Statistical Feature Extraction Research Articles

Physics-informed neural network for lithium-ion battery degradation stable modeling and prognosis

Accurate state-of-health (SOH) estimation is critical for reliable and safe operation of lithium-ion batteries. However, reliable and stable battery SOH estimation remains challenging due to diverse battery types and operating conditions. In this paper, we propose a physics-informed neural network (PINN) for accurate and stable estimation of battery SOH. Specifically, we model the attributes that affect the battery degradation from the perspective of empirical degradation and state space equations, and utilize neural networks to capture battery degradation dynamics. A general feature extraction method is designed to extract statistical features from a short period of data before the battery is fully charged, enabling our method applicable to different battery types and charge/discharge protocols. Additionally, we generate a comprehensive dataset consisting of 55 lithium-nickel-cobalt-manganese-oxide (NCM) batteries. Combined with three other datasets from different manufacturers, we use a total of 387 batteries with 310,705 samples to validate our method. The mean absolute percentage error (MAPE) is 0.87%. Our proposed PINN has demonstrated remarkable performance in regular experiments, small sample experiments, and transfer experiments when compared to alternative neural networks. This study highlights the promise of physics-informed machine learning for battery degradation modeling and SOH estimation.

An intelligent model to decode students’ behavioral states in physical education using back propagation neural network and Hidden Markov Model

This paper highlights the need for intelligent analysis of students’ behavioral states in physical education tasks. The hand-ring inertial data is used to identify students’ motion sequence states. First, statistical feature extraction is performed based on the acceleration and angular velocity data collected from the bracelet. After completing the filtering and noise reduction of the data, we perform feature extraction by Back Propagation Neural Network (BPNN) and use the sliding window method for analysis. Finally, the classification capability of the model sequence is enhanced by the Hidden Markov Model (HMM). The experimental results indicate that the classification accuracy of student action sequences in physical education exceeds 96% after optimization by the HMM method. This provides intelligent means and new ideas for future student state recognition in physical education and teaching reform.

Hole Edge Metrology and Inspection by Edge Diffractometry

Abstract This paper introduces a novel hole edge inspection and metrology technology by edge diffractometry, which occurs when light interacts with the hole edge. The proposed method allows for simultaneous characterization of hole part error and edge roughness conditions. Edge diffraction occurs as light bends at a sharp edge. Such a diffractive fringe pattern, the so-called interferogram, is directly related to edge geometry and roughness. Image-based diffractometry inspection technology was developed to capture the diffractive fringe patterns. The collected fringe patterns were analyzed through statistical feature extraction methods, and numerical results such as roundness index, concentricity, and via edge roughness (VER) were obtained. Through-focus scanning optical microscopy (TSOM) was also utilized to perform three-dimensional characterization of the hole features along the depth direction. As a result, the proposed method could characterize hole part error and evaluate its roughness conditions. This study showed the potential to be adapted for automatic optical inspection for advancing microelectronics and semiconductor packaging technology.

Machine learning for monitoring hobbing tool health in CNC hobbing machine

Utilizing Machine Learning (ML) to oversee the status of hobbing cutters aims to enhance the gear manufacturing process’s effectiveness, output, and quality. Manufacturers can proactively enact measures to optimize tool performance and minimize downtime by conducting precise real-time assessments of hobbing cutter conditions. This proactive approach contributes to heightened product quality and decreased production costs. This study introduces an innovative condition monitoring system utilizing a Machine Learning approach. A Failure Mode and Effect Analysis (FMEA) were executed to gauge the severity of failures in hobbing cutters of Computer Numerical Control (CNC) Hobbing Machine, and the Risk Probability Number (RPN) was computed. This numerical value aids in prioritizing preventive measures by concentrating on failures with the most substantial potential impact. Failures with high RPN numbers were considered to implement the Machine Learning approach and artificial faults were induced in the hobbing cutter. Vibration signals (displacement, velocity, and acceleration) were then measured using a commercial high-capacity and high-frequency range Data Acquisition System (DAQ). The analysis covered operating parameters such as speed (ranging from 35 to 45 rpm), feed (ranging from 0.6 to 1 mm/rev), and depth of cut (6.8 mm). MATLAB code and script were employed to extract statistical features. These features were subsequently utilized to train seven algorithms (Decision Tree, Naive Bayes, Support Vector Machine (SVM), Efficient Linear, Kernel, Ensemble and Neural Network) as well as the application of Bayesian optimization for hyperparameter tuning and model evaluation were done. Amongst these algorithms, J48 Decision tree (DT) algorithm demonstrated impeccable accuracy, correctly classifying 100% of instances in the provided dataset. These algorithms stand out for their accuracy and efficiency in building, making them well-suited for this purpose. Based on ML model performance, it is recommended to employ J48 Decision Tree Model for the condition monitoring of a CNC hobbing cutter. The emerging confusion matrix was crucial in creating a condition monitoring system. This system can analyze statistical features extracted from vibration signals to assess the health of the cutter and classify it accordingly. The system alerts the operator when a hobbing cutter approaches a worn or damaged condition, enabling timely replacement before any issues arise.

