Print ISSN: 1681-6900

Online ISSN: 2412-0758

Keywords : Neural Network

The Proposition of Three Approaching Ways to Implement Tan-sigmoid Activation Function in FPGA

Manal Talib Ali; Bassam H. Abed

Engineering and Technology Journal, 2022, Volume 40, Issue 2, Pages 311-321
DOI: 10.30684/etj.v40i2.2160

Hyperbolic tangents and Sigmoid are commonly used for Artificial Neural Networks as activation functions. The complex equation of the activation function is one of the most difficult to be implemented in hardware because containing division and exponential, which gives non-linear behavior. The challenge is building a tan-sigmoid function in hardware with efficient performance. Therefore, this work will focus on implementing the activation function in FPGA. To overcome this challenge, a different approach was proposed in this paper, efficient hardware-implemented for tan-sigmoid in terms of the number of slices occupied and the resources utilization are designed. In this work, three approaches are proposed: tan-sigmoid using the log approximation method, tan-sigmoid using segmentation method, and tan-sigmoid using the polynomial method. These approaches are efficiently implemented in the Xilinx Spartan-3A xc3s700a-4fg484 platform. Hardware synthesis and FPGA implementations illustrate that the proposed tan-sigmoid only takes up to 1% of logic resources in the first and second proposed approaches. While, 4% showed in the third proposed approach, with the best efficiency and significantly confirmed the lowest implementation costs than the traditional approach.

Design and FPGA Implementation of Intelligent Fault Detection in Smart Wireless Sensor Networks

Mohammed H. Hadi; Abbas Hussain Issa; Atheer Alaa Sabri

Engineering and Technology Journal, 2021, Volume 39, Issue 4A, Pages 653-662
DOI: 10.30684/etj.v39i4A.1951

In this paper, both the design and hardware of Fault Detection (FD) in Wireless Sensor Network (WSN) was implemented using FPGA NI myRIO kit, wireless temperature sensors network with small size, low cost, and low power consumption. Work data processing was performed using pattern recognition methods to detect residual generation. LabVIEW software environment was employed for system performance. In this paper. The design of the hardware circuit NI myRIO kit received temperature from the sensors. The examined system showed an ability to monitor and track any fault or fire that may occur; based on the results, if collected data is exceeded predetermined threshold, then the system is responding, a direct connection is using WIFI to process this data by LabVIEW.

Intelligent Monitoring for DC Motor Performance Based on FPGA

Bilal Z. Ahmed; Abbas H. Issa

Engineering and Technology Journal, 2016, Volume 34, Issue 13, Pages 2490-2499

This paper presents a fault monitoring of DC motors. A neural network is prepared to processes the inputs parameters “motor speed and current” collected from sensors and delivers condition states of the DC motors “good, fair or bad”. FPGA Spartan 3 kit board is used to implement the proposed monitoring network and the circuits are designed for data acquisition to makes an interface between motors analog collected data and FPGAs digitals inputs ports. The designed circuits are intended to gather analogs readings from the target motor and converting them into digitals to be compatibles with FPGAs inputs ports specifications. The neural networks which are designed based on backs propagation trainings are implemented using Xilinx Spartan-3A Starter FPGAs Kits boards.

Image Based Vehicle Traffic Measurement

Hamid M. Hasan

Engineering and Technology Journal, 2014, Volume 32, Issue 11, Pages 2722-2733

This research deals with measurement of the density of vehicles traffic. The traffic density is estimated from an image captured using the ordinary optical camera. An image processing methods is used and the edge of the objects is extracted. A two dimensional wavelet transform is used as a feature extraction. The extracted features were reduced by Multiple Region Centroid Estimation. A neural network is trained using many sets of images with different Traffic densities then it is used for traffic measurement. A classification rate of 98% can be achieved.

Edge Detection Using Scaled Conjugate Gradient Algorithm in Back propagation Neural Network

Walaa M. Khalaf; Mohammed Ali Tawfeeq; Kadhum Al-Majdi

Engineering and Technology Journal, 2014, Volume 32, Issue 2, Pages 385-395

This paper introduces a proposed method based on a backpropagation artificial neural network using Scaled Conjugate Gradient (SCG) training algorithm so as to gain the edges of any image. A new training image model is suggested to train this artificial neural network, then using this network to find the edges of any image. Computer experiments are carried out for extracting edge information from real images; the results presented are compared with those from classical edge detection methods like Canny. Using this new method does not need to tune any parameter to find the edge of any image, as well as using this method the false edges is reduced.

