Print ISSN: 1681-6900

Online ISSN: 2412-0758

Keywords : Artificial Neural Network

Determination Efficient Classification Algorithm for Credit Card Owners: Comparative Study

Raghad A. Azeez

Engineering and Technology Journal, 2021, Volume 39, Issue 1B, Pages 21-29
DOI: 10.30684/etj.v39i1B.1577

Today in the business world, significant loss can happen when the borrowers ignore paying their loans. Convenient credit-risk management represents a necessity for lending institutions. In most times, some persons prefer to late their monthly payments, otherwise, they may face difficulties in the loan payment process to the financial institution. Mainly, most fiscal organizations are considered managed and refined client classification systems, scanning a valid client from invalid ones. This paper produces the data mining idea, specifically the classification technique of data mining and builds a system of data mining process structure. The credit scoring problem will be applied using the Taiwan bank dataset. Besides that, three classification methods are adopted, Naïve Bayesian, Decision Tree (C5.0), and Artificial Neural Network. These classifiers are implemented in the WEKA machine learning application. The results show that the C5.0 algorithm is the best among them, it achieves 0.93 accuracy rates, 0.94 detection rates, 0.96 precision rates, and 0.95 F-Measure which is higher than Naïve Bayesian and Artificial Neural Network; also, the False Positive Rate in C5.0 algorithm achieves 0.1 which is less than Artificial Neural Network and Naïve Bayesian

A New Hybrid Technique for Face Identification Based on Facial Parts Moments Descriptors

Shaymaa M. Hamandi; Abdul Monem S. Rahma; Rehab F. Hassan

Engineering and Technology Journal, 2021, Volume 39, Issue 1B, Pages 117-128
DOI: 10.30684/etj.v39i1B.1903

Robust facial feature extraction is an effective and important process for face recognition and identification system. The facial features should be invariant to scaling, translation, illumination and rotation, several feature extraction techniques may be used to increase the recognition accuracy. This paper inspects three-moment invariants techniques and then determines how is influenced by the variation which may happen to the various shapes of the face (globally and locally) Globally means the whole face shapes and locally means face part's shape (right eye, left eye, mouth, and nose). The proposed technique is tested using CARL database images. The proposal method of the new method that collects the robust features of each method is trained by a feed-forward neural network. The result has been improved and achieved an accuracy of 99.29%.

Random Forest (RF) and Artificial Neural Network (ANN) Algorithms for LULC Mapping

Tay H. Shihab; Amjed N. Al-Hameedawi; Ammar M. Hamza

Engineering and Technology Journal, 2020, Volume 38, Issue 4A, Pages 510-514
DOI: 10.30684/etj.v38i4A.399

In this paper to make use of complementary potential in the mapping of LULC spatial data is acquired from LandSat 8 OLI sensor images are taken in 2019. They have been rectified, enhanced and then classified according to Random forest (RF) and artificial neural network (ANN) methods. Optical remote sensing images have been used to get information on the status of LULC classification, and extraction details. The classification of both satellite image types is used to extract features and to analyse LULC of the study area. The results of the classification showed that the artificial neural network method outperforms the random forest method. The required image processing has been made for Optical Remote Sensing Data to be used in LULC mapping, include the geometric correction, Image Enhancements, The overall accuracy when using the ANN methods 0.91 and the kappa accuracy was found 0.89 for the training data set. While the overall accuracy and the kappa accuracy of the test dataset were found 0.89 and 0.87 respectively

Investigation of the Effect of Nano Powder Mixed Dielectric on EDM Process

Rasha R. Elias

Engineering and Technology Journal, 2020, Volume 38, Issue 3A, Pages 295-307
DOI: 10.30684/etj.v38i3A.337

In this paper, Artificial Neural Network was adopted to predict the effect of current, the concentration of aluminum oxide (Al2O3) and graphite Nanopowders in dielectric fluid for the machining of Carbon steel 304 using Electrical Discharge Machining (EDM). The process variables were utilized to find their effect on Material Removal Rate (MRR), Surface Roughness (SR), and Tool Wear Rate (TWR). It was revealed from the experimental work that the addition of aluminum oxide and graphite Nanopowders into dielectric fluid maximizing MRR, minimized the SR and TWR at various variables. Minitab software was used in the design of experiments. Analysis of the process outputs of EDM indicates that graphite powder concentration greatly influencing SR also the discharge current whereas the current and Nanopowders concentration has more percentage of influence on the TWR and MRR.