An accurate hypertension detection model based on a new odd-even pattern using ballistocardiograph signals

BackgroundBallistocardiography (BCG) signals play an important role in identifying hypertension. These signals coupled with machine learning methods can be used for detecting hypertension. In this study, we presented a novel feature engineering architecture to detect hypertension using BCG signals. Methods and architectureIn this work, we used a publicly available BCG signal dataset to develop the novel model. Our model consists of a unique feature extraction process, named the Odd-Even Pattern (OEP). This technique generates three specific feature vectors based on alternating odd and even indices when fed with BCG signals. We employed singular pooling to address the shortcomings of OEP in extracting advanced-level features. This technique combines singular value decomposition with statistical feature extraction to capture the intricate details from the BCG signals. We obtained 14 features by merging OEP with statistical features. We then employed two advanced feature selectors yielding 28 selected feature vectors, hence the model is named as OddEven28. The classification is done using k-nearest neighbors (kNN) algorithm, along with iterative majority voting. ResultsOur proposed OddEven28 model achieved a classification accuracy of 97.78% using six different feature vectors on the dataset. ConclusionsOur developed OddEven28 architecture performed better than the deep learning-based models developed for the automated detection of hypertension using BCG signals. The robustness of the model can be improved by training with huge diverse data obtained from various centers.

Gender classification based on gait analysis using ultrawide band radar augmented with artificial intelligence

The identification of individuals based on their walking patterns, also known as gait recognition, has garnered considerable interest as a biometric trait. The use of gait patterns for gender classification has emerged as a significant research domain with diverse applications across multiple fields. The present investigation centers on the classification of gender based on gait utilizing data from Ultra-wide band radar. A total of 181 participants were included in the study, and data was gathered using Ultra-wide band radar technology. This study investigates various preprocessing techniques, feature extraction methods, and dimensionality reduction approaches to efficiently process Ultra-wide band radar data. The data quality is improved through the utilization of a two-pulse canceller and discrete wavelet transform. The hybrid feature dataset is generated through the creation of gray-level co-occurrence matrices and subsequent extraction of statistical features. Principal Component Analysis is utilized for dimensionality reduction, and prediction probabilities are incorporated as features for classification optimization. The present study employs k-fold cross-validation to train and assess machine learning classifiers, Decision Tree, Random Forest, Support Vector Machine, Logistic Regression, Multi-Layer Perceptron, K-Nearest Neighbors, and Extra Tree Classifier. The Multilayer Perceptron exhibits superior performance, achieving an accuracy of 0.936. The Support Vector Machine and k-Nearest Neighbors classifiers closely trail behind, both achieving an accuracy of 0.934. This research is of the utmost importance due to its capacity to offer solutions to crucial problems in multiple domains. The findings indicate that the utilization of UWB radar data for gait-based gender classification holds promise in diverse domains, including biometrics, surveillance, and healthcare. The present study makes a valuable contribution to the progress of gender classification systems that rely on gait patterns.