Designing a Nonlinear PID Neural Controller of Differential Braking System for Vehicle Model Based on Particle Swarm Optimization

Ahmed Sabah Al-Araji

Engineering and Technology Journal, 2014, Volume 32, Issue 1, Pages 197-214

This paper presents a nonlinear PID neural controller for the 2-DOF vehicle model in order to improve stability and performances of vehicle lateral dynamics by achieving required yaw rate and reducing lateral velocity in a short period of time to prevent vehicle from sliding out the curvature. The scheme of the discrete-time PID control structure is based on neural network and tuned the parameters of the nonlinear
PID neural controller by using a particle swarm optimization PSO technique as a simple and fast training algorithm. The differential braking system and front wheel steering angle are the outputs of the nonlinear PID neural controller that has automatically controlled the vehicle lateral motion when the vehicle rotates the curvatures. Simulation results show the effectiveness of the proposed control
algorithm in terms of the best transient state outputs of the system and minimum tracking errors as well as smoothness control signals obtained with bounded external disturbances.

Neural NetworkModeling of Oxidation Kinetics in Air of Steel-T21 Alloy Coated by Simultaneous Germanium-Doped Aluminizing-Silicon zing Process

Abbas Khammas; Mohanned M. H.AL-Khafaji

Engineering and Technology Journal, 2013, Volume 31, Issue 4, Pages 632-645

In this work a pack cementation of germanium-doped aluminum and silicon coatings on low alloy steel type-T21 has been applied. This gives significant improvement in the oxidation. Steel-T21 was coated with germanium-doped aluminizing-siliconizing. Diffusion coating was carried out at 1050oC for 6 h under an Ar atmosphere by simultaneous germanium-doped aluminizing-siliconzing process. Cyclic oxidation tests were conducted on the coated steel-T21 alloy in the temperature range oxide 300-900oC in air for 102 h at 3 h cycle. The results showed that the oxidation kinetics for coated system in air can be represented by parabolic curve .Oxide phases that formed on coated system are SiO2 and Cr2O3. A neural network model of oxidation kinetics has been proposed to model the oxidation kinetic. The neural model shows good agreement with the experimental data.

Design of Beam-Columns Using Artificial Neural Networks

Abdelmaseeh Bakos Keryou; Raid Rafi Al-Nima; Rafal Naheth Wadie

Engineering and Technology Journal, 2012, Volume 30, Issue 16, Pages 2843-2857

In this paper, manual design of beam-columns, based on the procedure adopted by
american society of steel construction, is described. an attempt has been taken to
apply artificial neural network to the design of steel beam-columns of hot-rolled
shapes. for this purpose, a set of data have been generated using the software package
staad pro, and then used in training and testing the neural network. the results showed
that artificial neural network after successful learning could specify the proper
sections with relatively high accuracy.

Semi – Chaotic Mutual Learning Platform for Key – Exchange Algorithm Using Neural Network

Enas H. Salih; Mohamad AB. Saleh; Mohammed Gheni Alwan

Engineering and Technology Journal, 2012, Volume 30, Issue 11, Pages 1971-1979

Neural network has been emerged the cryptography field as efficient tool for
both cryptanalysis and cryptography due to its amazing ability to explore solution
space for a given problem. One of the latest observations for the behavior of neural
networks is its ability to synchronize itself to other neural network based on mutual
learning rules; this phenomenon has been under the focus of specialist in
cryptographic field due to its significant usage as highly secure key exchange
This paper is presenting new approach to drive the synchronization based on
semi-chaotic mutual learning, where the output of each neural network will be
extracted through non-linear mapping to memory filled with balanced number of
1's and 0's as this paper will demonstrate.

Using Hamming Network to Decoding Binary Cyclic Code

Hind Abd Al-Razzaq

Engineering and Technology Journal, 2011, Volume 29, Issue 11, Pages 2093-2101

This work, efforts are concentrated on solving the problem of decoding
binary cyclic code, using hamming neural network. Therefore, this work
shows the ability of hamming network in solving one of the important
problems in coding theory. It presents the results of applying hamming
network as a decoding algorithm for cyclic code. The results prove the
relative efficiency of hamming network in decoding large linear cyclic
codes compared with other decoding algorithms.

Edge Detection Based on Standard Deviation Value and Back Propagation Algorithm of Artificial Neural Network

Ammar Sabr Majed; Mohammed Hussien Miry; Ali Hussien Miry

Engineering and Technology Journal, 2011, Volume 29, Issue 3, Pages 462-469

This paper presents a proposed neural network based edge detection
algorithm. we have used artificial neural network system to decide about whether
each pixel is edge or not. First standard deviation values are computed for mask
(3*3), Then after training a neural network system to recognize structural patterns
(these pattern represents edges), it decides on each pixel if its edge or not. Finally
we have test the proposed method on different images. Experimental results show
the ability and high performance of proposed algorithm.