J. Basic. Appl. Sci. Res., 3(6)850-859, 2013
© 2013, TextRoad Publication
ISSN 2090-4304
Journal of Basic and Applied
Scientific Research
www.textroad.com
*
Corresponding Author: Khashayar Teimoori, Young Researchers and Elites Club, Science and Research Branch, Islamic
Azad University, Tehran, Iran Email: kh.teimoori@srbiau.ac.ir
The Application of Synthetic Neural Network in the Process Control Device
Adel Maghsoudpour
a
, Khashayar Teimoori*
b
, Mustafa Korzaledin
c
a, c
Department of Mechanical and Aerospace Engineering, Science and Research Branch, Islamic Azad
University, Tehran Iran
b
Young Researchers and Elites Club, Science and Research Branch, Islamic Azad University, Tehran, Iran
ABSTRACT
As a matter of an academic fact, modeling neuron is one of the most influential fundamentals in guiding and
leading as well as the neural network efficiency, nonetheless the setting of the connections (Topology) in the
network is a major role from affecting viewpoints. In this research the prediction of characteristics such as;
temperatures, speeds and other parameters are as the outputs of an industrial process control device. First the
outputs related to the speed, pressure and temperature are collected from the prepared tables of renowned
corporations and then with the application of MATLAB software and the artificial neural network methods, a
pertinent cliché for data is selectively collected. The final goal of this study is to attain the ideal regression
which is recommended to be dealt with a Pre-assumption method. Besides, the final data calculated and
sketched by graphs and diagrams were regulated then the variety of the mentioned methods in accord with their
goals compared with one another and finally best ones of theirs chosen to be applied as the selective methods.
KEYWORDS: Artificial Neural Network, Multi layer Perceptron, Industrial System Information, neurons,
post-issuing algorithm, Process Control Device
1. INTRODUCTION
One of the most significant and fundamental issues in designing the system identification in industrial
segments is to identify the industrial system information (SI). There are different methodic approaches to
identify it. Recognizing the industrial SI is completely and wholly accomplished providing that the transfer
function of the SI is obtained nonetheless because of the complexity as well as being nonlinear of the industrial
SI, the transfer function of the mentioned SI is hardly obtainable for some clues. The synthetic neural network is
order to identify the industrial SI have demonstrated many capabilities on their own. The learning and correction
of neural system parameters illustrated on figure 1 accordingly.
Figure 1: Sample Actual Neuron (Adapted from Drugabuse.gov Neural Network Study)
From then on it was enabled to sequentially overcome the learning stages and the experimentations which
were majorly substituted as the major SI in computers so as to permit the control engineers to experiment the
various controllers on the system information which therefore is enabled to monitor by computers. This is led to
select the best function for the SI to design the target controller in order to finalize the SI controller. This
research is to identify the existing UNIPRO devices in process control laboratories. Having input and output
gates, the process of the identification stages is capably perceivable. Neural Networks are modeled as simplified
as possible out of the real neural SI which is massively applied to solve solutions at various sciences. The
domination of these networks is extremely vast which includes from the different categorizing applications to
the interpolating, estimating, and discovering applications. Having applied the neural networks and the
850