A machine learning-based data-driven approach to Alzheimer’s disease diagnosis using statistical and harmony search methods

Alzheimer’s disease (AD) is the most prevalent brain disorder which affects millions of people worldwide. Early detection is crucial for possible treatment. In this regard, machine learning (ML) approaches are widely utilized for AD detection. In this paper, we propose an ML-based method that drastically reduces the dimensionality of features while maintaining the relevant features and boosting the overall performance. To remove irrelevant features, first statistical feature extraction method is applied, and then further reduction among remaining features is applied by utilizing the harmony search method (HSM). The selected features are the most informative features that are fed to the different classifiers. To test the effectiveness of the proposed method, we deployed three classification techniques including support vector machine (SVM), k-nearest neighbor (k-NN), and decision tree (DT). The experimental results show that the proposed method has a higher performance while decreasing the dimensionality of feature space. To guarantee that the performance of the proposed method is accurate, we applied an ensemble of three classifiers (SVM, KNN, and DT) for classification. The results of the proposed method verify that this method can be successfully deployed for AD detection, due to its high performance and low dimensional features, and can help improve the accuracy and efficiency of Alzheimer’s disease diagnosis. The proposed method demonstrated a significant improvement, achieving high performance in AD/HC classification, with accuracy, sensitivity, specificity, F1-score, MCC, and Cohen’s Kappa rates reaching 95.5%, 97%, 94%, 95.56%, 0.9104, and 0.9109, respectively. AD/HC classification displayed the highest performance. Additionally, in the more challenging pMCI/sMCI classification, the method achieved an accuracy of 78.50%, sensitivity of 84.00%, specificity of 73.00%, F1-score of 79.62%, MCC of 0.57, and Cohen’s Kappa of 0.59.

Unlocking stroke prediction: Harnessing projection-based statistical feature extraction with ML algorithms

Non-communicable diseases, such as cardiovascular disease, cancer, chronic respiratory diseases, and diabetes, are responsible for approximately 71% of all deaths worldwide. Stroke, a cerebrovascular disorder, is one of the leading contributors to this burden among the top three causes of death. Early recognition of symptoms can encourage a balanced lifestyle and provide essential information for stroke prediction. To identify a stroke patient and risk factors, machine learning (ML) is a key tool for physicians. Due to different data measurement scales and their probability distributional assumptions, ML-based algorithms struggle to detect risk factors. Furthermore, when dealing with risk factors with high-dimensional features, learning algorithms struggle with complexity. In this study, rigorous statistical tests are used to identify risk factors, and PCA-FA (Integration of Principal Components and Factors) and FPCA (Factor Based PCA) approaches are proposed for projecting suitable feature representations for improving learning algorithm performances. The study dataset consists of different clinical, lifestyle, and genetic attributes, allowing for a comprehensive analysis of potential risk factors associated with stroke, which contains 5110 patient records. Using significant test (P-value <0.05), chi-square and independent sample t-test identified age, heart_disease, hypertension, work_type, ever_married, bmi, and smoking_status as risk factors for stroke. To develop the predicting model with proposed feature extraction techniques, random forests approach provides the best results when utilizing the PCA-FA method. The best accuracy rate for this approach is 92.55%, while the AUC score is 98.15%. The prediction accuracy has increased from 2.19% to 19.03% compared to the existing work. Additionally, the prediction results is robustified and reproducible with a stacking ensemble-based classification algorithm. We also developed a web-based application to help doctors diagnose stroke risk based on the findings of this study, which could be used as an additional tool to help doctors diagnose.

Deep neural networks based method to islanding detection for multi-sources microgrid

One of the significant issues in the field of microgrids is their islanding, where in many cases, the lack of awareness of microgrid islanding can lead to interference in the protective and control functions of the microgrid. Therefore, the accurate detection of microgrid islanding is of utmost importance. In this article, a method based on deep neural networks is presented. The proposed approach utilizes terminal parameters of microgrid resources, such as sequence current components, voltage, and other parameters, to detect islanding. Various operational states of the microgrid are simulated offline as standard test cases, and the parameters of each of them are recorded for later use. These data are then used to extract statistical features using discrete wavelet transform. Subsequently, the extracted features are fed into deep neural networks for training, and the training and evaluation results demonstrate an accuracy of over 99% for the proposed method in terms of precision and reliability. Furthermore, the accuracy of the proposed method is compared with some similar approaches for islanding detection.