Artificial Intelligent Technique for Power Management Lighting Based on FPGA

Hanan A. R. Akkar; Sameh J. Mohammed

Engineering and Technology Journal, 2020, Volume 38, Issue 2, Pages 232-239
DOI: 10.30684/etj.v38i2A.305

The modern technological advances gave rise to new intelligent ways of performance and management in various fields of our lives. The employment of the artificial intelligent techniques proved influential in enhancing the technological developments and in meeting the demands for new, more efficient, more reliable and faster ways of performing activities and tasks. Lighting systems are an important part of human life. For this reason, it is important to reduce and manage energy consumption properly. Light dimming paves the way for massive energy saving in lighting applications. The options include simply reducing the output during the night and achieve maximum saving with variable dimming. Advantage can be taken of off-peak times (no light needed) to reduce energy consumption significantly. Pulse Width Modulation (PWM) technique is used as dimming method. The proposed system offers intelligent management of lighting to reduce power consumption, extend lamp life and reduce maintenance. In this work, we will be using multiple sensors such as light dependent resistor (LDR) and Motion Sensor (PIR) for LED dimming system to achieve intelligent LED lighting system to manage energy consumption. The data collected by sensors is processed by Artificial Neural Network (ANN), which is implemented by using Field Programmable Gate Arrays (FPGAs), Spartan 3A starter kit that controls the light intensity of LED from changing the duty cycle of the PWM signals. FPGA was used to implement the design, because of the re-programmability of the FPGAs, which can support the re-configuration necessary to implement the design. VHDL program was used to describe the functions of all necessary components used. Xilinx ISE 14.7 design suite and MATLAB R2012A were used as software tools to perform Spartan 3A starter kit program. The Simulation results were obtained with Xilinx blocks found in MATLAB program.

A Hybrid Neural-Fuzzy Network Based Fault Detection and IsolationSystem for DC Motor of Robot Manipulator

Arkan A. Jassim; Abbas H. Issa; Qusay A. Jawad

Engineering and Technology Journal, 2019, Volume 37, Issue 8A, Pages 326-331
DOI: 10.30684/etj.37.8A.3

In this paper, the detecting and isolating fault that occurs in (actuator
and sensor) in robot manipulator, which is used as a mathematical model were proposed for fault detection, where the neural network was used to detect the fault. The neural network was trained on the data set obtained from the Input/output on the (DC motor).The output of the sensor or actuator was compared with the output of the model (neural network) after that the residual signal is used to detect the fault. The fuzzy logic circuit was used for fault isolation that is depending on the residual signal from any sensor or actuator that faults. There are three types of faults detected and isolated in this study abrupt fault, incipient fault and intermittent fault. The Matlab R2012a was used to the model steady state designed and simulated .The model has a high capacity for detecting faults.

Predictive Modeling of Multilayer Graphene Growth by Chemical Vapour Deposition on Co-Ni/Al2O3 Substrate using Artificial Neural Network

May A. Muslim; Zainab Yousif; Mohamed A. Abdel Ghany

Engineering and Technology Journal, 2019, Volume 37, Issue 1C, Pages 113-119
DOI: 10.30684/etj.37.1C.18

The uniqueness of multilayer graphene as extremely high carrier mobility, tune-able band gap and high elasticity has made it be considered as a high prospect engineering material that can be employed for several applications such as solar cells, field effect transistors, super-capacitors, batteries and sensors. In this study, the application of Artificial Neural Networks (ANN) for the predictive modeling of multilayer graphene (MLG) growth by chemical vapor deposition (CVD) on Co-Ni/Al2O3 substrate was investigated. Data comprises temperature, catalyst compositions, ethanol flowrates were generated using central composite experimental design and employed to obtain the MLG yield as the response. The data were subsequently used for predictive modeling using ANN. The findings show that the predictive values of the MLG yields were in good agreement with those obtained from the experimental runs having a coefficient of determination (R2 ) of 0.988.

