Radial Bias Function Research Articles

Machine Learning Based Fault Classification in Pilot Plant Batch Reactor: Using Support Vector Machine.

Identifying and diagnosing faults is a critical task in process industries to maintain effective monitoring of process and plant safety. Minimizing process downtime is critical for enhancing the quality of the product and minimizing production costs. Real-time categorization of issues across several levels is essential for the monitoring of processes. However, there are still notable obstacles, that must be addressed, such as the existence of robust correlations, the complexity of the data, and the lack of linearity. This study introduces a novel fault identification technique in batch reactor experimental trials that employs multikernel support vector machines (SVMs) to categorize internal and external issues, specifically reactor temperature, coolant temperature, and jacket temperature. The data set was obtained from empirical research. The classification has been conducted using a multikernel SVM. This article identified that the nonlinear classifier using the radial bias function results in an accuracy that is at least 22.08% superior to other methods.

Exploring Groundwater Quality Trends in Valliyar Sub-Basin, Kanniyakumari District, India through Advanced Machine Learning Techniques

The assessment of water quality assumes a position of utmost significance as it plays a critical role in upholding ecological balance and safeguarding the well-being of human populations. To achieve these goals, an in-depth consideration of water quality trends is essential, as it offers comprehension into the intricate interplay between various elements within aquatic ecosystems. As a consequence, the proposed work investigates the water quality trends specifically within the Valliyar sub-basin, which encompasses the geographical areas of Kattathurai, Colachal, Thuckalay, and Villukuri. The temporal scope of investigation spans from the year 2000 to 2018 using the dependent variable of water quality parameters with dependent variables of climate data. Recognizing the need for advanced methodologies to tackle the multifaceted nature of water quality dynamics, this research harnesses the power of pioneering machine learning techniques. Two notable approaches, the Radial Bias Function Neural Network (RBFNN) and the DenseNet-121-based Convolutional Neural Network (CNN), are brought into performance. The primary objective is to leverage these techniques to forecast water quality trends for the next twenty-two years. The effectiveness of various machine learning models in predicting water quality is evaluated using the following key performance metrics: the Mean-Square Error (MSE), Mean Absolute Percentage Error (MAPE) and Root-Mean-Square Error (RMSE). Notably, the DenseNet CNN model exhibits accurate prediction among the Artificial Neural Network (ANN), Support Vector Machine (SVM), and Deep Learning (DL) models. This research underscores the significance of machine learning techniques, with DenseNet CNN model emerging as a particularly promising tool in this domain.

Forecasting Bitcoin Price Utilizing Evolutionary Radial Bias Function Networks

A semi-parametric evolutionary method called the Fireworks method (FWA) and a radial bias function neural network (RBFN) are combined in this study to make a new type of network called RBFN-FWA. Effectively adjusting the RBFN's biases and weights is the goal of integrating FWA. We next used this proposed methodology to predict how Bitcoin's price will go in the future. The solo RBFN was trained using various optimization techniques such as GA, PSO, and GD for comparison purposes. This resulted in three new models: RBFN-GA, RBFN-PSO, and RBFN-GD. We also use all the other models that can do the same thing. Utilizing error metrics like Mean Absolute Percentage Error (MAPE) and Normalized Mean Squared Error (NMSE), we evaluated the models' performance. In terms of NMSE and MAPE, the experimental findings show that RBFNN-FWA is the best comparison model, demonstrating its higher predictive potential.A semi-parametric evolutionary method called the Fireworks method (FWA) and a radial bias function neural network (RBFN) are combined in this study to make a new type of network called RBFN-FWA. Effectively adjusting the RBFN's biases and weights is the goal of integrating FWA. We next used this proposed methodology to predict how Bitcoin's price will go in the future. The solo RBFN was trained using various optimization techniques such as GA, PSO, and GD for comparison purposes. This resulted in three new models: RBFN-GA, RBFN-PSO, and RBFN-GD. We also use all the other models that can do the same thing. Utilizing error metrics like Mean Absolute Percentage Error (MAPE) and Normalized Mean Squared Error (NMSE), we evaluated the models' performance. In terms of NMSE and MAPE, the experimental findings show that RBFNN-FWA is the best comparison model, demonstrating its higher predictive potential.

