International Journal of Neural Networks and Advanced Applications

ISSN: 2313-0563
Volume 3, 2016

Notice: As of 2014 and for the forthcoming years, the publication frequency/periodicity of NAUN Journals is adapted to the 'continuously updated' model. What this means is that instead of being separated into issues, new papers will be added on a continuous basis, allowing a more regular flow and shorter publication times. The papers will appear in reverse order, therefore the most recent one will be on top.

Main Page

Submit a paper | Submission terms | Paper format


Volume 3, 2016

Title of the Paper: Maximum Power Point Tracking of Photovoltaic Modules Comparison of Neuro-Fuzzy “ANFIS” and Artificial Network Controllers Performances


Authors: Z. Ons, J. Aymen, M. Mohamed Nejib, C. Aurelian

Pages: 53-57

Abstract: This paper makes a comparison between two control methods for maximum power point tracking (MPPT) of a photovoltaic (PV) system under varying irradiation and temperature conditions: the Neuro-Fuzzy logic and the Neural Network control. Both techniques have been simulated and analyzed by using Matlab/Simulink software. The power transitions at varying irradiation and temperature conditions have been simulated and the power tracking time realized by the Neuro-Fuzzy logic controller against the Neural Network controller has been evaluated.

Title of the Paper: On the Optimum Choice of the K Parameter in Hand-Written Digit Recognition by kNN in Comparison to SVM


Authors: Ivaylo Penev, Milena Karova, Mariana Todorova

Pages: 47-52

Abstract: The paper concerns the application of two machine learning algorithms – k-nearest neighbor (kNN) and support vector machines (SVM) for solving the problem of hand-written digit recognition. The main goal of the work is to derive recommendations for the choice of the K parameter in kNN (number of the nearest neighbors) so that the performance of kNN to be the near (or even better than) the performance of SVM – one of the most power machine learning known algorithms. The kNN distance function as well as the method for choosing a class of the recognized digit are explained. The presented experimental results show comparison of the kNN performance to SVM, regarding two criteria – percent of the correctly recognized digit images and run time for recognition. As a final result recommendations for the choice of the K value are summarized.

Title of the Paper: PID Parameterization of Cement Kiln Precalciner Based on Simplified Modeling


Authors: Dimitris C. Tsamatsoulis, Georgi Zlatev

Pages: 41-46

Abstract: This study aims in tuning a PID controller of cement kiln precalciner between the feed rate of the primary fuel and the temperature at precalciner exit. A simplified dynamic modeling has been used, including perfect mixers connected in series. The optimum number of tanks and the dynamical parameters has been computed using industrial data. The PID gains are determined by loop shaping technique using the maximum sensitivity as robustness criterion. The uncertainty of their values is computed based on the uncertainty of the dynamic parameters. Due to simplicity of the model, the tuning results could be used at least as initial PID values in real process.

Title of the Paper: Development of a Temperature and Heat Insensitive Lathe via Neural Network Inverse Analysis Control


Authors: Ikuo Tanabe

Pages: 34-40

Abstract: Since the beginning of the 21st century, the importance to manufacture products in an environmentally-conscious way has been highlighted. In this regard, manufacturers not only need to conserve energy, but they also need to scrutinize in order to save resources and reduce environmentally-harmful pollutants. Nevertheless, most machine tools highly depend on lubricating oil to achieve a smoother drive, large amounts of electricity during forced cooling to attain a high accuracy, as well as large amounts of cutting oil to accomplish lubrication and cooling effects. Since this represents a large environmental problem, a lathe insensitive to temperature and heat fluctuations was developed and evaluated. Specifically, the developed lathe was meant to be both a three-dimensional fixed-zero system structure and a forced self-cooling structure. Furthermore, an air flow speed control used on the forced self-cooling system was developed, using the neural network inverse analysis, for the reduction of thermal deformation on the bench lathe. Thereafter, the thermal deformation of the developed lathe present in several experiments was measured and evaluated. As a result it is concluded that: (1) even though there was no active forced cooling, the thermal deformation of the bench lathe was considerably small, and (2) the air flow speed control used on the forced self-cooling system, through the neural network inverse analysis, was effective in achieving a stable operation, disregarding weather variations.

