Robust Hybrid Control Using Recurrent Wavelet- Neural-Network Sliding-Mode Controller for Two- Axis Motion Control System

In this paper, a robust hybrid control system (RHCS) for achieving high precision motion tracking performance of a two-axis motion control system is proposed. The proposed AHCS incorporating a recurrent wavelet-neuralnetwork controller (RWNNC) and a sliding-mode controller (SMC) to construct a RRWNNSMC. The two-axis motion control system is an x-y table of a computer numerical control machine that is driven by two field-oriented controlled permanent-magnet synchronous motors (PMSMs) servo drives. The RWNNC is used as the main motion tracking controller to mimic a perfect computed torque control law and the SMC controller is designed with adaptive bound estimation algorithm to compensate for the approximation error between the RWNNC and the ideal controller. The on-line learning algorithms of the connective weights, translations and dilations of the RWNNC are derived using Lyapunov stability analysis. A computer simulation and an experimental are developed to validate the effectiveness of the proposed RHCS. All control algorithms are implemented in a TMS320C31 DSP-based control computer. The simulation and experimental results using star and four leaves contours are provided to show the effectiveness of the RHCS. The motion tracking performance is significantly improved using the proposed RHCS and robustness to parameter variations, external disturbances, cross-coupled interference and frictional torque can be obtained as well for the two-axis motion control system.

Use an Efficient Neural Network to Improve the Arabic Handwriting Recognition

Using an efficient neural network for recognition and segmentation will definitely improve the performance and accuracy of the results; in addition to reduce the efforts and costs. This paper investigates and compares between results of four different artificial neural network models. The same algorithm has been applied for all with applying two major techniques, first, neural-segmentation technique, second, apply a new fusion equation. The neural techniques calculate the confidence values for each Prospective Segmentation Points (PSP) using the proposed classifiers in order to recognize the better model, this will enhance the overall recognition results of the handwritten scripts. The fusion equation evaluates each PSP by obtaining a fused value from three neural confidence values. CPU times and accuracies are also reported. Experiments that were performed of classifiers will be compared with each other and with the literature.

COVID-LIBERTY, A Machine Learning Computational Framework for the Study of the Covid-19 Pandemic in Europe. Part 1: Building of an Artificial Neural Network and Analysis and Parametrization of Key Factors which Influence the Spread of the Virus.

Covid-19 is the most recent strain from the corona virus family that its rapid spread across the globe has caused a pandemic, resulting in over 200,000,000 infections and over 4,000,000 deaths so far. Many countries had to impose full lockdowns, with serious effects in all aspects of everyday life (economic, social etc.). In this paper, a computational framework is introduced, aptly named COVID-LIBERTY, in order to assist the study of the pandemic in Europe. Initially, the mathematics and details of the computational engine of the framework, a feed-forward, back-propagation Artificial Neural Network are presented. 5 European countries with similar population numbers were chosen and we examined the main factors that influence the spread of the virus, in order to be taken into consideration in the simulations. In this way lockdown, seasonal variability and virus effective reproduction were considered. The effectiveness of lockdown in the spread of the virus was examined and the Lockdown Index was introduced. Moreover, the relation of Covid- 19 to seasonal variability was demonstrated and the parametrization of seasonality presented.

COVID-LIBERTY, A Machine Learning Computational Framework for the Study of the Covid-19 Pandemic in Europe. Part 2: Setting up the Framework with Ensemble Modeling

The Covid-19 pandemic has caused within a period of one year and eight months over 200,000,000 infections and more than 4,000,000 deaths. It is of paramount importance to design powerful and robust tools in order to be able to predict the evolution of the disease. In this paper, the computational framework COVID-LIBERTY is introduced, in order to assist the study of the pandemic in Europe. In Part 1, important parameters that should be taken into consideration and their parametrizations were given, as well as the details and mathematics of the computational engine of COVID-LIBERTY, a feed-forward, back-propagation Artificial Neural Network. In Part 2, the CPRT index is introduced, the framework setup around the Artificial Neural Network is presented and the algorithm of ensemble modeling is discussed, which improves the accuracy of the predictions. In the simulations, 4 European countries with similar population numbers were considered. The capabilities of the COVID-LIBERTY framework for accurate predictions for periods up to 19 days will be demonstrated.

Forecasting Analysis of Covid-19 Cases with Wavelet Neural Network and Time Series Approach

COVID-19 is a respiratory disease caused by a novel coronavirus first detected in December 2019. As the number of new cases increases rapidly, pandemic fatigue and public disinterest in different response strategies are creating new challenges for government officials in tackling the pandemic. Therefore, government officials need to fully understand the future dynamics of COVID-19 to develop strategic preparedness and flexible response planning. In the light of the above-mentioned conditions, in this study, autoregressive integrated moving average (ARIMA) time series model and Wavelet Neural Networks (WNN) methods are used to predict the number of new cases and new deaths to draw possible future epidemic scenarios. These two methods were applied to publicly available data of the COVID-19 pandemic for Turkey, Italy, and the United Kingdom. In our analysis, excluding Turkey data, the WNN algorithm outperformed the ARIMA model in terms of forecasting consistency. Our work highlighted the promising validation of using wavelet neural networks when making predictions with very few features and a smaller amount of historical data.