Model-based neuroadaptive event-triggered tracking consensus control for nonlinear multiagent systems with input delay

In this article, a neuroadaptive event-triggered control methodology is proposed to tackle the leader-following tracking consensus issue of uncertain nonlinear multiagent systems (MASs) with input delay and external disturbances. The proposed approach incorporates radial bias function neural networks (RBF NNs) and the integral compensation technique, effectively addressing the issues of nonlinear uncertainty and input delay, respectively. For the sake of reducing the transmission of information, we adopt the strategy that the neural network (NN) weight update law is only changed at the event-triggered instant, and a novel adaptive model and related controller are constructed. Simultaneously, this paper establishes an impulsive dynamic system to analyze the stability of the closed-loop system by extending the Lyapunov theorem to continuous and reset dynamics. Sufficient conditions for achieving leader-following tracking consensus are derived without the Zeno behavior. Finally, a numerical example is presented to illustrate the validity of the proposed method.

UAV hyperspectral remote sensor images for mango plant disease and pest identification using MD-FCM and XCS-RBFNN.

To diminish disease transmission together with promoting effective management techniques, it is crucial to monitor plant health and detect pathogens earlier. The initial part in reducing losses sourced from plant diseases is to make an accurate and earlier identification. Thus, the usage of unmanned aerial vehicle (UAV) hyperspectral imaging (HSI) sensors for surveying and assessing crops, orchards, and forests has rapidly elevated over the last decade, particularly for the stress management like water, diseases, nutrition deficits, and pests. Using Minkowski Distance-based Fuzzy C Means (MD-FCM) clustering and Xavier initialization-adapted Cosine Similarity-induced Radial Bias Function Neural Network (XCS-RBFNN) techniques, a UAV HS imaging remote sensor for Spatial and Temporal Resolution (STR) of mango plant disease and pest identification is proposed in this scheme. Collecting the input UAV source (image or video) is eventuated initially along with the Region of Interest (ROI) calculated which is followed by preprocessing. Leaf segmentation is eventuated using Logistic U-net after preprocessing. Next, MD-FCM performs clustering to cluster the diseased leaves and pests individually. The disease and pest characteristics are then retrieved separately and classified further. The requisite features are then chosen from the retrieved features utilizing the Levy Flight Distribution-produced Butterfly Optimization Algorithm (LFD-BOA). Finally, the XCS-RBFNN classifier is utilized to categorize the various diseases together with pests found in the UAV input source using the chosen features. The proposed framework's experimental findings are then compared to some prevailing schemes, with the results revealing that the proposed work outperforms other benchmark techniques.

Experimental evaluation and modeling the mass and temperature of dried mint in greenhouse solar dryer; Application of machine learning method

This study is aimed to model the temperature and mass of dried mint in a Quonset type of Greenhouse Solar Dryer (GSD). The inputs including ambient air temperature (◦C), ambient air humidity (%) and solar radiation (Wm-2) and output data including temperature (◦C) and mass (gr) of dried mint were collected from a Quonset GSD. Artificial Neural Network (ANN) models including Multilayer Perceptron (MLP) and Radial Bias Function (RBF) and also, Gaussian Process Regression (GPR) by k-fold cross validation method were used for modeling. Levenberg-Marquardt (LM) learning algorithm with Sigmoid-Tangent transfer function by different combinations of neurons in the hidden layer were assessed for ANN models. The results showed that MLP and GPR have higher error than RBF model for predicting the temperature and mass of dried mint. The results of RBF optimization indicated that 3-15-1 and 3-18-1 topologies with using 60 and 50% of total dataset for training steps and having 0.4 and 0.3 spread factor values can predict the temperature and mass of dried mint with Mean Absolute Percentage Error (MAPE) of 1.4 and 1.82%, respectively. The results of t, F, and Kolmogorov–Smirnov tests indicated that there is no significant difference between actual and RBF output values.