Title of the Paper: Inhibition of Spikes in an Array of Coupled FitzHugh?Nagumo Oscillators by Means of Alternating Current


Authors: Arūnas Tamaševičius, Gytis Mykolaitis, Elena Adomaitienė, Skaidra Bumelienė

Pages: 30-33

Abstract: Damping of spikes in an array of coupled oscillators by injection of sinusoidal current is studied both experimentally and numerically. The effect is investigated using an array, consisting of thirty mean-field coupled FitzHugh?Nagumo type oscillators. The results are considered as a possible mechanism of the deep brain stimulation, used to avoid the symptoms of the Parkinson’s disease.

Title of the Paper: Artificial Neural Networks-Based Methodological Approach for Climatic Variables Prediction


Authors: Francklin Rivas-Echeverría, Edmundo Recalde, Iván Bedón, Stalin Arciniegas, David Narváez

Pages: 23-29

Abstract: In this work it’s presented a neural networks-based methodological approach for obtaining climatic variables prediction using artificial neural networks. In this methodology it’s included statistical data analysis techniques and has been used for temperature, humidity and Pressure prediction using data collected from a weather station in Ibarra, Ecuador.

Title of the Paper: Neural Network for Overall Porosity Prediction of Hollow Fiber Membrane


Authors: Mohammad Abbasgholipour Ghadim, Musa Bin Mailah, Intan Zaurah, A. F. Ismail, M. Rezaei Dashtarzhandi, Mahdi Abbasgholipour

Pages: 16-22

Abstract: This study aims to introduce an attractive and convenient method of calculating the overall porosity of Hollow Fiber Membrane (HFM). Artificial Neural Network (ANN) was used to estimate the overall porosity of HFMs. Overall porosity is predicted as a function of effective surface porosity of fabricated HFMs which depends on polymer solution composition, dope extrusion flow rate and bore fluid flow rate. The artificial neural network (ANN) converted the qualitative information based on quantitative results from the outer surface analyzed through a Field Emission Scanning Electron Microscopy (FESEM) images. A neural network with one hidden layer with three neurons was created to map the relationship between input and output. An image processing computer program was developed to measure the HFM surface porosity using the FESEM images. Obtained data by image processing program was used as input data for designed ANN. The calculated overall porosity of the HFM was compared with the achieved value from the mathematical model. It was found that there was no significant difference between the results of both methods, thereby confirming the applicability of ANN for assessing the membrane porosity.This work presents a novel approach and provides a useful framework to evaluate the overall porosity of HFM considering different dope compositions and spinning conditions.

Title of the Paper: Identification and Control of Nonlinear Dynamical Systems Using Levenberg-Marquardt Learning Algorithm for Recurrent Complex-Valued Neural Networks


Authors: Ieroham S. Baruch, Victor Arellano Quintana, Edmundo P. Reynaud

Pages: 8-15

Abstract: In this work, a recursive Levenberg-Marquardt (LM) learning algorithm in the complex domain is developed and applied to the learning of an adaptive control scheme composed by Complex-Valued Recurrent Neural Networks (CVRNN). We simplified the derivation of the LM learning algorithm using a diagrammatic method to derive the adjoint CVRNN used to obtain the gradient terms. Furthermore, we apply the CVRNN control scheme for a particular case of a nonlinear, oscillatory mechanical plant to validate the performance of the adaptive neural controller and the learning algorithm. The obtained simulation results using a flexible robot arm confirm a good performance of the derived control schemes and learning algorithms to suppress the occurred robot oscillations and tracking error.

Title of the Paper: Recent Advances of Adoptability of EEG Signals for Application Aimed at Improving the Life of Disabled People


Authors: Simona Petrakieva, Georgi Tsenov, Valeri Mladenov

Pages: 1-7

Abstract: In the recent years affordable wireless commercial EEG headsets have become widely available, thus opening the possibility of mass EEG signal application for various tasks. With the help of these devices the life of disabled people that are not able to walk or talk can be vastly improved by measuring and recognizing certain brain patterns with their manifestation in EEG signals. In this paper we present a survey on the recent advances in the field for recognizing brainwaves like in the cases that EEG signals are to be used for directional movements that can be used to control a mechanized wheelchair, or for typing on virtual keyboard by presenting on a screen alphabet symbols and recognizing the p300 signal, that is a specific brain signal appearing when the right presented pattern like the right letter symbol emerges, thus enabling the disabled person to communicate easier with others. With this paper we aim to present what is currently done and what could be done in the near future to improve the life of disabled people.