1.94 %, recall by 2.09 %, F1 score by 2.37 %, with a reduction in time by 1m1s. The study demonstrates that the combined improvement of GA and SENet enhances ResNet18 not only in feature selection and response enhancement but also significantly improves its performance compared to traditional ResNet18 networks, making it more effective in pipeline defect recognition tasks. This research is crucial for ensuring pipeline system integrity and preventing pipeline accidents.

Damage source localisation in complex geometries using acoustic emission and acousto-ultrasonic techniques: an experimental study on clear aligners

Passive non-destructive evaluation tools such as acoustic emission (AE) testing and acousto-ultrasonics (AU) approach present a complex problem in damage localisation in complex and nonhomogeneous geometries. A novel AU-guided AE frequency interpretation approach is proposed in this research work which aims at overcoming this limitation. For the experimental evaluation, the damage sources from a geometrically complex clear dental aligners are tested under cyclic compression load and their origins are evaluated. Despite the rapid worldwide diffusion of the clear aligners, their mechanical behaviour is poorly investigated. In this work, the frequency characteristics of the artificially simulated stress wave, generated from different dental positions of the clear aligners, are studied using the AU approach. These frequency characteristics are then used to analyse the AE signals generated by these aligners when subjected to cyclic compressive loading. In addition, the time domain characteristics of the AE signals are studied using their Time of Arrival (ToA). The Akaike Information Criterion (AIC) is used to estimate the ToA. These frequency and time domain characteristics of the AE signals are used to estimate the local damage origin in the clear dental aligners. This will help in identifying localised damage sources during the usage period of the aligners. Experimental results revealed significant damages in the left maxillary premolar and right maxillary third molar of the aligners.

Frequency-Domain Features and Low-Frequency Synchronization of Photoplethysmographic Waveform Variability and Heart Rate Variability with Increasing Severity of Cardiovascular Diseases.

Objective-Heart rate variability (HRV) and photoplethysmographic waveform variability (PPGV) are available approaches for assessing the state of cardiovascular autonomic regulation. The goal of our study was to compare the frequency-domain features and low-frequency (LF) synchronization of the PPGV and HRV with increasing severity of cardiovascular diseases. Methods-Our study included 998 electrocardiogram (ECG) and finger photoplethysmogram (PPG) recordings from subjects, classified into five categories: 53 recordings from healthy subjects, aged 28.1 ± 6.2 years, 536 recordings from patients with hypertension (HTN), 49.0 ± 8.8 years old, 185 recordings from individuals with stable coronary artery disease (CAD) (63.9 ± 9.3 years old), 104 recordings from patients with myocardial infarction (MI) that occurred three months prior to the recordings (PMI) (65.1 ± 11.0 years old), and 120 recordings from study subjects with acute myocardial infarction (AMI) (64.7 ± 11.5 years old). Spectral analyses of the HRV and PPGV were carried out, along with an assessment of the synchronization strength between LF oscillations of the HRV and of PPGV (synchronization index). Results-Changes in all frequency-domain indices and the synchronization index were observed along the following gradient: healthy subjects → patients with HTN → patients with CAD → patients with PMI → patients with AMI. Similar frequency-domain indices of the PPGV and HRV show little relationship with each other. Conclusions-The frequency-domain indices of the PPGV are highly sensitive to the development of any cardiovascular disease and, therefore, are superior to the HRV indices in this regard. The S index is an independent parameter from the frequency-domain indices.

Effects of rock pore and micromorphology on electromagnetic radiation characteristics

Electromagnetic radiation (EMR) is a crucial tool for monitoring and early warning of underground engineering disasters. Investigating the inherent pore characteristics of rocks is essential for a comprehensive understanding of the EMR phenomenon. The EMR was monitored during various types of rock splitting failures. The pore structure and micromorphology of rocks are studied using quantitative methods such as mercury intrusion porosimetry, fractal analysis, and the Gray Level Co-occurrence Matrix (GLCM). Results indicate that the fractal dimension of red sandstone is significantly lower than the other three rocks. The fractal complexity increases sequentially from red sandstone to marble, granite, and limestone. As the fractal dimension decreases, the signal waveform characteristics of the four rocks become more complex before the main fracture, with a significant increase in signals during the compaction and elasticity stages. Higher fractal dimensions lead to a shift in energy and count from elasticity stage to the post-peak stage. The main fracture amplitudes of the four rocks generally exhibited a consistent pattern, following the sequence of granite > marble > limestone > red sandstone. The main fracture amplitude decreases with increasing complexity of the rock's pore micromorphology. Rock pore characteristics affect frequency domain characteristics by influencing rock strength and crack expansion. An increase in the average pore diameter tends to decrease both the main and center frequencies.

