Principle Component Analysis Method Research Articles

Leveraging Machine Learning for Advanced Nanoscale X-ray Analysis: Unmixing Multicomponent Signals and Enhancing Chemical Quantification.

Energy dispersive X-ray (EDX) spectroscopy in the transmission electron microscope is a key tool for nanomaterials analysis, providing a direct link between spatial and chemical information. However, using it for precisely determining chemical compositions presents challenges of noisy data from low X-ray yields and mixed signals from phases that overlap along the electron beam trajectory. Here, we introduce a novel method, non-negative matrix factorization based pan-sharpening (PSNMF), to address these limitations. Leveraging the Poisson nature of EDX spectral noise and binning operations, PSNMF retrieves high-quality phase spectral and spatial signatures via consecutive factorizations. After validating PSNMF with synthetic data sets of different noise levels, we illustrate its effectiveness on two distinct experimental cases: a nanomineralogical lamella, and supported catalytic nanoparticles. Not only does PSNMF obtain accurate phase signatures, but data sets reconstructed from the outputs have demonstrably lower noise and better fidelity than from the benchmark denoising method of principle component analysis.

Assessment of Eco-Environmental Vulnerability Using Remote Sensing and GIS Tools in Maharashtra Region, India

Maharashtra region is prone to various disasters such as drought, floods, cyclones and earthquake and has been exposed to extreme weather events like dry spells. Communities within these dry lands are poor and face extreme conditions of water stress. This study has been carried out to analyze and quantify climatic and anthropogenic effect on eco-environmental vulnerability dynamic change. To achieve that a numerical model is set up, consisting of eight factors that are elevation, land use, drought, slope, NDVI, soil-type, soil erosion (water), and population density index & has been evaluated using the method of spatial principle component analysis (SPCA) on Remote Sensing and GIS platform. The integrated eco-environmental vulnerability index (EVI) of study area is estimated to analyse spatial-temporal dynamic vulnerability changes in the 11 years gap from 2000 and 2011. The results show that the study area has become eco-environmental vulnerable slightly (about 80% of the region) with an increased eco-environmental vulnerability integrated index (EVSI) value by more than 50% (i.e., about 74%) and the driving force of dynamic change is mainly caused by socio-economic activities. In addition the estimation has been regionalized into thirty-four districts to serve as a base for decision-making for eco-environmental recovering and rebuilding. It is found that the most vulnerable district in 2011 is Ratnagiri and the least one is Sangli. There are nine districts which shows more than 100% increase in EVSI value, with the highest increase in Hingoli(100.65%), indicating that the districts have become most environmental vulnerable in the study-period. The research concludes that the method, supported by G.I.S using SPCA can’t only represent distinctly the input spatial distribution of plain-mountain-belt feature, but also respect the whole river-valley as a single unit.

Has the Consumer Perception Changed? An Overview of Consumer Behavior towards Organized Retail Sectors in Kolkata, India

The research paper focuses on the change of consumers’ perception towards purchasing different commodities in the organized retail outlets of Kolkata. The study revealed that the behaviour of a consumer towards a particular outlet mainly depends upon factors like product availability, spending pattern, sales man services, store layout etc. The research shows that currently a large section of consumers is attracted towards the organised retail outlets because of wide range of services like variety of payment options provided by the store, visibility of the billing counter, wide range of brands availability, quality of the products, and better service at the store have majorly attributed to preference of organized retail store over the unorganized outlets. Also, parameters like visibility of display of MRP, discount offers, quality of service at the organized store, trolley service facilities etc. are some of the reasons of inclination of consumers towards organized retail stores as compared to traditional retail formats. The study has found mostly the younger generation consumers are tending to have a tendency towards making new experiences mainly for groceries and have an urge to make spontaneous purchases compared to older generation of consumers. Apart from the positive aspects, the study also observed a bunch of constraints that restricts a consumer from going to an organized retail store which are the factors like lower budget, inability of the lower income group of people due to financial constraints, lack of advertisements, unavailability of low-priced products, most of the necessary products being easily available in the traditional markets, illiteracy etc. are some of the major constraints found in the study which have contributed to 19.48% of variance and has been further renamed as consumers perception by using the Principle Component Analysis (PCA) method.

