Set Of Time Domain Features Research Articles

Helical Gearbox Defect Detection with Machine Learning Using Regular Mesh Components and Sidebands.

The current paper presents helical gearbox defect detection models built from raw vibration signals measured using a triaxial accelerometer. Gear faults, such as localized pitting, localized wear on helical pinion tooth flanks, and low lubricant level, are under observation for three rotating velocities of the actuator and three load levels at the speed reducer output. The emphasis is on the strong connection between the gear faults and the fundamental meshing frequency GMF, its harmonics, and the sidebands found in the vibration spectrum as an effect of the amplitude modulation (AM) and phase modulation (PM). Several sets of features representing powers on selected frequency bands or/and associated peak amplitudes from the vibration spectrum, and also, for comparison, time-domain and frequency-domain statistical feature sets, are proposed as predictors in the defect detection task. The best performing detection model, with a testing accuracy of 99.73%, is based on SVM (Support Vector Machine) with a cubic kernel, and the features used are the band powers associated with six GMF harmonics and two sideband pairs for all three accelerometer axes, regardless of the rotation velocities and the load levels.

Enhancing age-related postural sway classification using partial least squares-discriminant analysis and hybrid feature set

Feature sets in a machine learning algorithm can have an impact on the robustness, interpretability, and characterization of the data. To detect age-related changes, traditional linear methods for analyzing center of pressure (COP) signals offer limited insight into the complex nonlinear dynamics of postural control. To overcome this limitation, a novel approach that combines a partial least squares-discriminant analysis (PLS-DA) classifier with the nonlinear dynamics of COP time series was proposed. Three small feature sets were compared: time-domain features alone, entropy-based features alone, and a hybrid approach incorporating both types of features. The performance of the PLS-DA model was assessed in four different eyes and surface conditions by using the accuracy, sensitivity, selectivity, precision metrics, and ROC curves. The results indicated that the PLS-DA model utilizing the hybrid feature set achieved significantly higher accuracy than the time-domain and entropy-based feature sets. The best classification performance was observed when the eyes were open on a compliant surface, with an overall accuracy of 89% for training and 88% for cross-validation. For the old group, while the results indicated 93% sensitivity, 94% specificity, and 93% precision in the training, the results revealed 88% sensitivity, 93% specificity, and 91% precision in cross-validation. Notably, the hybrid feature set yielded an AUC value of 0.96, indicating a superior performance. This study emphasizes the robust classification capabilities of PLS-DA for age-related postural changes and highlights the effectiveness of utilizing a small hybrid feature set to improve classification accuracy and reliability.

A New Bearing Fault Diagnosis Approach based on Common Spatial Pattern Features

This study offers a novel intelligent detection method of rolling bearings that applied: the Common Spatial Pattern (CSP) method for feature extraction of bearing faults. CSP maximizes the variance ratio of the two class signal matrices coming from different sources. Different from the frequency features, characteristic vibration features from each measurement that can distinguish between the faulty and the healthy bearings were extracted. Artificial Neural Network (ANN), Support Vector Machine (SVM), and K-Nearest Neighbour (k-NN) methods were tested on two different data sets. An 88.5% accuracy was obtained with ANN for two-class fault detection and 93.5% for fault classification. The classification results were compared with classical time domain feature sets. CSP features have achieved higher accuracies in all two-class and multiclass cases. Therefore, the CSP features can be a possible means for condition monitoring of bearings through vibration data.

Features selection for estimating hand gestures based on electromyography signals

Hand prosthesis controlled by surface electromyography (sEMG) is promising due to the control capabilities and the noninvasive technique that machine learning (ML) offers to help physically disabled people during daily life. Nevertheless, dexterous prostheses are still infrequently popular due to control problems and limited robustness. This paper proposes a new set of time domain (TD) features to improve the EMG pattern recognition performance. The effect of five feature sets is evaluated based on the three classifiers k-nearest neighbor (KNN), linear discriminate analysis (LDA), and support vector machine (SVM). The EMG signals are obtained from database-5 (DB5) of the ninapro project datasets. In this study, the long-term signals of DB5 are segmented into short-term signals to perform short-term recognition. The results showed that the LDA classifier based on the proposed features achieved high classification accuracy for classifing 17 gestures. The LDA classifier achieved about 96.47% compared to 94.12%, and 93.82% for KNN and SVM classifiers, respectively. The results confirm that the suitable features extracted from short term signals with the appropriate classifier, has an important impact on improving the performance of gesture classification.