Cryptojacking Detection Using Genetic Search Algorithm

Mining cryptocurrency with an unauthorized and unlawful access to a victim’s computer’s processing power is called cryptojacking. With the rise of cryp- tocurrency in the markets people took advantage of this new piece of technol- ogy through mining it and earning from it. When the need for more perfor- mance and money emerged people came out with unlawful activities to mine cryptocurrency using other’s devices without their consent. These activities have led people and their devices vulnerable for the sake of greed. This is a dis- tributive approach which uses victim’s machine to mine cryptocurrency using all its resources i.e., CPU, GPU etc. This doesn’t need any software installa- tion but just to visit sites which are embedded with the malicious scripts that helps accessing user’s device. The analogy of this study is not limited to detect the presence of cryptojacking but also to enable users to detect the presence of any kind of suspicious activity is performed on the device. Through Genetic Search algorithm we extracted some of the key metrics used by computers for monitor its resources and further used these metrics to classify the presence of any unauthorized activity and achieved 100% accuracy for classifying these instances. The classification was done using several classification algorithms such as SVM (kernel – “linear”), SVM (kernel – “Radial Bias Function”), Logistic Regression, Decision Tree Classifier, Random Forest Classifier, Multi-Layer Perceptron. The Genetic Search algorithm mentioned earlier is a machine learning iterative technique which is based on natural selection it selects individuals from the current population as parents and uses them to generate the next generation of offspring and all this through a method called CfsSubsetEval which returns a fitness score to the child based on their dependency on other attributes. For comparison we also used several methods which also serves the same objective

Evaluating the Influence of Carbon Fiber on the Mechanical Characteristics and Electrical Conductivity of Roller-Compacted Concrete Containing Waste Ceramic Aggregates Exposed to Freeze-Thaw Cycling

Traditional methods of removing snow and ice from pavements using chemicals are combined with mechanical removal that involves a lot of manpower, advanced machinery, chemicals that are harmful to the environment, and damage to pavements. Furthermore, annually, large quantities of ceramic materials become waste due to their fragile nature during processing, transport, and installation, and their accumulation in the nature has brought about environmental and health-related concerns. Therefore, the study aims to investigate the effect of using waste ceramic as a replacement for fine aggregate in roller compacted concrete (RCC) and the application of carbon fiber to improve the mechanical properties and electrical conductivity of RCC. To achieve this goal, several tests such as compressive strength, indirect tensile strength, electrical resistance, chloride ion penetration, specific gravity, and skid resistance tests were carried out on the fabricated samples before and after freeze-thaw cycling exposure. The experimental results illustrated that replacing waste ceramics with fine-grained aggregate increased the compressive strength and tensile strength of RCC. Furthermore, carbon fiber increased tensile strength but had no noticeable influence on compressive strength. Freeze-thaw conditioning led to a reduction in the compressive and tensile strength regardless of the aggregate type and carbon fiber utilization. In the samples containing waste ceramic aggregate, the electrical conductivity was reduced, and by adding carbon fiber, its electrical conductivity was increased. Exposure to freeze-thaw cycling resulted in an increase in electrical resistance and the passing charge. Waste ceramic incorporation created a similar mixture in terms of skid resistance, while in contrast, the carbon fiber slightly reduced the skid resistance. In addition, freeze-thaw conditioning resulted in an increase in the skid resistance. Besides, in this study, kernelized support vector regression (KSVR) and radial bias function (RBF) neural network models were proposed to estimate the indirect tensile strength (ITS) and compressive strength (CS) values. The results showed that both models have high performance in estimating these values, but RBF was a more efficient model.