Causality-Based PCA Methods for Condition Modeling of Mechatronic Systems

This paper investigates the feature dimensionality reduction problem of high-dimensional data in condition modelling of complex mechatronic systems, aiming at improving performance, robustness and interpretability of classical principle component analysis (PCA) methods. First, the causal discovery algorithms are adopted to construct the corresponding causal network among monitoring variables. Then, two causality-based PCA methods are proposed, based on the corresponding qualitative reachability matrix and quantitative reachability strength matrix, respectively. Average treatment effect (ATE) values are considered as the quantitative measure of the causalities in a reachability strength matrix. The robustness and effectiveness of the proposed methods are verified on a synthetic dataset and four public datasets in comparison with two benchmark methods, i.e. PCA and autoencoder (AE). The proposed methods are also adopted in a high-speed train braking system for extracting informative features from the monitoring variables.

Hybrid model for brain age prediction on MRI images with modified texture features

Machine learning (ML) and diffusion magnetic resonance imaging (dMRI) approaches can be employed to identify individual brain aging conditions in both healthy people and patients with brain diseases. This research establishes a new brain age prediction system, which comprises different phases like pre-processing, feature extraction, feature selection, and prediction. The original magnetic resonance imaging (MRI) picture is first carried out in a pre-processing stage, which enhances image quality as well as visual appearances. Following that, the shape, intensity, as well as texture-based characteristics are retrieved. Furthermore, texture features like modified entropy, as well as modified correlation are retrieved along with the traditional features. Then, a modified principle component analysis (PCA) method is utilized for feature selection stage. In the prediction phase, hybrid classifiers such as multi-layer perception (MLP) as well as long short term memory (LSTM) are used. The weight of LSTM and MLP is optimally tuned with the aid of an implemented self-updated red deer algorithm (SURDA) model to make an accurate prediction. Moreover, the suggested work outperforms other traditional works in terms of MSE, MAPE, MSLE, and MAE, correspondingly.

De-noising of galaxy optical spectra with autoencoders

ABSTRACT Optical spectra contain a wealth of information about the physical properties and formation histories of galaxies. Often though, spectra are too noisy for this information to be accurately retrieved. In this study, we explore how machine learning methods can be used to de-noise spectra and increase the amount of information we can gain without having to turn to sample averaging methods such as spectral stacking. Using machine learning methods trained on noise-added spectra – Sloan Digital Sky Survey (SDSS) spectra with Gaussian noise added – we investigate methods of maximizing the information we can gain from these spectra, in particular from emission lines, such that more detailed analysis can be performed. We produce a variational autoencoder (VAE) model, and apply it on a sample of noise-added spectra. Compared to the flux measured in the original SDSS spectra, the model values are accurate within 0.3–0.5 dex, depending on the specific spectral line and signal-to-noise ratio. Overall, the VAE performs better than a principal component analysis method, in terms of reconstruction loss and accuracy of the recovered line fluxes. To demonstrate the applicability and usefulness of the method in the context of large optical spectroscopy surveys, we simulate a population of spectra with noise similar to that in galaxies at z = 0.1 observed by the Dark Energy Spectroscopic Instrument (DESI). We show that we can recover the shape and scatter of the mass–metallicity relation in this ‘DESI-like’ sample, in a way that is not possible without the VAE-assisted de-noising.

Industrial Process Monitoring Based on Parallel Global-Local Preserving Projection with Mutual Information

This paper proposes a parallel monitoring method for plant-wide processes by integrating mutual information and Bayesian inference into a global-local preserving projections (GLPP)-based multi-block framework. Unlike traditional multivariate statistic process monitoring (MSPM) methods, the proposed MI-PGLPP method transforms plant-wide monitoring into several sub-block monitoringtasks by fully taking advantage of a parallel distributed framework. First, the original datasets of the process are divided into a group of data blocks by quantifying the mutual information of process variables. The block indexes of new data are generated automatically. Second, each data block is modeled by the GLPP method. The variable information and local structure are well preserved during the whole projection. Third, Bayesian inference is introduced to generate final statistics of the process by the probability framework. To illustrate the algorithm performance, a detailed case study is performed on the Tennessee Eastman process. Compared with the principle component analysis and GLPP-based method, the proposed MI-PGLPP provides higher FDRs and superior performance for plant-wide process monitoring.