A Fault Diagnosis Method for Electric Vehicle Lithium Power Batteries Based on Dual Feature Extraction from the Time and Frequency Domains

Abstract This study focuses on the safety and reliability issues of lithium-ion batteries, proposing a fault diagnosis strategy that leverages dual feature extraction in the time-frequency domains. Additionally, by modifying the traditional autoencoder, the study proposes a feature-guided autoencoder as an unsupervised model for extracting features in the time domain. Initially, wavelet packet decomposition and its energy denoising treatment are employed to refine fault information within battery voltage signals. Subsequently, the reconstruction error outputted by the Feature-Guided Autoencoder is utilized as the time-domain fault feature, while the cosine similarity of the energy of signals in various frequency bands obtained after wavelet packet decomposition serves as the frequency-domain fault feature. Ultimately, this paper selects the Isolation Forest algorithm for two-dimensional outlier detection of time-frequency features. Experimental results demonstrate that the Feature-Guided Autoencoder proposed in this study not only enhances the sensitivity of time-domain fault features compared to traditional autoencoders and their variants but also optimizes issues related to training time and computational load. The effectiveness of the proposed dual feature fault diagnosis method in both the time and frequency domains is validated through data from two actual vehicles, showing superior early fault detection capability relative to single-feature fault diagnosis methods.

Research on the Unsteady Flow Characteristics of Solid–Liquid Two-Phase Flow in a Deep-Sea Mining Lift Pump and Model Experimental Verification

The deep-sea mining lift pump is one of the pivotal components in deep-sea mineral transportation systems and its internal flow is very complex; consequently, unraveling its unsteady flow behavior pattern holds immense practical value. This study adopts numerical methods to analyze the time-averaged distribution characteristics of the internal flow field in mining lift pumps, as well as the flow field’s pulsation intensity distribution characteristics, the vortex’s spatiotemporal evolution process in both moving and static cascades, and the time- and frequency-domain pulsation characteristics of internal pressure in each flow passage component under four different flow conditions are also investigated. The hydraulic properties of mining lift pumps under these four different conditions are also evaluated, and the outcomes are benchmarked against those of numerical predictions. Our findings reveal that the interplay between impeller blades and guide vanes significantly influences the pump’s flow characteristics, with the pump’s unsteady flow influencing its hydraulic properties. Experimental validation of this system confirms that the pump under study is in line with design specifications in terms of hydraulic properties. The method validation test on the prototype pump shows that the SST k-ω model is capable of successfully forecasting instability in the flow features of deep-sea mining lift pumps. These results will serve as a theoretical reference for regulating the flow state inside deep-sea mining lift pumps.

Transfer Learning and Deep Neural Networks for Robust Intersubject Hand Movement Detection from EEG Signals

In this research, five systems were developed to classify four distinct motor functions—forward hand movement (FW), grasp (GP), release (RL), and reverse hand movement (RV)—from EEG signals, using the WAY-EEG-GAL dataset where participants performed a sequence of hand movements. During preprocessing, band-pass filtering was applied to remove artifacts and focus on the mu and beta frequency bands. The initial system, a preliminary study model, explored the overall framework of EEG signal processing and classification, utilizing time-domain features such as variance and frequency-domain features such as alpha and beta power, with a KNN model for classification. Insights from this study informed the development of a baseline system, which innovatively combined the common spatial patterns (CSP) method with continuous wavelet transform (CWT) for feature extraction and employed a GoogLeNet classifier with transfer learning. This system classified six unique pairs of events derived from the four motor functions, achieving remarkable accuracy, with the highest being 99.73% for the GP–RV pair and the lowest 80.87% for the FW–GP pair in intersubject classification. Building on this success, three additional systems were developed for four-way classification. The final model, ML-CSP-OVR, demonstrated the highest intersubject classification accuracy of 78.08% using all combined data and 76.39% for leave-one-out intersubject classification. This proposed model, featuring a novel combination of CSP-OVR, CWT, and GoogLeNet, represents a significant advancement in the field, showcasing strong potential as a general system for motor imagery (MI) tasks that is not dependent on the subject. This work highlights the prominence of the research contribution by demonstrating the effectiveness and robustness of the proposed approach in achieving high classification accuracy across different motor functions and subjects.