Exploration of V–Cr–Fe–Co–Ni high-entropy alloys with high yield strength: A combination of machine learning and molecular dynamics simulation

Improving the strength of Cr–Fe–Co–Ni high-entropy alloys is a key issue in expanding their applicability. Herein, a framework combining machine learning and molecular dynamics is employed to improve the yield strength of Cr–Fe–Co–Ni high-entropy alloys through vanadium addition. The results indicate that by specifying the valence electron concentration, vacancy formation energy, and mismatch in cohesive energy as input features, a support vector regression model with a radial bias function kernel displays the highest performance among numerous combinations of machine learning models and material features. According to the Shapley additive explanation, the vacancy formation energy and valence electron concentration show a negative correlation with the yield strength above 1.66 eV and 7.60, respectively, and a positive correlation otherwise. The mismatch in the cohesive energy always shows a negative correlation. By utilizing a Bayesian adaptive alloy design, V5Cr16Fe9Co35Ni35 has been identified to have the highest yield strength. The simulated tensile deformation of polycrystalline high-entropy alloys confirms the predicted trend in yield strength, and the events observed during plastic deformation are consistent with previous experimental observations. The proposed framework provides a promising prospect for accelerating the design of high-entropy alloys with reduced dependence on costly trial-and-error experiments.

An optimized radial bias function neural network for intrusion detection of distributed denial of service attack in the cloud

SummaryCloud system is a set of resources meant to provide cloud customers with on‐demand services. Accessibility to cloud environment is provided through internet services, making data stored on cloud more accessible to attackers internally and externally. Several intrusion detection systems and authentication methods have been developed in previous studies to identify the intruders, however they are generally unsuccessful with certain drawbacks. Numerous existing researchers have focused on machine learning techniques for identifying intrusions. This article intends to introduce a new intrusion detection model in the cloud and it involves three processes like preprocessing, feature extraction and detection. Initially, preprocessing takes place and the preprocessed data are subjected to feature extraction process. The flow‐based features, statistical, and higher‐order statistical features with improved holoentropy features, and the technical indicators are extracted. After extracting the features, they are subjected to a detection process, where the optimized radial bias function neural network (RBF‐NN) is exploited. For precise detection, the weight of RBF‐NN is tuned optimally by (harmonic mean based poor and rich optimization) HMPRO Algorithm. The main aim of the proposed model is to detect attacks in the cloud. Finally, the performance of the presented scheme is computed over the existing approaches using the CICDDoS2019 and UNSW‐NB_15 dataset in terms of different metrics.

Evaluating the substitution potential of SBS with crumb rubber-polypropylene blends as asphalt binder and mixture modifiers

The study aimed to assess the suitability of substituting styrene–butadienestyrene with waste crumb rubber-polypropylene blends as modifiers to enhance the performance characteristics of asphalt binders and mixtures. Six crumb rubber-polypropylene blends including three styrene–butadienestyrene percentages were used to modify a base binder and then the modified binders were used to fabricate asphalt concrete mixtures. To obtain the physical and compatibility behavior of the unmodified and modified binders, various conventional binder tests such as softening point, penetration, and storage stability were employed. Dynamic shear rheometer, multiple stress creep recovery, and linear amplitude straintests were performed on unaged and long-term aged samples to achieve a broad understating of rheological characteristics at high and intermediate temperatures. In order to assess the influence of the modifiers on asphalt mixtures, various test methods were implemented including indirect tensile strength, resilient modulus, and four-point beam fatigue and dynamic creep tests. The findings of the physical tests represented a decline in penetration and a rise in softening point by the crumb rubber-polypropylene and styrene–butadienestyrene modifications compared with the base one. Results of the storage stability test proved that only 3% styrene–butadienestyrene modified was a storage-stable blend at high construction temperatures and the instability increased by incorporating more polymer content. According to the rotational viscosity test, all of the modifications increased the viscosity, reducing the workability of the modified mixture. Based on the rheological tests, the addition of the modifiers resulted in a reduction in phase angles, and an increase in complex modulus at high temperatures, thereby decreasing binder sensitivity to permanent deformation. With increasing the percentage of the polymer, more improvement would be obtained regarding rutting resistance. At intermediate temperatures, however, the complex modulus of modified binders decreased while no notable changes were observed for phase angle compared to the base binder. Considering the mechanical test results, the crumb rubber-polypropylene and styrene–butadienestyrene modified mixtures resulted in an enhancement in indirect tensile strength, resilient modulus, four-point beam fatigue, and dynamic creep tests results which would cause better resistance against rutting, fatigue damage, and moisture susceptibility compared with the base one. Besides, a radial bias function neural network model was designed to predict tensile strength ratio, and fracture energy ratio. The results showed the high performance of this method in the estimation of these values.