Performance evaluation of machine learning algorithms and impact of activation functions in artificial neural network classifier for bearing fault diagnosis

Machine learning algorithms are used to identify the bearing condition. In this work, different machine learning techniques such as decision tree, logistic regression, support vector machine (SVM), and artificial neural network (ANN) are used and compared to find healthy and faulty conditions of the bearing. The identification of the condition of the bearing is based on vibrations recorded using Fast Fourier Transform Analyzer. The vibration data recorded for the bearings have been used to categorize the condition of the bearing as healthy or faulty by applying machine learning techniques. The dataset of healthy and faulty bearings is collected using a four-channel Fast Fourier Transform Analyzer (FFT) analyzer. However, the statistical feature extraction technique has been used to evaluate the accuracy and performance of artificial neural network, support vector machine, logistic regression, and decision tree algorithms based on their classification accuracy and total costs. The result of the work reveals that the performance of the activation function based artificial neural network (ANN-AF) and SVM algorithm is better than logistic regression and decision tree models. However, it is observed that the use of appropriate activation functions within the ANN-AF technique improves the accuracy of the machine learning model.

Analyzing categorical time series with the R package ctsfeatures

Time series data are ubiquitous nowadays. Whereas most of the literature on the topic deals with real-valued time series, categorical time series have received much less attention. However, the development of data mining techniques for this kind of data has substantially increased in recent years. The R package ctsfeatures offers users a set of useful tools for analyzing categorical time series. In particular, several functions allowing the extraction of well-known statistical features and the construction of illustrative graphs describing underlying temporal patterns are provided in the package. The output of some functions can be employed to perform traditional machine learning tasks including clustering, classification and outlier detection. The package also includes two datasets of biological sequences introduced in the literature for clustering purposes, one dataset of sleep stages, and three interesting synthetic databases. In this work, the main characteristics of the package are described and its use is illustrated through various examples. Practitioners from a wide variety of fields could benefit from the valuable tools provided by ctsfeatures.

Iris-Based Biometric Identification Using a Combination of the Right - Left Iris Statistical Features

A combination between the information extracted for both right iris and left iris could increase the efficacy of the biometric recognition systems. In this paper, we propose a biometric identification method based on density of image patterns extracted from human iris images and the combination and comparison of the right iris and the left iris characteristics. The density of the patters approach for processed images can be a new biometric feature used to implement a biometric recognition system with high performance when a small feature dimension is used. In this way, we can maximize the retention of the effective biometric information. The experiments were conducted on the MMU Iris Database containing 225 images of the left eye and 225 images of the right eye. Two morphological Top-hat and Hit or Miss transforms were implemented to find out the particular pattern of pixels. They allow for the enhancement of detail in images. Then, a statistical feature extraction technique is employed to derive the density of the patterns in morphological transformed images. To assess the density of the patterns differences between the right and left iris data groups, the Pearson’s correlation coefficient (PCC) is computed. We reported very good results with a PCC of 0.6164 (strong and positive correlation) for Top-hat morphological operation whilst the Hit or Miss transform returns a PCC of 0.0127 so there is no relationship between the density of the patterns in the right and left irises.

Orthogonal projection based statistical feature extraction for continuous process monitoring

Multivariate statistical techniques have been widely applied in industrial processes to detect abnormal behaviors, while their performance could be unsatisfactory due to insufficient extraction of complex data characteristics. A method named Orthogonal transformed statistics Mahalanobis distance (OTSMD) is developed to handle this issue. As a feature-based method, OTSMD simultaneously considers various data characteristics through monitoring statistical features of process variables. Orthogonal transformed components (OTCs) are first calculated to capture variable correlation, and a set of statistical features is determined to extract other crucial characteristics, especially for the process non-stationarity. Statistical features of OTCs, which reveals implied process information, are continuously obtained using a sliding window, and a Mahalanobis distance index is utilized for fault detection. Compared with existing methods, OTSMD extracts data characteristics more comprehensively with a lower dimension, making it more effective in monitoring various faults. The results are illustrated through a numerical example, and two chemical industrial processes.

