Least Squares Support Vector Machine Research Articles

희박한 최소 절대 편차 지지벡터기계

The support vector machine solves a quadratic programming problem with linear inequality and equality constraints. However, it is not trivial to solve the quadratic problem. The least squares support vector machine(LS-SVM) solves a linear system by equality constraints instead of inequality constraints. LS-SVM is a popular method in regression and classification problems, because it effectively solves simple linear systems. There are two issues with the LS-SVM solution : the lack of robustness to outliers and the absence of sparseness. In this paper, we propose a sparse and robust support vector machine for regression problems using the least absolute deviation support vector machine (LAD-SVM) and recursive reduced LS-SVM (RR-LS-SVM). The split-Bregman iteration gives the exact solution for the LAD-SVM problem, while RR-LS-SVM gives a sparse solution with a much smaller number of all support vectors. Numerical experiments with simulation and benchmark data demonstrate that the proposed algorithm can achieve comparable performance to other methods in terms of robustness and sparseness.

Estimation of CO2 solubility in aqueous solutions of commonly used blended amines: Application to optimised greenhouse gas capture

One of the key concerns in the 21st century, alongside the growing population, is the increase in energy consumption and the resulting global warming. The impact of CO2, a prominent greenhouse gas, has garnered significant attention in the realm of CO2 capture and gas purification. CO2 absorption can be enhanced by introducing some additives into the aqueous solution. In this study, the accuracies of some of the most up-to-date computational approaches are investigated. The employed machine learning methods are hybrid-adaptive neuro-fuzzy inference system (Hybrid-ANFIS), particle swarm optimization-adaptive neuro-fuzzy inference system (PSO-ANFIS), least-squares support vector machines (LSSVM) and genetic algorithm-radial basis function (GA-RBF). The developed models were used in estimating the solubility of CO2 in binary and ternary amines aqueous solutions. i.e. blends of monoethanolamine (MEA), triethanolamine (TEA), aminomethyl propanol (AMP), and methyldiethanolamine (MDEA). This modeling study was undertaken over relatively significant ranges of CO2 loading (mole of CO2/mole of solution) as a function of input parameters, which are 0.4–2908 kPa for pressure, 303–393.15 K for temperature, 36.22–68.89 g/mol for apparent molecular weight, and 30–55 wt % for total concentration. In this work, the validity of approaches based on different statistical graphs was investigated, and it was observed that the developed methods, especially the GA-RBF model, are highly accurate in estimating the data of interest. The obtained AARD% values for the developed models are 18.63, 8.25, 12.22, and 7.54 for Hybrid-ANFIS, PSO-ANFIS, LSSVM, and GA-RBF, respectively.

A survival prediction model based on PCA-HSIDA-LSSVM for patients with esophageal squamous cell carcinoma.

Esophageal squamous cell carcinoma (ESCC) is a type of cancer and has some of the highest rates of both incidence and mortality globally. Developing accurate models for survival prediction provides a basis clinical judgment and decision making, improving the survival status of ESCC patients. Although many predictive models have been developed, there is still lack of highly accurate survival prediction models for ESCC patients. This study proposes a novel survival prediction model for ESCC patients based on principal component analysis (PCA) and least-squares support vector machine (LSSVM) optimized by an improved dragonfly algorithm with hybrid strategy (HSIDA). The original 17 blood indicators are condensed into five new variables by PCA, reducing data dimensionality and redundancy. An improved dragonfly algorithm based on hybrid strategy is proposed, which addresses the limitations of dragonfly algorithm, such as slow convergence, low search accuracy and insufficient vitality of late search. The proposed HSIDA is used to optimize the regularization parameter and kernel parameter of LSSVM, improving the prediction accuracy of the model. The proposed model is validated on the dataset of 400 patients with ESCC in the clinical database of First Affiliated Hospital of Zhengzhou University and the State Key Laboratory of Esophageal Cancer Prevention and Control of Henan Province. The experiment results demonstrate that the proposed HSIDA-LSSVM has the best prediction performance than LSSVM, HSIDA-BP, IPSO-LSSVM, COA-LSSVM and IBA-LSSVM. The proposed model achieves the accuracy of 96.25%, sensitivity of 95.12%, specificity of 97.44%, precision of 97.50%, and F1 score of 96.30%.