URBAN MOBILITY DURING THE PANDEMIC IN YOGYAKARTA CITY

The urban situation has changed during COVID-19 pandemic since 2020. The swift of mobility behavior happened in all over the city in the world. The WHO’s suggestion to maintain NPIs; the physical distancing is a push factor for people to reduce traveling and mobility. The slogan ‘stay at home’ became part of the new normal until nowadays, because publication of it was spread on all media to reach out the society. Yogyakarta City responds to the pandemic issue with the same strategy, minimizing mobility and travel. However, there is uniqueness in the practice on minimizing mobility and travel in 2020 and 2021. In 2020, the control of mobility and travel is based on community and in 2021 the control of mobility and travel is based on government policy. The security level of travel in 2020 is housing and urban kampung, and in 2021 is on the city level. The research is focused on the perception of respondent on minimizing travel and mobility between these two years. The data was analyzed using the Principle Component Analysis method. The quantitative result will be compared with observation, so it shows the important value to respond during uncertainty on travel and mobility issue. The study will give how local wisdom and public policy collaborate to produce urban resilience in Yogyakarta City.

Automated facial expression recognition using novel textural transformation

Facial expressions demonstrate the important information about our emotions and show the real intentions. In this study, a novel texture transformation method using graph structures is presented for facial expression recognition. Our proposed method consists of five steps. First the face image is segmented and resized. Then the proposed graph-based texture transformation is used as feature extractor. The exemplar feature extraction is performed using the proposed deep graph texture transformation. The extracted features are concatenated to obtain one dimensional feature set. This feature set is subjected to maximum pooling and principle component analysis methods to reduce the number of features. These reduced features are fed to classifiers and we have obtained the highest classification accuracy of 97.09% and 99.25% for JAFFE and TFEID datasets respectively Moreover, we have used CK + dataset to obtain comparison results and our textural transformation based model yielded 100% classification accuracy on the CK + dataset. The proposed method has the potential to be employed for security applications like counter terrorism, day care, residential security, ATM machine and voter verification.

Fault diagnosis of H-bridge cascaded five-level inverter based on improved support vector machine with gray wolf algorithm

The reliability of power electronics is very important to the safety and efficiency operation of the multilevel inverters, and it is very necessary to quickly detect and locate the faults. To solve the difficulty in locating open-circuit faults and improve the diagnosis accuracy for the H-bridge cascaded five-level inverters, an improved support vector machine (SVM) with gray wolf (GWO) optimization method has been elaborated and proposed in this paper especially for single switch faults and double switches faults in one phase. Firstly, the output voltage signals are selected as the sampling data, and whose eigenvectors have been taken as the input signals of the wavelet packet energy entropy method, and the dimensionality of these extracted eigenvectors is furtherly reduced with a probabilistic principle component analysis (PPCA) method to reduce the calculations and improve the diagnosis efficiency. Then, the SVM method is used for the fault classification, during which the GWO optimization has been introduced to find the optimal value of SVM, so as to improve the diagnostic efficiency. Experimental results show that this diagnosis method can not only achieve higher diagnostic accuracy, smaller identification error, and has greater efficiency compared with ordinary SVM or some other traditional diagnosis methods and other intelligent optimization algorithms, which promotes the fault diagnosis research for these multilevel inverters.