Mapping LCLU Using Python Scripting

Oday Z. Jasim; Khalid I. Hasoon; Noor E. Sadiqe

Engineering and Technology Journal, 2019, Volume 37, Issue 4A, Pages 140-147
DOI: 10.30684/etj.37.4A.5

Land cover land use changes constantly with the time at local, regional, and global scales, therefore, remote sensing provides wide, and broad information for quantifying the location, extent, and variability of change; the reason and processes of change; and the responses to and consequences of change. And considering to the importance of mapping of (LCLU). For that reason this study will focus on the problems arising from the traditional classification (LCLU) that based on spatial resolution only which leads to prediction a thematic map with noisy classes, and using a new method that depend on spectral and spatial resolution to produce an acceptable classification and producing a thematic map with an acceptable database by using artificial neural network (ANN) and python in additional to other program. In this study the methods of classification were studied through using two images for the same study area , rapid eye image which has three spectral bands with high spatial resolution(5m) and Landsat 8 image (high spectral resolution with eight bands), also several programs like ENVI version 5.1, Arc GIS version 10.3, Python 3, and GPS. The result for this research was sensuousness as geometrics accuracy accepted in map production.

A Simplified Recurrent Neural Network Trained by Gbest-Guided Gravitational Search Algorithm to Control Nonlinear Systems

Omar F. Lutfy; Ahmed L. Jassim

Engineering and Technology Journal, 2018, Volume 36, Issue 12A, Pages 1290-1301
DOI: 10.30684/etj.36.12A.11

This paper presents a feedback control strategy using a Simplified Recurrent Neural Network (SRNN) for nonlinear dynamical systems. As an enhancement for a previously reported modified recurrent network (MRN), the proposed SRNN structure is used as an intelligent Proportional-Integral-Derivative (PID)-like controller. More precisely, the enhancement in the SRNN structure was realized by employing unity weight values between the context and the hidden layers in the original MRN structure. The newly developed Gbest-guided Gravitational Search Algorithm (GGSA) was adopted for optimizing the parameters of the SRNN structure. To show the efficiency of the proposed PID-like SRNN controller, three different nonlinear systems were considered as case studies, including a control valve, and a complex difference eq.. From an extensive set of evaluation tests, which includes a control performance test, a disturbance rejection test, and a generalization test, the proposed PID-like SRNN controller demonstrated its effectiveness with regards to precise control and good robustness and generalization abilities. Furthermore, compared to other Neural Network (NN) structures, including the original MRN and the Multilayer Perceptron (MLP) NN, the SRNN structure attained superior results due to the utilization of a reduced set of parameters. From another study, the GGSA accomplished the best optimization results in terms of control precision and convergence speed compared to the original Gravitational Search Algorithm (GSA).

Applying Modern Optimization Techniques for Prediction Reaction Kinetics of Iraqi Heavy Naphtha Hydrodesulferization

Zaidoon M. Shakor; Anfal H. Sadeiq

Engineering and Technology Journal, 2018, Volume 36, Issue 11A, Pages 1171-1175
DOI: 10.30684/etj.36.11A.6

In this study, a powerful modern optimization techniques such as Genetic Algorithm (GA), Particle Swarm Optimization (PSO) and Artificial neural network (ANN) were applied to estimate the optimal reaction kinetic parameters for Heavy naphtha Hydrodesulferization (HDS), the hydrodesulferization unit located in AL-Daura refinery-Baghdad/Iraq. The reactions was carried out in a fixed-bed reactor packed with Co-Mo/γ-Al2O3 catalyst and the operating was 315-400 °C temperature 35 bar Pressure and 0.5-2.1 hr-1 liquid hourly space velocity. The result showed that hydrodesulferization of heavy naphtha follows the pseudo-first order reaction kinetics. This study signifies that the reaction kinetic parameters calculated by Genetic Algorithm was found to be more accurate and gives the highest correlation coefficient (R2= 0.9507) than the other two methods. ANN technology by using the topology of (3-3-1-1) provides an effective tool to simulate and understand the non-linear behavior of the process. The model result showed very good agreement with the experimental data with less than 5%. mean absolute error.