Modeling and optimization of oil adsorption capacity on functionalized magnetic nanoparticles using machine learning approach

Using magnetic nanoparticles (MNPs) has emerged as a promising solution to capture oil from emulsified oily wastewater due to their high oil adsorption capacity, low toxicity, and reusability. Various factors affect the oil adsorption process using MNPs; thus, optimization of the process is required to achieve higher oil adsorption capacity. Smart models based on artificial intelligence (AI) are becoming popular as advanced computational tools to assess the non-linear relationships of variables in complex processes. In this study, least squares support vector machines (LSSVM) hybridized with the coupled simulated annealing (CSA) algorithm, adaptive network-based fuzzy inference system (ANFIS), and optimization techniques such as gene expression programming (GEP) are used to predict the oil adsorption capacity as a target variable. Oil concentration, mixing time, and MNP dosage as effective parameters are selected as input variables. After conducting experiments, 149 data points are obtained, 80 % of which is used in the training process and the remaining 20 % for the testing step. The performances of the developed models are evaluated using statistical parameters, including the coefficient of determination (R2), mean percentage error (MPE), and mean absolute percentage error (MAPE). According to the results, ANFIS and LSSVM-CSA models have a better performance than the GEP model in estimating the oil adsorption capacity with higher values of R2 (>0.99) and smaller relative errors (close to zero) for all training, testing, and total datasets. Detailed model evaluation and error analysis indicate that the LSSVM-CSA model predicts slightly better than ANFIS with the highest R2 of 0.9921 and a very small MAPE = 3.7597 % over the total dataset. Although the developed GEP model shows an acceptable prediction with R2 > 0.95, the higher distribution of relative errors of the developed model results in a larger MAPE. Moreover, the GEP model computational time is considerably greater than that of the other models. The relative importance analysis using Pearson’s and Spearman’s correlation coefficients indicates that the oil concentration and MNP dosage are the most influential variables that affect the oil adsorption capacity.

Fault diagnosis of anti-friction bearings based on Bi-dimensional ensemble local mean decomposition and optimized dynamic least square support vector machine

In order to ensure the normal operation of rotating equipment, it is very important to quickly and efficiently diagnose the faults of anti-friction bearings. Hereto, fault diagnosis of anti-friction bearings based on Bi-dimensional ensemble local mean decomposition and optimized dynamic least square support vector machine (LSSVM) is presented in this paper. Bi-dimensional ensemble local mean decomposition, an extension of ensemble local mean decomposition from one-dimensional signal processing to Bi-dimensional signal processing, is used to extract the features of anti-friction bearings. Moreover, an optimized dynamic LSSVM is used to fault diagnosis of anti-friction bearings. The experimental results show that Bi-dimensional ensemble local mean decomposition is superior to Bi-dimensional local mean decomposition, optimized dynamic LSSVM is superior to traditional LSSVM, and the proposed Bi-dimensional ensemble local mean decomposition and optimized dynamic LSSVM method is effective for fault diagnosis of anti-friction bearings.

A robust identification method for inherent parameters of lithium battery considering thermal distribution and state of charge

In charge/discharge processes, the internal resistances of lithium-ion batteries (LIBs) change dynamically with the current, temperature and state of charge (SOC). However, these resistances are usually assumed to be constant after their value is offline identified, and the influence of the factors, especially SOC and temperature distribution on them is ignored. Besides, noise is less considered during the identification, resulting in poor robustness of the model. Aiming for these problems, a robust identification approach is proposed for the inherent parameters of soft pack LIBs in consideration with the effect of temperature distribution and state of charge as data is polluted by noise. First, to solve the modeling problem of the thermal process with unknown dynamics and unknown boundaries, a spatiotemporal least square support vector machine (LS-SVM) model is developed to reconstruct the temperature distribution of the LIB. Then, a robust adaptive neuro fuzzy (R-ANFIS) approach is developed to construct the model between the internal resistances and the external parameters, i.e., SOC, current and distributed temperature. It develops the kernel strategy to reduce the model dimension and minimizes both mean and variance of error to improve the model robustness. Using experiments at different discharge rates, the effectiveness of this method is tested and verified.