Exploring optimal dispersion process parameters for fabrication of graphene-reinforced cement composites

Graphene and its derivatives have been studied as nano-reinforcement in cement-based composites to improve their mechanical, transport and durability characteristics. However, the agglomeration effect due to the strong van der Waals force between graphene sheets makes the uniform dispersion difficult and limit the efficacy of graphene. The goal of this paper is to determine an optimal dispersion process for incorporating graphene nanoplatelets (GNPs) into cement-based composites. Surfactant-assisted sonication was used for the dispersion of graphene sheets. The effects of sonication amplitude, sonication duration, sample volume, surfactant type, and surfactant-to-GNP ratio on the dispersion performance were studied. As for the surfactant type, two polycarboxylate superplasticizers, two ionic surfactants, and one non-ionic surfactant were considered. Taguchi method of experimental design with an orthogonal array of L25 was employed to minimize the number of experiments needed to assess the effects of selected factors on the dispersion process. The average particle size, UV–vis absorbance, residual absorbance after 24 h, and zeta potential were measured through dynamic light scattering and UV–vis spectroscopy tests. Two multi-criteria decision making methods, namely Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS) and Principle Component Analysis (PCA) methods, were then employed to determine the optimal values of experimental factors for the dispersion. Finally, graphene-mortar composites were fabricated using two different surfactants and the properties of the developed composites were evaluated. The results indicates a 32% and 52% improvement in compressive strength and volume of permeable voids of graphene-enhanced mortar composites.

Soil texture identification using LIBS data combined with machine learning algorithm

A machine learning based model is developed to classify the soil texture [Clay, Sandy clay (SC), Sandy clay loam (SCL), Sandy loam (SL) and Sand]. The model utilizes the data obtained with Laser Induced Breakdown Spectroscopy (LIBS) method. An investigation on LIBS experiment with and without black slit placed on the top of the soil surface is carried out to determine the appropriate input data format for the model. The considered input formats are raw spectrum, relative intensity (Iλ/I393nm) and relative intensity (Iλ/I553nm). The selection of I553nm as reference line during the computation of relative intensity results in projecting the reflection phenomenon at other wavelengths. Also the reflection from the soil surface is highly dependent on incident plasma emission wavelength and soil texture. The Principle Component Analysis (PCA) method is used to explore the classification power of the input data format. The input data formats are given to machine learning classifiers such as k- Nearest Neighbor (k-NN), Support Vector Machine (SVM), Decision Tree (DT), Random Forest (RB) and Naïve Bayes (NB). From the experimental analysis, it is observed that the Naïve Bayes classifier achieved higher F1-score (Clay = 0.93, SC = 1.00, SCL = 0.86, SL = 0.85 and Sand = 1.00), when it is subjected with Iλ/I553nm data format. This improvement in performance metric is mainly attributed to the consideration of plasma reflection phenomenon at longer wavelength.

Combined GRACE and GPS to Analyze the Seasonal Variation of Surface Vertical Deformation in Greenland and Its Influence

The geophysical effects are the main factor that causes the nonlinear motion of the station, and a comprehensive analysis of the relationship between the GRACE seasonal load deformation and the GPS station coordinates is helpful to study the physical mechanism that causes the nonlinear motion of the station. Aiming at the continuous GPS coordinate time series in Greenland, this paper comprehensively analyzes the correlation between GRACE seasonal load deformation and GPS station coordinates. First, in order to improve the accuracy of GPS station coordinates, the principle component analysis (PCA) method was used to eliminate the common mode error (CME) of the station coordinates. The results show that this method effectively reduces the uncertainty of the station coordinates time series. Secondly, when extracting seasonal signals, it is found that the singular spectrum (SSA) method can effectively obtain the time-varying part of seasonal signals, and its extraction effect is better than that of the least square fitting (LSF) method. Finally, the seasonal relationship between GRACE load deformation and GPS station coordinates is analyzed from the aspects of time series change, correlation, and WRMS reduction. It is found that there are differences in the amplitude and phase parts of the time series. The mean value of correlation is 0.73, the maximum reduction of WRMS is 55.20% (QAQ1 station), and the minimum is −22.69% (KMJP station), indicating that most stations mainly exhibit seasonal load deformation, while individual stations cannot effectively reflect. In addition, the influence of GRACE seasonal load deformation on the station coordinate parameters is quantitatively analyzed. The results show that the best noise model of the station is mainly WN + FN, which effectively reduces the velocity uncertainty of the station coordinate, and weakens the seasonal term oscillation.