Geotechnical characterisation of coal spoil piles using high-resolution optical and multispectral data: A machine learning approach

Geotechnical characterisation of spoil piles has traditionally relied on the expertise of field specialists, which can be both hazardous and time-consuming. Although unmanned aerial vehicles (UAV) show promise as a remote sensing tool in various applications; accurately segmenting and classifying very high-resolution remote sensing images of heterogeneous terrains, such as mining spoil piles with irregular morphologies, presents significant challenges. The proposed method adopts a robust approach that combines morphology-based segmentation, as well as spectral, textural, structural, and statistical feature extraction techniques to overcome the difficulties associated with spoil pile characterisation. Additionally, it incorporates minimum redundancy maximum relevance (mRMR) based feature selection and machine learning-based classification. This automated characterisation will serve as a proactive tool for dump stability assessment, providing crucial data for improved stability models and contributing to a greener and more responsible mining industry .

A domain adaptation network with feature scale preservation for remaining useful life prediction of rolling bearings under variable operating conditions

Transfer learning and domain adaptation (DA) methods have been utilized in bearing prognostic and health management, but most of the current DA methods do not take into account the feature scale change of degraded features when aligning the feature distribution, and these methods are more suitable for the classification problem, which is more robust to the feature scale change. However, they perform poorly in regression problems. In addition, most of the remaining useful life (RUL) prediction methods require preprocessing such as statistical feature extraction on the signal, which makes the prediction process complicated. To solve the above problems, a DA method based on the representation subspace distance (RSD) is proposed for predicting the bearing RUL under different operating conditions. First, the proposed convolutional neural network (CNN) self-attention (SA) long short term memory network model is utilized to extract the deep features from the original signal, which overcomes the limitations of the CNN in extracting time series. Then, the RSD in the Riemannian geometry of the Grassmann manifold is proposed as a domain transfer loss to learn domain invariant features. The modified method can align the feature distribution of the source domain and the target domain without changing the feature scale. At the same time, the bases mismatch penalization is introduced to avoid destroying the semantic information of the features in the process of domain alignment. Finally, the effectiveness of the proposed method is verified by experiments on four types of transfer tasks, and its superiority is also demonstrated by comparison with other advanced methods.

A New Framework Based on Supervised Joint Distribution Adaptation for Bearing Fault Diagnosis across Diverse Working Conditions

To address the degradation of diagnostic performance due to data distribution differences and the scarcity of labeled fault data, this study has focused on transfer learning-based cross-domain fault diagnosis, which attracts considerable attention. However, deep transfer learning-based methods often present a challenge due to their time-consuming and costly nature, particularly in tuning hyperparameters. For this issue, on the basis of classical features-based transfer learning method, this study introduces a new framework for bearing fault diagnosis based on supervised joint distribution adaptation and feature refinement. It first utilizes ensemble empirical mode decomposition to process raw signals, and statistical features extraction is implemented. Then, a new feature refinement module is designed to refine domain adaptation features from high-dimensional feature set by evaluating the fault distinguishability and working-condition invariance of feature data. Next, it proposes a supervised joint distribution adaptation method to conduct improved joint distribution alignment that preserves neighborhood relationships within a manifold subspace. Finally, an adaptive classifier is trained to predict fault labels of feature data across varying working conditions. To prove the cross-domain fault diagnosis performance and superiority of the proposed methods, two bearing datasets are applied for experiments, and the experimental results verify that the model built by the proposed framework can achieve desirable diagnosis performance under different working conditions and that it apparently outperforms comparative models.