Features selection for estimating hand gestures based on electromyography signals

Hand prosthesis controlled by surface electromyography (sEMG) is promising due to the control capabilities and the noninvasive technique that machine learning (ML) offers to help physically disabled people during daily life. Nevertheless, dexterous prostheses are still infrequently popular due to control problems and limited robustness. This paper proposes a new set of time domain (TD) features to improve the EMG pattern recognition performance. The effect of five feature sets is evaluated based on the three classifiers k-nearest neighbor (KNN), linear discriminate analysis (LDA), and support vector machine (SVM). The EMG signals are obtained from database-5 (DB5) of the ninapro project datasets. In this study, the long-term signals of DB5 are segmented into short-term signals to perform short-term recognition. The results showed that the LDA classifier based on the proposed features achieved high classification accuracy for classifing 17 gestures. The LDA classifier achieved about 96.47% compared to 94.12%, and 93.82% for KNN and SVM classifiers, respectively. The results confirm that the suitable features extracted from short term signals with the appropriate classifier, has an important impact on improving the performance of gesture classification.

Statistical Machine Learning Strategy and Data Fusion for Detecting Incipient ITSC Faults in IM

The new technological developments have allowed the evolution of the industrial process to this new concept called Industry 4.0, which integrates power machines, robotics, smart sensors, communication systems, and the Internet of Things to have more reliable automation systems. However, electrical rotating machines like the Induction Motor (IM) are still widely used in several industrial applications because of their robust elements, high efficiency, and versatility in industrial applications. Nevertheless, the occurrence of faults in IMs is inherent to their operating conditions; hence, Inter-turn short-circuit (ITSC) is one of the most common failures that affect IMs, and its appearance is due to electrical stresses leading to the degradation of the stator winding insulation. In this regard, this work proposes a diagnosis methodology capable of performing the assessment and automatic detection of incipient electric faults like ITSC in IMs; the proposed method is supported through the processing of different physical magnitudes such as vibration, stator currents and magnetic stray-flux and their fusion of information. Certainly, the novelty and contribution include the characterization of different physical magnitudes by estimating a set of statistical time domain features, as well as their fusion following a feature-level fusion approach and their reduction through the Linear discriminant Analysis technique. Furthermore, the fusion and reduction of information from different physical magnitudes lead to performing automatic fault detection and identification by a simple Neural-Network (NN) structure since all considered conditions can be represented in a 2D plane. The proposed method is evaluated under a complete set of experimental data, and the obtained results demonstrate that the fusion of information from different sources (physical magnitudes) can lead to achieving a global classification ratio of up to 99.4% during the detection of ITSC in IMs and an improvement higher than 30% in comparison with classical approaches that consider the analysis of a unique physical magnitude. Additionally, the results make this proposal feasible to be incorporated as a part of condition-based maintenance programs in the industry.

Cross-Domain Conv-TasNet Speech Enhancement Model with Two-Level Bi-Projection Fusion of Discrete Wavelet Transform

Nowadays, time-domain features see wide use in speech enhancement (SE) networks such as frequency-domain features to achieve excellent performance in eliminating noise from input utterances. This study primarily investigates how to extract information from time-domain utterances to create more effective features in SE. We extend our recent work by employing sub-signals which dwell in multiple acoustic frequency bands in the time domain and integrating them into a unified time-domain feature set. The discrete wavelet transform (DWT) is applied to decompose each input frame signal to obtain sub-band signals, and a projection fusion process is performed on these signals to create the ultimate features. The corresponding fusion strategy is either bi-projection fusion (BPF) or multiple projection fusion (MPF). In short, MPF exploits the softmax function to replace the sigmoid function in order to create ratio masks for multiple feature sources. The concatenation of fused DWT features and time features serves as the encoder output of two celebrated SE frameworks, the fully convolutional time-domain audio separation network (Conv-TasNet) and the dual-path transformer network (DPTNet), to estimate the mask and then produce the enhanced time-domain utterances. The evaluation experiments are conducted on the VoiceBank-DEMAND and VoiceBank-QUT tasks, and the results reveal that the proposed method achieves higher speech quality and intelligibility than the original Conv-TasNet that uses time features only, indicating that the fusion of DWT features created from the input utterances can benefit time features to learn a superior Conv-TasNet/DPTNet network in SE.