Missing Traffic Data Imputation for Artificial Intelligence in Intelligent Transportation Systems: Review of Methods, Limitations, and Challenges

Missing data in Intelligent Transportation Systems (ITS) could lead to possible errors in the analyses of traffic data. Applying Artificial Intelligence (AI) in these circumstances can mitigate such problems. Past works focused only on specific data imputation methods, such as tensor factorization or a specific neural network model. While there are review papers covering singular topics regarding missing data, there are none in the field of traffic, to the best of our knowledge, that introduces the process of missing data collection and the viability of the traffic data collected while also broadly covering the popularly used models of recent years. This has led to non-uniformity of the terms used in missing data imputation, limited research in areas where datasets are not available, and a narrowed view of the methods used for data imputation. Hence, this paper aims to standardize the terms used in missing data classifications, look into the limitations of using available public or private datasets for urban traffic research, and discuss popular statistical and data-driven methods used by recent AI and ITS papers. It was found that tensor decomposition-based methods are the most popular for missing data imputation, followed by Generative Adversarial Networks and Graph Neural Networks, all of which rely on a large training dataset. Meanwhile, Probability Principle Component Analysis (PPCA) methods provide valuable insights via traffic analysis and are used for real-time traffic imputation. This paper also highlights the need for more efficient and reliable methods for traffic data collection, such as online APIs.

Clustering Techniques for Recommendation of Movies

A recommendation system employs a variety of algorithms to provide users with recommendations of any kind. The most well-known technique, collaborative filtering, involves users with similar preferences although it is not always as effective when dealing with large amounts of data. Improvements to this approach are required as the dataset size increases. Here, in our suggested method, we combine a hierarchical clustering methodology with a collaborative filtering algorithm for making recommendations. Additionally, the Principle Component Analysis (PCA) method is used to condense the dimensions of the data to improve the accuracy of the outcomes. The dataset will receive additional benefits from the clustering technique when using hierarchical clustering, and the PCA will help redefine the dataset by reducing its dimensionality as needed. The primary elements utilized for recommendations can be enhanced by applying the key elements of these two strategies to the conventional collaborative filtering recommendation algorithm. The suggested method will unquestionably improve the precision of the findings received from the conventional CFRA and significantly increase the effectiveness of the recommendation system. The total findings will be applied to the combined dataset of TMDB and Movie Lens, which is utilized to suggest movies to the user in accordance with the rating patterns that each individual user has generated.

An anomaly pattern detection for bridge structural response considering time-varying temperature coefficients

To improve the detection rate in the bridge anomaly detection, this study proposes an anomaly pattern detection method based on the time-varying temperature–displacement relationship. First, the empirical wavelet transform method is used to extract thermal-induced displacement components. Then, based on the daily variation of solar radiation, a pattern is defined as three groups of regression parameters associated with the temperature–displacement relationships. To measure the uncertainty in the anomaly detection, distributions of the parameters in the defined pattern are determined by Bayesian estimation. To overcome occasionality, a new index — weekly mean detection rate is defined. The effectiveness of the proposed temperature-driven anomaly pattern detection method is validated by using actual girder end displacements from a large span suspension bridge in China. As a result, the anomalous boundary condition is detected with a weekly mean detection rate of 0.91. Whilst, the weekly mean detection rate is 0.67 when applying the conventional anomaly detection method without considering the time-varying factor. Finally, the detection rate and the false detection rate of the proposed method are compared with those of cointegration and moving principle component analysis methods. The proposed method has a higher detection rate and a lower false detection rate for the anomalous boundary condition.

Blind-noise image denoising with block-matching domain transformation filtering and improved guided filtering

The adaptive block size processing method in different image areas makes block-matching and 3D-filtering (BM3D) have a very good image denoising effect. Based on these observation, in this paper, we improve BM3D in three aspects: adaptive noise variance estimation, domain transformation filtering and nonlinear filtering. First, we improve the noise-variance estimation method of principle component analysis using multilayer wavelet decomposition. Second, we propose compressive sensing based Gaussian sequence Hartley domain transform filtering to reduce noise. Finally, we perform edge-preserving smoothing on the preprocessed image using the guided filtering based on total variation. Experimental results show that the proposed denoising method can be competitive with many representative denoising methods on the evaluation criteria of PSNR. However, it is worth further research on the visual quality of denoised images.