Certain Investigations on Feature Selection Technique Using Artificial Immune Systems for EEG Color Visualization Classification

This paper aims to extract and select the significant features of electroencephalogram (EEG) signals and classify the visual stimulation of distinct colors. In this work, a novel method for selecting distinct colors using EEG signals called affinity artificial immune and Daubechies wavelet time-based learning (AAIDWTL) is proposed. Initially, the EEG signals were collected in a controlled environment and an in-built band-pass filter was applied to remove the artifacts. The filtered signals were converted into frequency domain signals using least squarebased short-term Fourier transform. After that, by utilizing Daubechies wavelet statistical time-based feature extraction model the time domain features were extracted. Followed by, computationally efficient features were selected using an affinity artificial immune-based feature selection model. The selected features were classified using a polynomial kernel multiclass classification-based machine learning algorithm and achieved an accuracy of 97.5% when compared with other methods like linear discriminant analysis (LDA) which obtained only 92%. Furthermore, while utilizing the proposed method classification time was considerably less when compared to LDA. The experimental result shows that the proposed color stimulation of the EEG signals method achieved greater improvement in terms of both classification time and classification accuracy with a minimum false positive rate.

RPA Technology Enables Highly Automated Development of Corporate Financial Accounting Processes

Abstract The article discusses the application of RPA (Robotic Process Automation) technology in financial management and demonstrates its remarkable effectiveness in improving efficiency and reducing costs. The article establishes an automated accounting method for enterprise finance using RPA technology through statistical analysis, classification processing, and feature extraction of enterprise financial data. The consistency check method is used to verify the fit and reasonableness of the method. Practically applied to the financial accounting process of Enterprise A, the method dramatically reduces the time for reconciling bonds and funds from 60 minutes to 2 seconds, significantly improving the accounting efficiency. In addition, the method ensured that the accuracy of statement filling reached 100% and reduced the labor cost of enterprise accounting from 17.58 million yuan to 6.88 million yuan, a reduction of 2.56 times. The article emphasizes that the established financial accounting method increases the enterprise’s profitability and lays a solid foundation for the development of financial accounting automation.

Classification of six banana ripeness levels based on statistical features on machine learning approach

&lt;span&gt;Banana plants are often cultivated because they have many benefits. In producing, we need to maintain the quality of bananas by looking at banana ripeness levels before being distributed to markets. The level of banana ripeness is related to marketing reach. If the marketing reach is far, bananas should be harvested when the ripeness level of bananas is still relatively low. A system that can classify the degree of ripeness of bananas can help overcome this problem. In this study, our dataset includes 6 ripeness levels of bananas, more than in previous related studies. Furthermore, we use the statistical features extraction method to find the parameters that affect the level of banana ripeness, considering the texture and color of the banana peel which determines the level of ripeness visually. The extraction used is features extraction based on a histogram, then we employ four features, i.e., mean, skewness, energy descriptor, and smoothness, generated from the image dataset. In the next stage, we perform classification based on the features that have been obtained. In this study, we use Naive Bayes classifier and support vector machine (SVM) algorithms. Based on the result of this research, the best performance is the Naive Bayes classifier, with an accuracy is 86.67%, a weighted average precision of 83.55%, and a weighted average recall of 86.67%.&lt;/span&gt;

Motor and bearing multi-fault diagnosis with sound quality metrics

The occurrence of multi-faults is more critical as compared to single faults, and it creates unfavourable working conditions. The number of contact-based sensors and their placement plays a crucial role in diagnosing the multi-faults. The use of non-contact-based acoustic sensors could be a cost-effective potential solution to this problem. In this work, two rotating components, a motor and a separate bearing of a system situated away from the motor, were considered for forming system-level multi-fault conditions. The acoustic signature of various multi-fault conditions was acquired at different rotational speeds. Thereafter, a moving window was employed on the acquired raw acoustic signals to extract statistical and sound quality features. A feature fusion approach was implemented on the extracted features, and a machine learning model was trained with optimised hyperparameters. A comparative study was performed to analyse the model's performance, which was trained with the extracted statistical, sound quality and fused features. It was found that the proposed feature fusion-based methodology was helpful for diagnosing multi-faults with acoustic signatures. The importance of psychoacoustical features on model performance was also observed. It mitigated the complexity of mounting the sensors and showed the potential usage of sound quality metrics for multi-fault diagnosis.