Improved Whale Optimization Algorithm with Deep Learning-Driven Retinal Fundus Image Grading and Retrieval

Several Deep Learning (DL) and medical image Machine Learning (ML) methods have been investigated for efficient data representations of medical images, such as image classification, Content-Based Image Retrieval (CBIR), and image segmentation. CBIR helps medical professionals make decisions by retrieving similar cases and images from electronic medical image databases. CBIR needs expressive data representations for similar image identification and knowledge discovery in massive medical image databases explored by distinct algorithmic methods. In this study, an Improved Whale Optimization Algorithm with Deep Learning-Driven Retinal Fundus Image Grading and Retrieval (IWOADL-RFIGR) approach was developed. The presented IWOADL-RFIGR method mainly focused on retrieving and classifying retinal fundus images. The proposed IWOADL-RFIGR method used the Bilateral Filtering (BF) method to preprocess the retinal images, a lightweight Convolutional Neural Network (CNN) based on scratch learning with Euclidean distance-based similarity measurement for image retrieval, and the Least Square Support Vector Machine (LS-SVM) model for image classification. Finally, the IWOA was used as a hyperparameter optimization technique to improve overall performance. The experimental validation of the IWOADL-RFIGR model on a benchmark dataset exhibited better performance than other models.

Modeling of Monthly Rainfall–Runoff Using Various Machine Learning Techniques in Wadi Ouahrane Basin, Algeria

Rainfall–runoff modeling has been the core of hydrological research studies for decades. To comprehend this phenomenon, many machine learning algorithms have been widely used. Nevertheless, a thorough comparison of machine learning algorithms and the effect of pre-processing on their performance is still lacking in the literature. Therefore, the major objective of this research is to simulate rainfall runoff using nine standalone and hybrid machine learning models. The conventional models include artificial neural networks, least squares support vector machines (LSSVMs), K-nearest neighbor (KNN), M5 model trees, random forests, multiple adaptive regression splines, and multivariate nonlinear regression. In contrast, the hybrid models comprise LSSVM and KNN coupled with a gorilla troop optimizer (GTO). Moreover, the present study introduces a new combination of the feature selection method, principal component analysis (PCA), and empirical mode decomposition (EMD). Mean absolute error (MAE), root mean squared error (RMSE), relative RMSE (RRMSE), person correlation coefficient (R), Nash–Sutcliffe efficiency (NSE), and Kling Gupta efficiency (KGE) metrics are used for assessing the performance of the developed models. The proposed models are applied to rainfall and runoff data collected in the Wadi Ouahrane basin, Algeria. According to the results, the KNN–GTO model exhibits the best performance (MAE = 0.1640, RMSE = 0.4741, RRMSE = 0.2979, R = 0.9607, NSE = 0.9088, and KGE = 0.7141). These statistical criteria outperform other developed models by 80%, 70%, 72%, 77%, 112%, and 136%, respectively. The LSSVM model provides the worst results without pre-processing the data. Moreover, the findings indicate that using feature selection, PCA, and EMD significantly improves the accuracy of rainfall–runoff modeling.

Non-destructive detection of protein content in mulberry leaves by using hyperspectral imaging.