SEMG signal filtering study using synchrosqueezing wavelet transform with differential evolution optimized threshold

Most gesture recognition studies based on surface electromyography (sEMG) signals focus on filtering, in which the lack of diversity for considered noises can still be the problem. In this work, a denoising method based on Synchrosqueezing Wavelet Transform with Differential Evolution optimized threshold (DEOT-SWT) is proposed. The sEMG signals of ten gestures with three mixed noises, including power line interference (PLI), baseline drift (BW), and white Gaussian noise (WGN), are firstly investigated and filtered by DEOT-SWT, which are collected from seven subjects by utilizing two wearable sEMG signal sensors. Then, the most commonly used Hudgins time-domain feature set is extracted for recognizing ten gestures. Three metrics are adopted to evaluate filtering performance: signal-to-noise ratio (SNR), root mean square error and R-squared value. The gesture recognition accuracy is utilized to verify the practical effect of DEOT-SWT in sEMG-based gesture recognition applications. The results of the experiments demonstrate that the DEOT-SWT algorithm accomplishes desirable denoising performance with an average recognition accuracy of 95.95% (±3.88) in comparison to the classic Infinite Impulse Response (IIR) algorithm and the empirical mode decomposition (EMD) algorithm.

Data-driven systems to detect physical weakening from daily routine: A pilot study on elderly over 80 years old.

The use of telemonitoring solutions via wearable sensors is believed to play a major role in the prevention and therapy of physical weakening in older adults. Despite the various studies found in the literature, some elements are still not well addressed, such as the study cohort, the experimental protocol, the type of research design, as well as the relevant features in this context. To this end, the objective of this pilot study was to investigate the efficacy of data-driven systems to characterize older individuals over 80 years of age with impaired physical function, during their daily routine and under unsupervised conditions. We propose a fully automated process which extracts a set of heterogeneous time-domain features from 24-hour files of acceleration and barometric data. After being statistically tested, the most discriminant features fed a group of machine learning classifiers to distinguish frail from non-frail subjects, achieving an accuracy up to 93.51%. Our analysis, conducted over 570 days of recordings, shows that a longitudinal study is important while using the proposed features, in order to ensure a highly specific diagnosis. This work may serve as a basis for the paradigm of future monitoring systems.

Searching for quasi-periodic eruptions using machine learning

Abstract Quasi-periodic eruption (QPE) is a rare phenomenon in which the X-ray emission from the nuclei of galaxies shows a series of large amplitude flares. Only a handful of QPEs have been observed but the possibility remains that there are as yet undetected sources in archival data. Given the volume of data available a manual search is not feasible, and so we consider an application of machine learning to archival data to determine whether a set of time-domain features can be used to identify further light curves containing eruptions. Using a neural network and 14 variability measures we are able to classify light curves with accuracies of greater than $94{{\ \rm per\ cent}}$ with simulated data and greater than $98{{\ \rm per\ cent}}$ with observational data on a sample consisting of 12 light curves with QPEs and 52 light curves without QPEs. An analysis of 83 531 X-ray detections from the XMM Serendipitous Source Catalogue allowed us to recover light curves of known QPE sources and examples of several categories of variable stellar objects.