Breast Ultrasound Computer-Aided Diagnosis System Based on Mass Irregularity Features in Frequency Domain

Our study develops a computer-aided diagnosis (CAD) system for breast ultrasound by presenting an innovative frequency domain technique for extracting mass irregularity features, thereby significantly boosting tumor classification accuracy. The experimental data consists of 5252 ultrasound breast tumor images, including 2745 benign tumors and 2507 malignant tumors. A Support Vector Machine was employed to classify the tumor as either benign or malignant, and the effectiveness of the proposed features set in distinguishing malignant masses from benign ones was validated. For the constructed CAD system, the performance indices’ accuracy, sensitivity, specificity, PPV, and NPV were 92.91%, 89.94%, 91.38%, 90.29%, and 91.45%, respectively, and the area index in the ROC analysis (AUC) was 0.924, demonstrating our method’s superiority over traditional spatial gray level dependence (SGLD), the ratio of depth to width, the count of depressions, and orientation features. Therefore, the constructed CAD system with the proposed features will be able to provide a precise and quick distinction between benign and malignant breast tumors with minimal training time in clinical settings.

Remaining useful life prediction of rolling bearings based on time convolutional network and transformer in parallel

Abstract Remaining useful life prediction (RUL) is crucial for maintaining the reliability and safety of industrial equipment. Recently, the Transformer has been widely used due to its ability to effectively extract global feature information in rolling bearing RUL prediction. However, the Transformer is weak in acquiring local feature information and cannot extract temporal features from the degradation process. Conversely, a temporal convolutional network (TCN) can effectively extract local features but is weak in global feature extraction. Therefore, to address the above problems, this paper proposes a prediction method based on the parallel combination of TCN and Transformer. The method first extracts the time domain, frequency domain, and time-frequency domain features from the original vibration signals. After screening through the evaluation indices, the features are fused using K-means clustering and principal component analysis (PCA) to construct health indicators (HI) that characterize the degradation of rolling bearings. Then, a TCN-Transformer parallel network model is constructed to extract both local and global features, and a feed-forward neural network (FNN) is used for the prediction of RUL. Finally, experimental validation is carried out on the PHM2012 bearing dataset, and the results show that the method achieves higher RUL prediction accuracy compared to other methods.

Social anxiety prediction based on ERP features: A deep learning approach

BackgroundSocial Anxiety Disorder is traditionally diagnosed using subjective scales that may lack accuracy. Recently, EEG technology has gained importance for anxiety detection due to its ability to capture stable and objective neurophysiological activities. However, existing methods mainly focus on extracting EEG features during resting states, with limited use of psychologically features like Event-Related Potential (ERP) in task-related states for anxiety detection in deep learning frameworks. MethodsWe collected EEG data from 63 participants exposed to four facial expressions and extracted task-relevant features. Using the EEGNet model, we predicted social anxiety and evaluated its performance using metrics such as accuracy, F1 score, sensitivity, and specificity. We compared EEGNet's performance with Deep Convolutional Neural Network (DeepConvNet), ShallowConvNet, Bi-directional Long Short-Term Memory (BiLSTM), and SVM. To assess the generalizability of the results, we carried out the same procedure on our prior dataset. ResultsEEGNet outperformed other models, achieving 99.16 % accuracy with Late Positive Potential (LPP). ERP components yielded higher accuracy than time-domain and frequency-domain features for social anxiety recognition. Accuracy was better for neutral and negative facial stimuli. Consistency across two datasets indicates stability of findings. LimitationsDue to limited publicly available task-state datasets, only our own were used. Future studies could assess generalizability on larger datasets from different sources. ConclusionsWe conducted the first test of ERP features in anxiety recognition tasks. Results show ERP features have greater potential in social anxiety recognition, with LPP exhibiting high stability and accuracy. Outcomes indicate recognizing social anxiety with negative or neutral facial stimuli is more useful.