Protein content is one of the most important indicators for assessing the quality of mulberry leaves. This work is carried out for the rapid and non-destructive detection of protein content of mulberry leaves using hyperspectral imaging (HSI) (Specim FX10 and FX17, Spectral Imaging Ltd., Oulu, Finland). The spectral range of the HSI acquisition system and data processing methods (pretreatment, feature extraction, and modeling) is compared. Hyperspectral images of three spectral ranges in 400-1,000 nm (Spectral Range I), 900-1,700 nm (Spectral Range II), and 400-1,700 nm (Spectral Range III) were considered. With standard normal variate (SNV), Savitzky-Golay first-order derivation, and multiplicative scatter correction used to preprocess the spectral data, and successive projections algorithm (SPA), competitive adaptive reweighted sampling, and random frog used to extract the characteristic wavelengths, regression models are constructed by using partial least square and least squares-support vector machine (LS-SVM). The protein content distribution of mulberry leaves is visualized based on the best model. The results show that the best results are obtained with the application of the model constructed by combining SNV with SPA and LS-SVM, showing an R 2 of up to 0.93, an RMSE of just 0.71 g/100 g, and an RPD of up to 3.83 based on the HSI acquisition system of 900-1700 nm. The protein content distribution map of mulberry leaves shows that the protein of healthy mulberry leaves distributes evenly among the mesophyll, with less protein content in the vein of the leaves. The above results show that rapid, non-destructive, and high-precision detection of protein content of mulberry leaves can be achieved by applying the SWIR HSI acquisition system combined with the SNV-SPA-LS-SVM algorithm.

Rapid and comprehensive grade evaluation of Keemun black tea using efficient multidimensional data fusion

To develop a comprehensive evaluation method for Keemun black tea, we used micro-near-infrared spectroscopy, computer vision, and colorimetric sensor array to collect data. We used support vector machine, least-squares support vector machine (LS-SVM), extreme learning machine, and partial least squares discriminant analysis algorithms to qualitatively discriminate between different grades of tea. Our results indicated that the LS-SVM model with mid-level data fusion attained an accuracy of 98.57% in the testing set. To quantitatively determine flavour substances in black tea, we used support vector regression. The correlation coefficient for the predicted sets of gallic acid, caffeine, epigallocatechin, catechin, epigallocatechin gallate, epicatechin, gallocatechin gallate and total catechins were 0.84089, 0.94249, 0.94050, 0.83820, 0.81111, 0.82670, 0.93230, and 0.93608, respectively. Furthermore, all compounds exhibited residual predictive deviation values exceeding 2. Hence, combining spectral, shape, colour, and aroma data with mid-level data can provide a rapid and comprehensive assessment of Keemun black tea quality.

Classification of Chinese traditional cereal vinegars and antioxidant property predication by fluorescence spectroscopy.

In this work, a rapid and accurate strategy for classification of Chinese traditional cereal vinegars (CTCV) and antioxidant property predication was proposed by using the combination fluorescence spectroscopy and machine learning. Three characteristic fluorescent components were extracted by parallel factor analysis (PARAFAC), which have correlations greater than 0.8 with antioxidant activity of CTCV obtained by Pearson correlation analysis. Machine learning methods, including linear discriminant analysis (LDA), partial least squares-discriminant analysis (PLS-DA) and N-way partial least squares discriminant analysis (N-PLS-DA), were used for the classification of different types of CTCV, and the correct classification rates was higher than 97%. The antioxidant property of CTCV were further quantified by using optimized variable-weighted least-squares support vector machine based on particle swarm optimization (PSO-VWLS-SVM). The proposed strategy provides a basis for further research on antioxidant active ingredients and antioxidant mechanisms of CTCV, and enable the continued exploration and application of CTCV from different types.

Research on Defect Detection in Kubo Peach Based on Hyperspectral Imaging Technology Combined with CARS-MIV-GA-SVM Method.