On Noise Effect in Semi-supervised Learning

The article deals with the problem of noise effect on semi-supervised learning. The goal of this article is to analyze the impact of noise on the accuracy of binary classification models created using three semi-supervised learning algorithms, namely Simple Recycled Selection, Incrementally Reinforced Selection, and Hybrid Algorithm, using Support Vector Machines to build a base classifier. Different algorithms to compute similarity matrices, namely Radial Bias Function, Cosine Similarity, and K-Nearest Neighbours were analyzed to understand their effect on model accuracy. For benchmarking purposes, datasets from the UCI repository were used. To test the noise effect, different amounts of artificially generated randomly-labeled samples were introduced into the dataset using three strategies (labeled, unlabeled, and mixed) and compared to the baseline classifier trained with the original dataset and the classifier trained on the reduced-size original dataset. The results show that the introduction of random noise into the labeled samples decreases classifier accuracy, while a moderate amount of noise in unmarked samples can have a positive effect on classifier accuracy.

Applying machine learning to explore the association between biological stress and near misses in emergency medicine residents.

Physician stress is associated with near misses and adverse medical events. However, little is known about physiological mechanisms linking stress to such events. We explored the utility of machine learning to determine whether the catabolic stress hormone cortisol and the anabolic, anti-stress hormone dehydroepiandrosterone sulfate (DHEA-S), as well as the cortisol to DHEA-S ratio relate to near misses in emergency medicine residents during active duty in a trauma 1 emergency department. Compared to statistical models better suited for inference, machine learning models allow for prediction in situations that have not yet occurred, and thus better suited for clinical applications. This exploratory study used multiple machine learning models to determine possible relationships between biomarkers and near misses. Of the various models tested, support vector machine with radial bias function kernels and support vector machine with linear kernels performed the best, with training accuracies of 85% and 79% respectively. When evaluated on a test dataset, both models had prediction accuracies of around 80%. The pre-shift cortisol to DHEA-S ratio was shown to be the most important predictor in interpretable models tested. Results suggest that interventions that help emergency room physicians relax before they begin their shift could reduce risk of errors and improve patient and physician outcomes. This pilot demonstrates promising results regarding using machine learning to better understand the stress biology of near misses. Future studies should use larger groups and relate these variables to information in electronic medical records, such as objective and patient-reported quality measures.

An enhanced support vector regression model for agile projects cost estimation

The appearance of agile software development techniques (ASDT) since 2001 has encouraged many organizations to move to an agile approach. ASDT presents an opportunity for researchers and professionals, but it has many challenges as well. One of the most critical challenges is agile effort prediction. Hence, many studies have investigated agile software development cost estimation (ASDCE). The objective of this study is twofold: First, to propose an improved model based on support vector regression with radial bias function kernel (SVR-RBF) enhanced by the optimized artificial immune network (Optainet). Second, to perform a detailed comparative analysis of the proposed method compared to other existing optimization techniques in the literature and applied for ASDCE. The experimental evaluation was carried out by assessing the performance of the proposed method using some trusted measures like standardized accuracy (SA), mean absolute error (MAE), prediction at level p (Pred(p)), mean balanced relative error (MBRE), mean inverted balanced relative error (MIBRE), and logarithmic standard deviation (LSD). Throughout a dataset with 21 agile projects using the leave-one-out cross-validation (LOOCV) technique. The results obtained prove that the proposed model enhances the accuracy of the SVR-RBF model, and it outperforms the majority of existing models in the literature.