EMAHA-DB1: A New Upper Limb sEMG Dataset for Classification of Activities of Daily Living

In this paper, we present electromyography analysis of human activity - database 1 (EMAHA-DB1), a novel dataset of multi-channel surface electromyography (sEMG) signals to evaluate the activities of daily living (ADL). The dataset is acquired from 25 able-bodied subjects while performing 22 activities categorised according to functional arm activity behavioral system (FAABOS) (3 - full hand gestures, 6 - open/close office draw, 8 - grasping and holding of small office objects, 2 - flexion and extension of finger movements, 2 - writing and 1 - rest). The sEMG data is measured by a set of five Noraxon Ultium wireless sEMG sensors with Ag/Agcl electrodes placed on a human hand. The dataset is analyzed for hand activity recognition classification performance. The classification is performed using six state-of-the-art machine learning classifiers, including Random Forest (RF), Fine K-Nearest Neighbour (KNN), Ensemble KNN (sKNN), Linear Discriminant Analysis (LDA), Support Vector Machines (SVM), and a hybrid deep learning architecture of Convolutional Neural Network (CNN) and Bidirectional- Long Short Term Memory (Bi-LSTM) layers. In the classical methods, ten combinations of time domain and frequency domain feature sets are analyzed. The state-of-the-art classification accuracy on five FAABOS categories is 83.21% by using the SVM classifier with the third order polynomial kernel using energy feature and auto regressive feature set ensemble. The classification accuracy on 22 class hand activities is 75.39% by the same SVM classifier with the log moments in frequency domain (LMF) feature, modified LMF, time domain statistical (TDS) feature, spectral band powers (SBP), channel cross correlation and local binary patterns (LBP) set ensemble. The analysis depicts the technical challenges addressed by the dataset. The developed dataset can be used as a benchmark for various classification methods as well as for sEMG signal analysis corresponding to ADL and for the development of prosthetics and other wearable robotics.

Myoelectric Control With Fixed Convolution-Based Time-Domain Feature Extraction: Exploring the Spatio–Temporal Interaction

The role of feature extraction in electromyogram (EMG) based pattern recognition has recently been emphasized with several publications promoting deep learning (DL) solutions that outperform traditional methods. It has been shown that the ability of DL models to extract temporal, spatial, and spatio–temporal information provides significant enhancements to the performance and generalizability of myoelectric control. Despite these advancements, it can be argued that DL models are computationally very expensive, requiring long training times, increased training data, and high computational resources, yielding solutions that may not yet be feasible for clinical translation given the available technology. The aim of this paper is, therefore, to leverage the benefits of spatio–temporal DL concepts into a computationally feasible and accurate traditional feature extraction method. Specifically, the proposed novel method extracts a set of well-known time-domain features into a matrix representation, convolves them with predetermined fixed filters, and temporally evolves the resulting features over a short and long-term basis to extract the EMG temporal dynamics. The proposed method, based on Fixed Spatio–Temporal Convolutions, offers significant reductions in the computational costs, while demonstrating a solution that can compete with, and even outperform, recent DL models. Experimental tests were performed on sparse-and high-density EMG (HD-EMG) signals databases, across a total of 44 subjects performing a maximum of 53 movements. Despite the simplification compared to deep approaches, our results show that the proposed solution significantly reduces the classification error rates by 3% to 10% in comparison to recent DL models, while being efficient for real-time implementations.

An automated detection of atrial fibrillation from single‑lead ECG using HRV features and machine learning

BackgroundAtrial fibrillation (AF) is a disorder of the heart rhythm where irregular and rapid heartbeats are observed. This supraventricular arrhythmia may increase the risk of blood clots, stroke, heart failure, and other serious heart complications. Automatic analysis of AF that is based on machine learning (ML) plays an important role in detecting this heart disease. MethodsA new approach for automated AF detection is presented using heart rate variability (HRV) features and machine learning. A set of time-domain, frequency-domain and nonlinear features are extracted from the R-R intervals. A new method for frequency-domain analysis of R-R intervals using the Fourier Decomposition Method is presented, which provides promising results as compared to the usual method of power spectral density estimation. We train the algorithm on the Massachusetts Institute of Technology-Beth Israel Hospital (MIT-BIH) atrial fibrillation database and perform a comprehensive analysis using statistical tests to obtain the results without any intra-patient bias. ResultsThe proposed method is able to achieve average result of 95.16% sensitivity, 92.46% specificity and 94.43% accuracy and its performance is better than the existing approaches. Furthermore, the efficacy of the proposed algorithm is tested on eight records from a previously unseen MIT-BIH Arrhythmia Database. ConclusionThis work shows that the proposed HRV features and ML approach can be effectively used for the analysis, detection, and classification of AF.