Due to the dark red surface of ripe fresh peaches, their internal injury defects cannot be detected using the naked eye and conventional images. The rapid and accurate detection of fresh peach defects can improve the efficiency of fresh peach classification. The goal of this paper was to develop a nondestructive approach to simultaneously detecting internal injury defects and external injuries in fresh peaches. First, we collected spectral data from 347 Kubo peach samples using hyperspectral imaging technology (900-1700 nm) and carried out pretreatment. Four methods (the competitive adaptive reweighting algorithm (CARS), the combination of CARS and the average influence value algorithm (CARS-MIV), the combination of CARS and the successive projections algorithm (CARS-SPA), and the combination of CARS and uninformative variable elimination (CARS-UVE)) were used to extract the characteristic wavelength. Based on the characteristic wavelength extracted using the above methods, a genetic algorithm optimization support vector machine (GA-SVM) model and a least-squares support vector machine (LS-SVM) model were used to establish classification models. The results show that the combination of CARS and other feature wavelength extraction methods can effectively improve the prediction accuracy of the model when the number of wavelengths is small. Among them, the discriminant accuracy of the CARS-MIV-GA-SVM model reaches 93.15%. In summary, hyperspectral imaging technology can accomplish the accurate detection of Kubo peaches defects, and provides feasible ideas for the automatic classification of Kubo peaches.

Classification and quantification of multiple adulterants simultaneously in black tea using spectral data coupled with chemometric analysis

Black tea is a popularly consumed beverage across the world. However, the adulteration of tea powder with extraneous colors is causing serious health threats. In the present study four such color adulterants viz., tartrazine, sunset yellow, carmoisine and ponceau 4R were simultaneously detected using FT-IR spectral data coupled with chemometric tools. PLS-DA (partial least squares discriminant analysis) was used for the classification of adulterants, whereas, PLS2 (multi response PLS) and LS-SVM (least-squares support vector machines) were used for quantification purpose. RCGA (real coded genetic algorithm) was used as a feature selection algorithm to obtain fewer fingerprints from the FT-IR spectrum of the adulterated tea powders. PLS-DA was able to predict with 100% accuracy for the samples spiked with all four adulterants when only 30 fingerprints for each adulterant were used. Amongst the regression algorithms, LS-SVM was observed to be superior over PLS2 having lower root-mean-square error of prediction (RMSEP) for carmoisine and ponceau 4R when randomized statistical test was conducted. The prediction results had high regression coefficient (R2p >0.89) with REP values ranging between 10.90% and 18.40%, and RPD values in the range of 2.82–5.81, for simultaneous quantification four adulterants using 30 variables LS-SVM model. To study matrix effect, an additional experiment was carried to observe the effectiveness of above established methodologies. Towards this, six different indigenous commercial tea brands were selected and the robustness of these chemometric methods were demonstrated. This study provides a robust tool for infrared based simultaneous detection and quantification of four different color adulterants in black tea powders.

Monitoring and detecting coal miners' fatigue status using MPA-LSSVM in the vision of smart mine

With booming computer technology and diverse computer-based smart applications, intelligent monitoring and detection of the fatigue state of coal miners (miners) has attracted extensive attention from enterprises. In order to accurately and fast knowledge the fatigue status of Miners and reduce production accidents. The article proposes a Fatigue Monitoring and Detection (FMD) model based on a fusion Machine Learning (ML) approach: Marine Predators Algorithm (MPA)-optimized Least Squares Support Vector Machine (LSSVM). Firstly, the physiological information of electroencephalogram (ECG) of coal miners before and after manual handling operations was collected using the MP160 recorder (multiconductance physiological 160) produced by BIOPAC, USA. Using the paired-samples t-test method, the characteristic indicators reflecting miners' fatigue were extracted from the ECG. Secondly, Principal Component Analysis (PCA) was used to optimize the selected feature indicators and establish the depth fatigue feature parameter set to characterize the fatigue level of miners. Finally, the proposed MPA-LSSVM-based FMD model is applied to recognize Miners' fatigue levels. The results show that the selected indexes can effectively reflect the fatigue status of Miners. The proposed MPA-LSSVM-based FMD model has higher recognition accuracy than SVM and LSSVM models (13.99% and 18.68% higher, respectively) and better robustness. Therefore, the proposed MPA-LSSVM-based FMD model can accurately and effectively identify the fatigue status of Miners.