Anomaly detection model of mooring system based on LSTM PCA method

Aiming at the difficulty of damage detection of the mooring system, this paper proposes an anomaly detection framework combined with statistical analysis and neural network methods. Specifically, based on the real design parameters of a semi-submersible platform, a hydrodynamic model was established. Taking the damage degree of 5%, 10%, 15%, and 20% as examples, the analysis of the responses was carried out when the anchor chain was damaged in different positions. And then, based on the Long Short-Term Memory (LSTM) network and normal state data, a joint prediction model of the platform's six-degree-of-freedom (6DoF) motion was established considering the marine environment loads. The motion prediction error was only 0.0078. The Principle Component Analysis (PCA) together with the prediction residual sequences was tailored and employed to build the anomaly detection model. The Upper Confident Level (UCL) of the model was also given. The recognition accuracy reached 100% which can guide the safe service of the platform.

Depth detection of spar cap defects in large-scale wind turbine blades based on a 3D heat conduction model using step heating infrared thermography

The defects dispersed in a spar cap often lead to the failure of large-scale wind turbine blades. To predict the residual service life of the blade and make the repair, it is necessary to detect the depth of spar cap defects. Step-heating thermography (SHT) is a common infrared technique in this domain. However, the existing methods of SHT on defect depth detection are generally based on 1D models, which are unable to accurately detect the depth of spar cap defects due to ignoring material anisotropy and in-plane heat flow. To improve the depth detection accuracy of spar cap defects, a 3D model based on the theory of heat transfer is established by using the equivalent source method (ESM), and a defect depth criterion is proposed based on the analytical solution of the heat conduction equation. The modeling process is as follows. The heat conduction model of SHT was established by ESM. Then, coordinate transformation, variables separation, and Laplace transformation were utilized to solve the 3D heat conduction equation. A defect depth criterion was proposed based on emerging contrast Cr. A glass fiber reinforced plastic composite plate containing 12 square flat-bottom holes with different sizes and depths was manufactured to represent a spar cap with large thermal resistance defects, such as delamination and cracks. The experimental results demonstrate the validity of the 3D model. Then, the model was applied to an on-site SHT test of a 1.5 MW wind turbine blade. The test results prove that the depth detection accuracy of spar cap defects can be significantly improved by using the 3D model. In addition, by using an improved principle component analysis (PCA) method containing a contrast enhancement factor, artifacts can be reduced and the recognition time of defects can be shortened. The 3D model provides a tool for detecting the depth of deep-lying defects in a thick composite structure, and the SHT technology is optimized by improved PCA.

An electronic nose based on carbon nanotube -titanium dioxide hybrid nanostructures for detection and discrimination of volatile organic compounds

Detection and classification of Volatile Organic Compounds (VOCs) that pose risks to human health even at very low concentrations, is a common requirement in the industry. Herein, we apply a modification to the aligned Carbon Nanotubes (CNTs) grown at low temperature (250 °C) using Plasma Enhanced Chemical Vapor Deposition (PECVD). CNTs are used to identify VOCs in an improved sensitivity condition to gas at room temperature. We improve the sensitivity and stability of carbon nanotube sensors via growing Titanium Dioxide Nanowires (TiO2-NW) on the surface of CNTs through the hydrothermal method. A threefold increase in sensitivity and a 30-second decrease in response time were observed as a result of the CNT-TiO2 sensor compared to the pristine CNT sensor. In order to obtain new data, we deposit four different electrodes on the TiO2 surface as virtual arrays of Electronic nose (E-nose). Temperature modulation and Discrete Wavelet Transform (DWT) of virtual arrays, extract new features of the E-nose data. We select 14 different features from E-nose data and apply Principle Component Analysis (PCA) method for data reduction. We categorize the extracted features by using Support Vector Machine (SVM) algorithms and data covariance properties. This paper achieves 97.5% accuracy in classifying four different VOC gases.