New Strategies for Associative Memories

Azad R. Kareem; Saja A. Talib

Engineering and Technology Journal, 2018, Volume 36, Issue 2A, Pages 207-212
DOI: 10.30684/etj.36.2A.13

Associative memory is a neural network used to save collection of input and output data at its layers. Each output data is produced coincide with a given input. It can be useful as an artificial memory in many applications like (military, medical, data security systems, error detection and correction systems …etc.). There are two matters which limit the uses of associative memory; the limited storage capacity, and the error occurred in the reading of output data. A modified strategy is suggested to overcome these limitations by introducing a new algorithm to the design of the associative memory. This method provides a software solution to the problems. The obtained results from the test examples proved that the proposed associative memory net could train and recall unlimited patterns in different sizes efficiently and without any errors.

Multibiometric Identification System based on SVD and Wavelet Decomposition

R.A. Hussein; H.A. Jeiad; M.N. Abdullah

Engineering and Technology Journal, 2017, Volume 35, Issue 1A, Pages 61-67
DOI: 10.30684/etj.2017.127311

Biometric systems refer to the systems used for human recognition based on their characteristics. These systems are widely used in security institutions and access control. In this work three biometric sources were used for identification purposes. Singular value decomposition (SVD) was employed as a tool for feature extraction and artificial neural network (ANN) was used as pattern recognition for the model. High accuracy was obtained from this work with 95% recognition rate.

Surface Roughness Prediction Using Circular Interpolation Based on Artificial Neural Network in Milling Operation

Maan Aabid Tawfiq; Ahmed A.A. Duroobi; Safaa K. Ghazi

Engineering and Technology Journal, 2016, Volume 34, Issue 5, Pages 983-998

This paper presents a method togenerate tool path and get G-codes for complex shapes depending on mathematical equations without using the package programs that use linear interpolation. Circular interpolation (G02, and G03) were used to generate tool path. This needs to define the tool radius and radius of curvature in addition to the cutting direction whether clockwise or counter clockwise. In addition many other factors had been considered in the machining process of the proposed surface to find the best tool path and G-code. Side step, feed rate and cutting speed had been studied as machining factors affecting tool path generation process. Artificial Neural Network technique had also been considered to find the best tool path depending on the cutting parameters proposed while surface roughness was the characteristic that the tool path process and G-code generation depend on. The impact of the machining parameters on the surface roughness was determined by the use of analysis of variance (ANOVA) that detects more influence for side step (85%, 53%, and 67%).From this study, it has been learned that less side step (0.2) mm and feed speed (1000) mm/min and high value for cutting speed (94.2) m/mingive better tool path to be used in machining operations. This study would help engineers and machinists to select the best tool path for their products.

Computation of Seepage through Homogenous Earth Dams with Horizontal Toe Drain

Raad Hoobi Irzooki

Engineering and Technology Journal, 2016, Volume 34, Issue 3, Pages 430-440

This investigation concerns to find a new equation for computing the quantity of seepage through homogenous earth dam with horizontal toe drain. For this purpose the computer program SEEP/W (which is a sub-program of Geo-Studio) was used. The SEEP/W runs were carried out with three different downstream slopes of the dam, three different upstream slopes, three variable horizontal toe drain lengths, three different free boards, three different top widths and three different heights of the dam. For each run the quantity of seepage was determined. The results show that the seepage discharge increased with increasing upstream slope, downstream slope, upstream reservoir water depth and length of horizontal toe drain. Also, the results show that the seepage discharges decreased with increasing the top width of the dam and the height of the free board. Using SEEP/W results with helping a dimensional analysis theory, a new easy and reliable empirical equation for computing seepage discharge through homogenous earth dams with horizontal toe drain was developed. The analysis of the results by Artificial Neural Network (ANN) shows that the length of horizontal toe drain (L) is the more geometrical variable effect on the seepage discharge, while the upstream slope (tanθ) of the earth dam has a little effect.