Groundwater Quality Evaluation of Fractured Aquifers Using Machine Learning Models and Hydrogeochemical Approaches to Sustainable Water-Irrigation Security in Arid Climate (Central Tunisia)

The primary aims of this research paper involve the creation and verification of machine learning-based quality models that utilize Integrated Irrigation Water Quality Indices (IIGWQIs) through an integrated GIS approach. We utilize the Least-Squares Support Vector Machines (LS-SVM) and the Pearson Correlation Fuzzy Inference-based System (PC-FIS) to establish forecasts for groundwater quality in the Meknassy basin. This basin serves as a representative case of an irrigated region in a mining environment under arid climatic conditions in central Tunisia. The evaluated factors for groundwater quality encompass the Irrigation Water Quality Index (IWQIndex), Sodium Adsorption Ratio Index (SARIndex), Soluble Sodium Percentage Index (SSPIndex), Potential Salinity Index (PSIndex), Kelley Index (KIndex), and Residual Sodium Carbonate Index (RSCIndex). These factors were determined based on measurements from 53 groundwater wells, which included various physico-chemical parameters. The hydrogeochemical facies identified included Ca-Mg-SO4, mixed Ca-Mg-Cl-SO4, and Na-Cl facies, revealing processes such as carbonate weathering, carbonate dissolution, interactions between rocks and groundwater, and mixing ionic substitution. In terms of the irrigation suitability categories, the IWQIndex, SSPIndex, PSIndex, Kindex, and RSCIndex indicated no limitation or minimal limitation (77.36%), secure (92.45%), favorable to excellent (66.04%), favorable (81.13%), and average to secure (88.68%), respectively. However, only 15.09% were considered favorable, according to SARIndex. The evaluation of the predictive models revealed the effectiveness of both the PC-FIS model and the LS-SVM model in accurately forecasting the IIGWQIs.

Performance Comparison between SVM and LS-SVM for Rice Leaf Disease detection

INTRODUCTION: Automatic detection of rice plant diseases at early stage from its images is quite beneficial over traditional verification methods. OBJECTIVES: Recent years machine learning (ML) approaches are more efficient in disease classification task. In current generation the statistical machine learning algorithm which shows state-of-arts performance is Support Vector Machine (SVM) and variants of SVM. METHODS: SVM has an excellent learning performance for linear and non-linear data samples. It works for Quadratic Programming Problems (QPP) due to which it has the drawback of computational complexity. However QPP can be solved linearly with the help of Least Square SVM(LS-SVM) approach. In LS-SVM the epsilon tube and slack variables of SVM are replaced with error variables. The distance is calculated by error square value. RESULTS: In this research performance comparison is made between SVM and LS-SVM for rice leaf diseases such as Bacterial Leaf Blight (BLB), Brown spot(BS), Leaf smut(LS) and Leaf Blast using two datasets (DS1 and DS2).Accuracy of LS-SVM is found to be 91.3% and 98.87% for DS1 and DS2 respectively whereas accuracy of SVM is 83.3% and 98.75% for DS1 and DS2 respectively. CONCLUSION: Performance of LS-SVM outperformed than SVM in terms of accuracy.

Simultaneous Determination of Estradiol Cypionate and Medroxyprogesterone Acetate Hormones in Injectable Suspension by UV Spectrophotometry Based on Least-Squares Support Vector Machine and Fuzzy Inference System: Comparison with HPLC.