Genetic Programming-Based Feature Construction for System Setting Recognition and Component-Level Prognostics

Extracting representative feature sets from raw signals is crucial in Prognostics and Health Management (PHM) for components’ behavior understanding. The literature proposes various methods, including signal processing in the time, frequency, and time–frequency domains, feature selection, and unsupervised feature learning. An emerging task in data science is Feature Construction (FC), which has the advantages of both feature selection and feature learning. In particular, the constructed features address a specific objective function without requiring a label during the construction process. Genetic Programming (GP) is a powerful tool to perform FC in the PHM context, as it allows to obtain distinct feature sets depending on the analysis goal, i.e., diagnostics and prognostics. This paper adopts GP to extract system-level features for machinery setting recognition and component-level features for prognostics. Three distinct fitness functions are considered for the GP training, which requires a set of statistical time-domain features as input. The methodology is applied to vibration signals extracted from a test rig during run-to-failure tests under different settings. The performances of constructed features are evaluated through the classification accuracy and the Remaining Useful Life (RUL) prediction error. Results demonstrate that GP-based features classify known and novel machinery operating conditions better than feature selection and learning methods.

Fast Detection for Cyber Threats in Electric Vehicle Traction Motor Drives

While cyber-physical security of electric vehicles (EV) is gaining increased concerns due to the fast development of vehicle onboard communication networks, the existing literature focuses on the vehicle level and it does not explicitly address cyber-threat detection for the EV powertrain traction motor drives. Therefore, in this paper, we propose a fast, model-free approach to detect cyber-threats in electric vehicle traction motor drives with only four easy-to-get, trustworthy sensor signals. First, the trustworthy motor current signals are selected to undermine the impacts of the vehicle’s random driving cycles. Then a set of innovative time-domain current features that are the most sensitive to a wide range of anomalies are selected to reduce the number of observations needed, thus vastly reducing the computational burden and the time-to-detect. Next, four binary classifiers are developed to detect cyber-threats while a majority vote mechanism is adopted to reduce the false alarm rate. Finally, the proposed method is validated by the real-time hardware-in-the-loop simulations. Validation results show that the proposed detection method achieves much faster detection compared to traditional current signature analysis. Furthermore, the proposed detection methods achieve an accuracy higher than 98% with the false alarm rate less than 0.01%.

Feature Optimization for Gait Phase Estimation with a Genetic Algorithm and Bayesian Optimization

For gait phase estimation, time-series data of lower-limb motion can be segmented according to time windows. Time-domain features can then be calculated from the signal enclosed in a time window. A set of time-domain features is used for gait phase estimation. In this approach, the components of the feature set and the length of the time window are influential parameters for gait phase estimation. However, optimal parameter values, which determine a feature set and its values, can vary across subjects. Previously, these parameters were determined empirically, which led to a degraded estimation performance. To address this problem, this paper proposes a new feature extraction approach. Specifically, the components of the feature set are selected using a binary genetic algorithm, and the length of the time window is determined through Bayesian optimization. In this approach, the two optimization techniques are integrated to conduct a dual optimization task. The proposed method is validated using data from five walking and five running motions. For walking, the proposed approach reduced the gait phase estimation error from 1.284% to 0.910%, while for running, the error decreased from 1.997% to 1.484%.

Automated and Rapid Seal Wear Classification Based on Acoustic Emission and Support Vector Machine