Neural Network Modelling of Tds Concentrations in Shatt Al-Arab River Water

Ahmed Naseh Ahmed; Ammar S. Dawood

Engineering and Technology Journal, 2016, Volume 34, Issue 2, Pages 334-345
DOI: 10.30684/etj.34.2A.12

River water salinity is a big concern in many countries, considering agricultural and drinking usages. Therefore, prediction of amount of Total Dissolved Solid (TDS) is a necessary tool for planning and management of water resources. Shatt Al-Arab river basin in Basrah which is located in south of Iraq suffer from high salinity, therefore use of the water for irrigation and drinking has become problematic. In this regard, prediction of future TDS of Shatt Al-Arab river basin was studied using Artificial Neural Network (ANN).
Data measured monthly from January 2007 up to December 2012 at monitoring station in the middle point along to the Shatt Al-Arab river has been used for training of the selected ANN.
Some of water quality parameters such as, power of hydrogen (pH), Total Hardness (TH), Magnesium hardness (MgSO4), Calcium hardness (CaSO4), Chlorides (Cl), Sulphates (SO4), turbidity (TU) and electrical conductivity (EC) were considered as inputs for the ANN and Total Dissolved Solid (TDS) was the output of the model.
The validation of the neural network model showed very good agreement for predictions of the TDS concentrations between observed and simulated values.
The coefficient of correlation (R), during the validation process was found to be (1), and the mean squared error (MSE) was (0.075). This work supports the concept that the neural network approach is a successful method of modelling such complex and nonlinear behavior of TDS in the rivers with different environmental conditions.

Hydraulic Characteristics of Flow Over Triangular Broad Crested Weirs

Raad Hoobi Irzooki; Mohammad Faiq Yass

Engineering and Technology Journal, 2015, Volume 33, Issue 7, Pages 86-96

In the present work, the hydraulic characteristics of flow over triangular broad crested weirs with triangular front or back face have been experimentally studied. The main objective of this research is to obtain empirical equation to estimate the value of the discharge coefficient (Cd) for this kind of weir and determine the factors that affect on it. For this purpose 18 models were constructed with different dimensions made of plexiglass and were tested in a laboratory flume of 6m length, 30cm width and 40cm height. These models divided into two groups, each group consists of 9 models. In the first group 108 experiments were conducted by changing the upper face angle of the weir three times (90°, 120°, 150°), the angle of the triangular front or back face (α) is also changed three times (90°, 120°, 150°), for each model six different discharges were passed. In the second group 54 experiments were carried out on models with a straight face on the front and back (α=180°) with changing the upper face angle (θ) three times (90°, 120°, 150°) and changing the height of the edge of the weir (P) three times ( 20 , 18 , 16 cm), for each model six different discharges were passed. Dimensional analysis was performed to obtain the dimensionless parameters that the discharge coefficient (Cd) depends on it. Results showed that the change in the angle of the triangular front or back face (α) have little effect on the discharge over these weirs, while it was noted that the height of the edge of the weir (P) affects on the discharge coefficient, where (Cd) increased with increasing (P). Also, the upper face angle of the weir (θ) has an effect on the discharge coefficient, where the discharge coefficient increased with decreasing the value of angle (θ). A simple empirical equation was predicted, in terms of the application, for the calculation of the discharge coefficient (Cd) of weirs that used in this study, there was a good agreement between the results obtained from this equation with the experimental results.

Artificial Neural Networks Based Fingerprint Authentication

Abbas H. Issa

Engineering and Technology Journal, 2015, Volume 33, Issue 5, Pages 1255-1271

Fingerprint authentication and recognition is an important subject that has been widely used in various applications because of its reliability and accuracy in the process of authenticating and recognizing the person's identity. In this paper, an Intelligent Fingerprint Authentication Model (IFAM) based upon the neural network has been proposed. The proposed work consists of two main phases which are the features extraction and the authentication. The features extraction phase has been regarded via proposing a statistical and geometrical approach for determining and isolating the features of the fingerprint images. The proposed approach is called the Features Ring Approach which is abbreviated by FRA. The approach creates a circular ring centered at the core point of the fingerprint to bind the valuable features that are invariant under rotation and translation. The radius of the outer circle of the ring is suggested to be variable to give a variable area for the established circular ring.
The authentication phase of IFAM suggests the neural network to hold the job of verification of the extracted feature patterns resulted by FRA for a fingerprint image of certain person. This is done using a neural network trained with a collection of features patterns extracted from fingerprint images. Backpropagation (BP) is suggested as a training algorithm for the structured neural network.