The combination of estradiol cypionate (ECA) and medroxyprogesterone acetate (MPA) is used to prevent pregnancy in women. The analysis of the ECA and MPA combination reveals a challenge due to the strong overlap of the spectra of these compounds. Spectrophotometry techniques along with chemometrics methods are simple, fast, precise, and low-cost for the simultaneous determination of ECA and MPA in a combined pharmaceutical dosage form. Two developed approaches, the least-squares support vector machine (LSSVM) and fuzzy inference system (FIS), along with a spectrophotometric method were proposed to solve such a challenging overlap. Based on the cross-validation method, the regularization parameter (γ) and width of the function (σ) in the LSSVM model were optimized and the root mean square error (RMSE) values were found to be 0.3957 and 0.2839 for ECA and MDA, respectively. The mean recovery values were 99.87 and 99.63% for ECA and MDA, respectively. The FIS coupled with principal component analysis (PCA) showed mean recovery percentages equal to 99.05 and 99.50% for ECA and MDA, respectively. Also, the RMSE of both components was lower than 0.3. The analysis results of a real sample (injection suspension) using the proposed methods were compared with HPLC by a one-way analysis of variance (ANOVA) test, and no significant differences were found in the results. Intelligent methods were proposed for the simultaneous determination of ECA and MPA. The least-squares support vector machine and fuzzy inference system along with spectrophotometry were used. HPLC as a reference method was performed and compared with chemometrics methods. The benefits of the proposed approaches are that they are rapid, simple, low-cost, and accurate.

Crosstalk tolerance analysis of coupled-line structures using least square-support vector machine technique

In this paper, crosstalk sensitivity analysis of a microwave coupled-line structure due to the fabrication imperfections is investigated using Least Square-Support Vector Machine (LS-SVM) method. Since LS-SVM uses a set of linear equations instead of a convex quadratic programming problem, the computational cost is extremely reduced compared to that of the well-known Monte Carlo (MC) analysis or even Support Vector Machine (SVM) without decreasing the accuracy. Using this method, the geometrical parameters of the coupled-line are assumed to be randomly distributed using the Latin Hypercube function and the variation range of each parameter is set to ± 50% around its central value. The frequency response of the coupled-line is estimated and compared with those of the measured and simulation ones for a few well-known practical case studies. The results show that the LS-SVM procedure quickly predicts the worst-case crosstalk expectation values and accurately anticipates the probability of obtaining various outcomes of the coupled-line for the specified parameter variation over a wide frequency range.

A NEW ROLLER BEARING FAULT DIAGNOSIS METHOD BASED ON VMD ENERGY ENTROPY AND BSOA-LSSVM

This paper presents a new method for roller bearing fault diagnosis based on least square support vector machine (LSSVM) with parameters optimized by Backtracking Search Optimization Algorithm (BSOA), namely BSOA-LSSVM. First, roller bearing acceleration vibration signals are decomposed into functions by using Variational Mode Decomposition (VMD) method. Second, initial feature matrices are extracted from those functions by energy entropy to obtain feature matrix. Third, these values serve as input vector for BSOA - LSSVM classifier. Experimental results showed that the proposed method gave the higher classification accuracy (100%) and shorter computational time than other method.

Deviation of peak hours for metro stations based on least square support vector machine.

The station-level ridership during the peak hour is one of the key indicators for the design of station size and relevant facilities. However, with the operation of metro system, it cannot be ignored that, in many cities, the station peak and the city peak may not be simultaneously occurred. As the current ridership forecasting methods use the city peak as the point of reference, stations with wide differences of ridership in between would experience disorders due to serious underestimates of passenger demand during the actual peak. Accordingly, this study fully considers the phenomenon that the metro station peak is not identical to the city peak and focuses on the concept of the peak deviation coefficient (PDC), the ratio of the station peak ridership to the city peak ridership. It investigates how metro ridership determinants affects the PDC using the least square support vector machine (LSSVM) model. A land-use function complementarity index is employed as one of the independent variables, which is newly proposed in this study that describes the relationship of the commute land use around an individual station with that along the whole network. This method can help to resolve the ridership amplification indicator for a fine-grained station-level forecasting. The results for Xi'an metro indicate that the LSSVM is an effective method to scrutinize the nonlinear effects of e.g., land use attributes, on the temporal distribution features of the metro ridership. Compared to the ratio of commute land use measured for individual stations, the land-use function complementarity index can better explain and predict the severity of peak deviation phenomenon, controlling other independent variables in the model.