Comparison between conventional and deep learning-based surrogate models in predicting convective heat transfer performance of U-bend channels

Deep neural networks are efficient methods to achieve real-time visualization of physics fields. The main concerns that prevented deep learning from being implemented in the field of energy conversion were the risks of overfitting and the lack of data. Therefore, it is necessary to evaluate different kinds of surrogate modeling methods and provide guidelines for designers to choose models. In this study, three conventional models (Artificial Neural Network, Radial Bias Function, and Kriging), and two deep learning-based models (Convolutional Neural Network and Conditional Generative Adversarial Neural Network) were established to predict the flow and heat transfer performance of a U-bend with variable geometries. The models were detailly compared in terms of the single-point prediction accuracy, response accuracy, sensitivity to sample size, and other characteristics of interest. Results showed that the conventional models had slightly higher single point accuracy and the relative error of pressure loss and heat transfer were within ±6.6% and ±5.7% respectively, while those of the deep learning-based models were within ±8.0% and ±6.3% respectively. Nevertheless, the deep learning-based models had higher response accuracy and could reconstruct the distributions of surface pressure and wall heat flux with the pixel-wise absolute error within ±2.0 Pa and ±45 W/m2 respectively. The results indicated that deep learning was a promising surrogate modeling approach due to its acceptable prediction error and ability to reconstruct physical fields. This effort was expected to serve as a guide for establishing more reliable data-driven surrogate models for energy conversion and heat transfer problems.

Machine learning models for prediction the Higher Heating Value (HHV) of Municipal Solid Waste (MSW) for waste-to-energy evaluation

Time-resolved knowledge of physico-chemical properties of Municipal Solid Waste (MSW) materials and their thermal energy content is one of the important subjects needed to build waste incineration power plants around the world. For this purpose, machine learning models were used to predict the Higher Heating Value (HHV) of MSW based on the initial materials. Four types of machine learning methods: Radial Bias Function Artificial Neural Network (RBF-ANN), Multilayer Perceptron Artificial Neural Network (MLP-ANN), Support Vector Machine (SVM) and Adaptive Nero-Fuzzy Inference System (ANFIS) were used for modeling the HHV with six different inputs (carbon, water, hydrogen, oxygen, nitrogen, sulfur, and ash). The results showed that RBF-ANN can predict the HHV of MSW with higher accuracy than other models. The overall Mean Absolute Percentage Error (MAPE) for MLP-ANN, SVM and ANFIS models were 7.3, 11.77 and 23.76%, respectively. The MAPE of the best topology for RBF model (6-17-1) with spread factor of 0.8 reach to 0.45%. Finally, the results of this study proved that ANN's can be used as a practical tool with high accuracy and reliability for design and management of waste incineration plants.

Prediction of Soil Penetration Resistance with Three Different Artificial Neural Networking Methods

Soil compaction is one of the major problems of the agricultural sector affecting negatively the soil structure and impedes plant root growth. Soil penetration resistance should be measured from many points of the production land to determine the effects of plant growth problems. Collection of soil penetration values from huge lands is time-consuming, tiring, and tedious for researchers. Also, the number of measured points to what extent will be sufficient to evaluation on whole production area is not clear. To eliminate this ambiguity, soil penetration values of the unmeasured points should be estimated to evaluate the whole land. Artificial Neural Networks (ANN) are one of the most popular mathematical computing and modeling method used to estimate unknown data with the help of known data. In this study, we collected 1603 samples of geographical position and soil penetration value from 40 cm depth within the 20 ha field. From the 1603 values, 24% records were selected for testing and the remaining 76% records were used for training. Soil penetration values of the unmeasured points were estimated using Generalized Regression Neural Network (GRNN), Multi-Layer Perceptron (MLP) and Radial Bias Function (RBF) methods in MATLAB. In addition to mean squared error (MSE), root mean square error (RMSE) and mean absolute error (MAE) has been also used for evaluation of prediction accuracy on these methods. RBF results showed very good agreement between the predicted and the measured real values of soil resistance (MSE: 0.1608; RMSE: 0.3717; MAE: 0.3682).