Seal wear in hydraulic cylinders results in fluid leakage, and instability of the piston rod movement. Therefore, regular inspection of seals is required using automated approaches to improve productivity and to reduce unscheduled maintenance. In literature, successful attempts have been made using acoustic emission-based condition monitoring to classify the seal wear. However, limited attempts have been made to implement automated approaches to classify seal wear using acoustic emission features. Therefore, this article presents an automated approach for rapid and computationally economical diagnosis of seal wear using acoustic emission. The experiments were performed at varying pressure conditions on a hydraulic test rig that can simulate fluid leakage conditions similar to that of hydraulic cylinders. From the acoustic emission spectrum, the frequency bands were identified with peak power and by heterodyning the signal. This technique results in 10X downsampling without losing the information of interest. Further, the signal was divided into smaller “snapshots” to facilitate rapid diagnosis. In these tests, the diagnosis was made on short-time windows, as low as 0.3 seconds in length. A set of time and frequency domain features were designed, and the Support Vector Machine (SVM) was trained on 60 % of the test cases generated with seals of three different conditions. The SVM was able to accurately classify the status irrespective of the pressure conditions, with an accuracy of ~99 %. Therefore, the proposed automated seal wear classification technique based on acoustic emission and SVM can be used for real-time monitoring of seal wear in hydraulic cylinders.

Сочетание вручную построенных признаков и признаков глубоких сетей для автоматического обнаружеия дефектов в процессе визуализации квадратичной частотно-модулированной тепловой волны

Recent advancements of non-destructive testing and evaluation (NDT&E) reached the fourth revolution with machine learning, artificial intelligence, and the internet of things as key enablers in parallel with industry 4.0. Nevertheless, Active thermography (AT) is a non-contact, whole field, safe, remote, cost-efficient, and widely used NDT technique for subsurface anomaly detection. In AT, the automatic defect detection is modelled as object localization and semantic segmentation in thermograms. This paper presents a feature fusion network that fuses the global features extracted using a deep neural network (DNN) with the deep features extracted using a convolutional neural network (CNN). A set of handcrafted time-domain statistical and frequency domain features of thermal profiles are given to the DNN sub-network whereas, the CNN sub-network is fed with the thermal profiles in the feature fusion network. Experimentation is carried out over carbon fiber reinforced polymer (CFRP) sample with artificially drilled flat bottom holes excited by quadratic frequency-modulated optical stimulus. Experimental results showed that the feature fusion enhanced the defect detection capability compared to the local networks with a significant increment in signal-to-noise ratio, accuracy, and F-score.

Detection of Early Core Browning in Pears Based on Statistical Features in Vibro-Acoustic Signals

The purpose of this work was to achieve non-destructive detection of early core browning in Korla pear using the vibro-acoustic method. The response signals of a piezoelectric sensor were analyzed to extract 18 statistical features. Three types of feature sets were then designed: time-domain feature set (11 features), frequency-domain feature set (7 features), and combined feature set (18 features). The minimum number of features in each feature set was selected using distance evaluation technology to train classifiers based on support vector machine. Two classifiers were constructed for disorder detection: a browning classifier for moderate disorder and a slight browning classifier for the mild disorder. The browning classifier obtained a high overall accuracy of 93.9% with three time-domain features and one frequency-domain feature. The overall accuracy of the slight browning classifier was 86.4% using two time-domain features and one frequency-domain feature. For these two classifiers, the F1 values from confusion matrix analysis were 85.9% and 93.5%, respectively. Therefore, our constructed classifiers could be used in the application of the vibro-acoustic method for the detection of internal browning in pears.

Feature Engineering for Human Activity Recognition

Human activity recognition (HAR) techniques can significantly contribute to the enhancement of health and life care systems for elderly people. These techniques, which generally operate on data collected from wearable sensors or those embedded in most smart phones, have therefore attracted increasing interest recently. In this paper, a random forest-based classifier for human activity recognition is proposed. The classifier is trained using a set of time-domain features extracted from raw sensor data after being segmented into windows of 5 seconds duration. A detailed study of model parameter selection is presented using the statistical t-test. Several simulation experiments are conducted on the WHARF accelerometer benchmark dataset, to compare the performance of the proposed classifier to support vector machines (SVM) and Artificial Neural Network (ANN). The proposed model shows high recognition rates for different activities in the WHARF dataset compared to other classifiers using the same set of features. Furthermore, it achieves an overall average precision of 86.1% outperforming the recognition rate of 79.1% reported in the literature using Convolution Neural Networks (CNN) for the WHARF dataset. From a practical point of view, the proposed model is simple and efficient. Therefore, it is expected to be suitable for implementation in hand-held devices such as smart phones with their limited memory and computational resources.