Investigating Forward kinematic Analysis of a 5-axes Robotic Manipulator using Denavit-Hartenberg Method and Artificial Neural Network

Israa R. Shareef; Iman A. Zayer; Izzat A. Abd Al Kareem

Engineering and Technology Journal, 2014, Volume 32, Issue 11, Pages 2700-2713

Robot Forward kinematic equations' analysis is an essential and important manner to analysis the position and orientation of the end effectors of a robotic manipulator, where in this paper, Denavit-Hartenberg notation and method (D – H) is used to represent the relative kinematic relationships precisely between each two adjacent links of this robot , besides a kind of artificial neural network (ANN) which is known as the supervised learning training sets network is investigated to solve the problem of kinematic analysis of this laboratory five - axes robot, it shows that the using of the artificial intelligence method which is the neural networks had offered the facility of dealing with this non linear robotic system by a simple manner with acceptable faster solution as compared with the traditional forward kinematic equations analysis method .

OFDM Channel Estimation Based on Intelligent Systems

Ismail Mohammad Jaber; Hanan A. R. Akkar; Haraa Raheim Hatem

Engineering and Technology Journal, 2014, Volume 32, Issue 2, Pages 305-324

This work is dedicated to the study of reducing Bit Error Rate (BER) when transferring data in the system Orthogonal Frequency Division Multiplexing (OFDM) by estimating the carrier channel in different ways. The proposal design for Artificial Neural Network (ANN) is considered as a tool to improve performance BER and compared with the traditional method based on the use of the Least Square estimation algorithm (LS) to estimate the impulse response of frequency selective Rayleigh fading channel. A MATLAB 7.14 program is used in simulation.
The proposed method which integrates algorithm LS with ANN includes the following:
1. Training the neural network by Back-Propagation (BP) and using the trained neural network with algorithm (LS) to estimate the channel in different paths.
2. Using Resilient Back propagation algorithm (RProp)in the training of the neural network.
3. UsingLevenberg-Marquardt algorithm (LM) in the training of the neural network.
4.The comparison of results between the traditional method and the proposed method when taking BER = 0.001 at various tracks (one path, two path and three path) and showed that there profit of (1.5dB, 2dB, 2dB) between using the traditional method and the proposed method using RProp algorithm and a profit of (2dB,3dB, 2dB) using an algorithm LM. There is also comparison between the performance ofRProp algorithm and LMalgorithm and the results showed that the LM algorithm better thanRProp algorithm.

Prediction of Square Footing Settlement under Eccentric Loading on Gypseous Soil through Proposed Surface for Dry and Soaked States

Bushra S. Z. Albusoda; Abdul-Kareem E; R. S.Hussein

Engineering and Technology Journal, 2013, Volume 31, Issue 20, Pages 217-237

Gypseous soils as any other soils deform under loading, this deformation differs greatly between its dry state and its soaked state. This deformation also differs when the loading is applied with eccentricity.
An experimental work was conducted on a square footing model (100 mm  100 mm) above gypseous soil 450 mm thick. Loading was applied at the center of the footing (e/B = 0) and at an eccentricity of (e/B = 0.05, 0.1, 0.15, 0.2) for its dry state and its soaked state. Settlement was obtained at the center and at the base soil of the footing for each state.
The data obtained was normalized and a proposed surface was obtained for each of the two states (dry and soaked) and at two places (center and edge). Four proposed equations were obtained represented four cases of research i) Dry center, ii) Dry edge, iii) Soaked center, and iv) Soaked edge. The four equations showed very good agreement with the data obtained from the experiment.
Artificial Neural Network model was also used to obtain a neural network representing the proposed surface for the abovementioned four cases and also a very good agreement was obtained.
It is concluded that a proposed surface for the central and eccentric loading on square footing for gypseous soil showed a good agreement with the experimental data and therefore may be used for settlement prediction.