The Streamers Dynamics Study by an Intelligent System Based on Neural Networks

The formation and propagation of streamers is an important precursor to determine the characteristics of electrical breakdown of many HV electrode configurations. Understanding of the study of the interaction between the polymer surface and the development process of the streamer is of major importance when we want to improve internal and external performance insulation systems. In this context, a numerical tool using neural networks is developed. This model allows evaluating the speed of streamers as a function of the amplitude of voltage initiation and the nature of the insulating materials. For this, a database was created to train the neural model from a laboratory model. This investigation builds a database for predicting the propagation of streamers on the polymers surface by different neuronal methods and this presents an interesting tool for estimating the propagation phenomena in functions of very important parameters

Modelling Student’s Performance Using Data Mining Techniques in a Higher Learning Environment in the Pacific

The students’ performance in higher education has become one of the most widely studied area. Modelling student performance play a pivotal role in forecasting students’ performance where the data mining applications are now becoming most widely used techniques in this study. There are various factors, which determine the student performance. Eight attributes are used as input, which is considered most influential in determining students’ performance in the Pacific. Statistical analysis is done to see which attribute has the highest influence to student performance. In this research, different algorithms are utilized for building the classification model, each of them using various classification techniques. Some of classification techniques used are Artificial Neural Network, Decision Tree, Decision Table, and Naïve Bayes. The WEKA explorer application and R software are used for correlation test between different variables. The dataset used in this research is an imbalanced set, which is later transformed to balance set through under sampling. Neural Network is one of the classification techniques that has done well on both, imbalanced and balanced dataset. Another technique which has done well is Decision tree. Statistical analysis shows that internal assessment has weak positive relationship with student performance while demographic data is not. Further observations are reported in this research in relation to two types of datasets with application to different classification techniques

Fault Detection and Classification for Slider Attachment Process using Convolution Neural Network

Hard Disk Drive (HDD) utilizes automation machines for the assembly processes used in the industry to achieve higher production rates and lower costs. The Head Gimbal Assembly (HGA) production process has two main parts: glue dispensing and slider attaching by an Auto Core Adhesion mounting Machine (ACAM). The slider attaching process produces a mounted head to the suspension utilizing vacuum pressure to hold and position a slider. The errors from a vacuum leak from any step trigger system alarms resulting in machine downtime and slider loss defective (SLD). This paper proposes a classification algorithm derived from 250x250 micron images of mounted heads are 4 different categories: Good, Fault I, Fault II and Fault III using Convolution Neural Networks (CNN). CNN is a performance model for predictive maintenance before failure. The method has achieved a 95 % accuracy for detection and classification

A Compact Gradient Based Neural Network for Capon Spectral Estimation

This paper describes the use of a novel gradient based recurrent neural network to perform Capon spectral estimation. Nowadays, in the fastest algorithm proposed by Marple et al., the computational burden still remains significant in the calculation of the autoregressive (AR) Parameters. In this paper we propose to use a gradient based neural network to compute the AR parameters by solving the Yule-Walker equations. Furthermore, to reduce the complexity of the neural network architecture, the weights matrixinputs vector product is performed efficiently using the fast Fourier transform. Simulation results show that proposed neural network and its simplified architecture lead to the same results as the original method which prove the correctness of the proposed scheme.

Detection of Brain Tumor Using K-Nearest Neighbor (KNN) Based Classification Model and Self Organizing Map (SOM) Algorithm

Knowledge discovery is also known as Data mining in databases, in recent years that technique plays a major role in research area. Data mining in healthcare domain has noteworthy usage in real world. The mining method can enable the healthcare field for the enhancement of institutionalization of its administrations and become quicker with best in class technologies. Innovation utilization isn't restricted to basic leadership in undertakings, yet spread to different social statuses in all fields. In this paper a novel approach for the detection of brain tumor is proposed. The novel approach uses the classification technique of K-nearest neighbor (KNN) and for ignoring the error of the dataset image SOM (self-organizing map) algorithm has been used. Discrete wavelet transform (DWT) is used for transforming input image data set, in which RGB color of input data image has been converted into gray scale. Then it has been classified using KNN after that the error avoiding algorithm has been carried out. This will help to differentiate tumor cells and the normal cells. The presence of tumor in brain image is detected using parametric analysis by simulation.