Investigating the Effects of Waste Cr-Pp Combinations on the Properties of Asphalt Binder and Mixture in Comparison to a Commercial Thermoplastic Elastomer, as Well as a Moisture Susceptibility Prediction Model Based on Radial Bias Function Neural Network

Investigating the Effects of Waste Cr-Pp Combinations on the Properties of Asphalt Binder and Mixture in Comparison to a Commercial Thermoplastic Elastomer, as Well as a Moisture Susceptibility Prediction Model Based on Radial Bias Function Neural Network

Empirical Mode Decomposition with Fuzzy Weight Beetle Swarm Optimization (EMD-FWBSO) Denoising and Enhanced Kernel Support Vector Machine (EKSVM) Classifier for Arrhythmia in Electrocardiogram Recordings

Elderly persons are generally prone to CHDs (Chronic Heart Diseases). Arrhythmia is a persistent CHD with high mortalities resulting from cardiac failures, heart strokes, and CADs (Coronary Artery Diseases). Arrhythmia can be detected using ECG (Electrocardiogram) signals. ECG signals need to be pre-processed for removing noises present in signals. Since denoising is a significant step in ECG signals. Recently Support Vector Machine -Radial Bias Function (SVM-RBF) classifier is introduced for arrhythmia classification, it doesn’t remove noises presented from the ECG signals. The major aim of the work is to design a new classifier with removed noises and enhanced ECG signal. In this work, EMDs (Empirical Mode Decompositions) is introduced for noise removing which works recursively and dependent on signals called sifting. In EMD, IMFs (Intrinsic Mode Functions) decompose noisy signals into intrinsic oscillatory components adaptively using sifting. Further, FWBSOs (Fuzzy Weight Beetle Swarm Optimizations) are used in this work for optimizing EMDs and IMFs. This work in the initial phase reconstructs ECG signals which are filtered by IMFs. These filters are followed by extraction of morphological features from waves of P-QRS-T while ECG segments are selected using PCAs and DTWs. In the final phase, EKSVMs (Enhanced Kernel Support Vector Machines) classifies extracted features automatically by categorizing ECG signals into Normal and Ventricular Ectopic Beats. This work’s resulted are evaluated with performance metrics of Sensitivity, F-measure, Positive Productivity and Accuracy. This work uses database of MIT-BIH arrhythmia in a 5 fold cross validation for its predictions. The proposed EKSVMs classifier is compared to existing classifiers such as K-Nearest Neighbors (KNN), Enhanced Particle Swarm Optimisation-Multiple Layer Perception (EPSO-MLP) and SVM-RBF. The experiments of the proposed classifier and existing methods are carried out on MATLAB R2018a.

Neural Integral Backstepping Hierarchical Sliding Mode Control for a Ridable Ballbot Under Uncertainties and Input Saturation

This article proposes a novel adaptive backstepping hierarchical sliding mode control with balance and transfer for a ridable ballbot. The algorithm is applied to the dynamics of coupling two vertical planes. The dynamic model is a multiple-input–multiple-output underactuated system. Based on the backstepping and hierarchical sliding mode schemes, a robust nonlinear control law is obtained to govern the balance and transfer of the ridable ballbot; this scheme considers input constraints as well as matched and mismatched uncertainties. The influence of actuator saturation is examined by proposing an auxiliary system. The stability of the whole system is guaranteed via Lyapunov theory and Barbalat’s lemma, which shows that the switching gains are limited by the upper bound of the matched uncertainty. An adaptive law is proposed to estimate the matched uncertainty using radial bias function networks. As a result, the robustness is enhanced, and the constraints of the switching gains on the matched uncertainty can be relaxed. The stabilizing and tracking problems of the ridable ballbot are investigated through comparative simulations. The results demonstrate the effectiveness and robustness of the proposed control algorithm in the presence of parameter variations, external disturbances, and input saturation.