ECG Signal Diagnoses Using Intelligent Systems Based on FPGA

Ali M. Abdul Kareem; Hanan A.R. Akkar

Engineering and Technology Journal, 2013, Volume 31, Issue 7, Pages 1351-1364

This paper presents the use of Particle Swarm Optimization (PSO), neural networks with the most promising supervised learning algorithms for automatic detection of cardiac arrhythmias based on analysis of the Electrocardiogram (ECG). Artificial Neural Network (ANN) has three layers with ten nodes in the input layer, five nodes in the hidden layer and five nodes in the output layer, which is trained using the PSO algorithm. The trained network was able to classify the ECG signal in normal signal, atrial flutter, ventricular tachycardia, sever conducting tissue and wandering a trial pacemaker. Field Programmable Gate Arrays (FPGAs) have been used to implement ANN trained by the supervised learning algorithms and PSO, because of their speed benefits, as well as the re-programmability of the FPGAs which can support the reconfiguration necessary to program a neural network. A VHDL Design of ANN platform is proposed to evolve the architecture ANN circuits using FPGA-Spartan 6 Evaluation board. The VHDL design platform creates ANN design files using WebPACKTM ISE 13.3 program. All the algorithms used to train the ANN showed high effectiveness with 100% classification.

Text Hiding Using Artificial Neural Networks

Haider Tarish Haider; Faiq Sabar Baji; Ahmad Saeed Mohammad

Engineering and Technology Journal, 2012, Volume 30, Issue 20, Pages 3553-3564

The growth of information technology and data transfer led to increase the data
attacks, so that information security becomes an important issue to keep the data
saved during information exchanges in computer networks. Steganography
techniques used to protect the information from being detected. The art of
steganography will hide secret information into cover data, which will be sending
without any change so the attack does not recognize any change into cover image.
This paper use the Steganography and artificial neural networks to presents an
information hiding procedure for hiding text in cover image, the secret text will be
converted to binary code, also the cover image will be converted to the binary data
in form of vectors. The supervised learning of neural networks will use binary
patterns of hidden text as set of input values, and the corresponding cover image
data as target that used as teacher signal to neural network. The generated weights
from neural network and the coordinate of data block of cover image have been
saved and then used to extract hidden text data.

Early Detection of Disease-Viral Hepatitis Type-C Using Elman Artificial Neural Network

Ghaidaa Kaain Salih

Engineering and Technology Journal, 2012, Volume 30, Issue 12, Pages 2150-2164

The problem of founding important information in complex medical images which are needed in diagnosing of diseases with the complex data considered as one of the predication problem these days, so it is necessary to find aided means for diagnosing process. Artificial neural network (ANN) is one of them. This paper deals with the designing and implementation a classification ANN module for Lever Hepatitis(class-C)
or type-C which doesn’t have any vaccine these days. The different in diagnosing between hepatitis and other liver diseases is often difficult on purely clinical grounds in addition the damage to the liver causes changes in the pattern of the serum enzymes and
in recent years this has led to develop disease testing and its vaccine. Elman neural networks (NN) have been applied for automated detection of various medical diseases. Like its application on blood sample tests extracted from on line microscope (like it used
in this research).That feature selection is an important issue by removing features that do not encode important data information from the images used.This helps physicians to extract features which aided them in diagnosing process. Kernal principle component analysis (PCA) is used to represent blood images as eigen-features of training images in addition to extract mathematical module for classification of it. Finally a neural network (NN) is trained to perform the typical images and classify them (diagnosing process). The produced NN system produces used a matlab package in order to design and diagnose the proposed module. The object of this system used in our work is to diagnosing lever Hepatitis type-C in samples of blood images wherever difficulties in practical experiments by finding an optimal feature from specialists whom work in laboratories.

Training Artificial Neural Network Using Back-Propagation & Particle Swarm Optimization for Image Skin Diseases

Hanan A. R. Akkar; Samem Abass Salman

Engineering and Technology Journal, 2011, Volume 29, Issue 13, Pages 2739-2755

This work is devoted to compression Image Skin Diseases by using Discrete Wavelet Transform (DWT) and training Feed-Forward Neural Networks (FFNN) by using Particle Swarm Optimization(PSO) and compares it with Back-Propagation (BP) neural networks in terms of convergence rate and accuracy of results .The comparison between the two techniques will be mentioned. A MATLAB 6.5 program is used in simulation.
The structure Artificial Neural Network (ANN) of training image skin diseases is proposed as follows: 1- The proposed structure of NN that performs three compressions Images Skin training by BP algorithms with log sigmoid activation function, and three neurons in output layer.
2- The proposed structure of FFNN using PSO that performs three compressions Images Skin with hardlim activation function, and three neurons in output layer. The results obtained using PSO are compared to those obtained using BP. Learning iterations (602-4700 epoch), convergence time (1sec.- 100 sec.), number of initial
weights (1set - 75set), number of derivatives (0 - 38 derivatives) and accuracy (60% - 100%) are used as performance measurements. The obtained Mean Square Error (MSE) is 7 10 - to check the performance of algorithms. The results of the proposed neural networks performed indicate that PSO can be a superior training algorithm for
neural networks, which is consistent with other research in the area.

Training Artificial Neural Networks by PSO to Perform Digital Circuits Using Xilinx FPGA

Hanan A. R. Akkar; Firas R. Mahdi

Engineering and Technology Journal, 2011, Volume 29, Issue 7, Pages 1329-1344

One of the major constraints on hardware implementations of Artificial Neural
Networks (ANNs) is the amount of circuitry required to perform the multiplication
process of each input by its corresponding weight and there subsequent addition. Field
Programmable Gate Array (FPGA) is a suitable hardware IC for Neural Network (NN)
implementation as it preserves the parallel architecture of the neurons in a layer and
offers flexibility in reconfiguration and cost issues. In this paper the adaption of the
ANN weights is proposed using Particle Swarm Optimization (PSO) as a mechanism
to improve the performance of ANN and also for the reduction in the ANN hardware.
For this purpose we modified the MATLAB PSO toolbox to be suitable for the taken
application. In the proposed design training is done off chip then the fully trained
design is download into the chip, in this way less circuitry is required. This paper
executes four bit Arithmetic Logic Unit (ALU) implemented using Xilinx schematic
design entry tools as an example for the implementation of digital circuits using ANN
trained by PSO algorithm.

Transmission System On –Line Fault Location Using ArtificialNeural Network

Adil Hameed Ahmed; Hatim Ghadhban Abood

Engineering and Technology Journal, 2010, Volume 28, Issue 5, Pages 964-979

In this work, protection systems for overhead transmission lines are
investigated and an efficient technique for on –line fault location based on
Artificial Neural Network(ANN ) is suggested. First, Studying and investigating
the power transmission lines short circuit modeling and analysis, and then
developing a MATLAB programs to calculate fault currents and voltages for
different fault types depending on the location of the fault in the transmission line
and finding the location of this fault. The proposed technique for the fault
location is the two -end data technique. The pre-fault data plus the fault data
construct a training set for the neural network programs which contain two types,
one for fault detection and classification, and the other for the fault location. Then,
these programs are applied on the Iraqi super grid (400 kV).

Integrating Neural Network With Genetic Algorithms For The Classification Plant Disease

Alia Karim Abdul Hassan; Sarah Sadoon Jasim

Engineering and Technology Journal, 2010, Volume 28, Issue 4, Pages 686-701

In this work Aِِrtificial Neural Network (ANN) is used as a classifier capable of
recognizing the most important features of the plant disease, with minimum error
value. Genetic algorithm has been used to minimize error values of the ANN
classifier. Error value of ANN classifier is defined as more than (%5). This ratio
is a threshold (cut-of-value) to determine if GA is executed or not after the ANN
classifier execution. Genetic algorithm execution results in either optimal solution
(%100) recognition or suggests a modified parameter to the ANN classifier
(specifically learning rate and number of neurons).The result obtained from
integrating neural network with genetic algorithm for classification plant diseases
indicates that the classifier recognizes most of input pattern with accuracy (96%).
Integrating neural network with genetic algorithm for classification plant diseases
implemented using Visual Basic version 6 programming