Abbas Toloie- Eshlaghy et al./ Elixir Mgmt. Arts 41 (2011) 5877-5881 5877
Introduction
Smart systems have many applications as one of the modern
tools in different fields of sciences for optimization and
prediction. These evolutionary achievements in information
technology have allowed processing of the information in
parallel. Human being has tried to imitate brain mechanism to
discover the secrets of creation in order to find this complex
architecture of creation leading to development of smart
systems.
This major change in human knowledge led to an important
achievement in science and scientific methods and has been used
as scientific complex tool to solve problems of human beings.
This tool has been mostly applied in reliability and many
progresses were made in this field.
In this paper which is done with use of one of the advanced
techniques in this field i.e. neuro-fuzzy networks, we try to
predict reliability of tire in passenger cars. Neuro-fuzzy
networks –based models are able to predict variable behavior
property and desirably and can predict its future behavior. Our
goal is to model a nonlinear prediction for reliability.
With regard to promotion and development of automotive
industries and competitive atmosphere in this industry,
important issues such as useful life of the car and safety were
raised. In this regard, reliability entered automotive industry as
an important index which can answer some of the above cases.
Reliability is defined as the possibility that a component or
system performs its assumed tasks properly under defined
conditions and at special time.
Material and methods
Fuzzy deduction system
Fuzzy set theory is a relatively new mathematical theory
which was presented by Iranian student, Professor Lotfi
Asgarzadeh alias Zadeh, in order to find mathematical models
which are compatible with human thinking and deductive
manner as well as natural and real models. Word fuzzy means
confused and unspecified and is used for description of unreal
and unspecified phenomena. A fuzzy system is a system based
on fuzzy rules of if-then which images input space on output
space with use of fuzzy variables and fuzzy decision making. In
summary, starting point of a fuzzy system design is to find a set
of if-then rules from knowledge of the experts or the studied
knowledge.
Artificial neural network
Main idea of neural networks is human brain. At present,
scientists try to have access to information processing by
researching function of human brain and algorithm and
modeling of its internal mechanism. Main component of these
models is information processing unit which simulates function
of neural neurons which are called neural neuron.
Although it seems that a neuron is a simple entity, neurons
can be connected to each other in a system structure. More than
40 traits of natural neurons have been recorded in information
processing. More information about neurons and neural system
is presented in detail.
Human brain neurons ANN
It is a system which has been created by connecting some
factors to each other. It is inspired by biological study of neural
systems of living creatures especially human being. On the other
hand, neural networks try to make the machines which act like
human brain. In order to make such machines, some components
which act like biological neurons are used. Function of a neural
system is as follows: When an input model is defined for it, it
should be able to produce an output model (Toloie Eshlaghy,
2008).
Figure 1 shows simple scheme of a neural system model. As
figure 1 show, each neural system has been composed of three
layers including: input layer, output layer and hidden layer.
Figure 1: simple scheme of a neural system model
Tele:
E-mail addresses: toloie@gmail.com
© 2011 Elixir All rights reserved
Prediction of reliability of vehicle tire with use of Neuro-fuzzy networks
Abbas Toloie- Eshlaghy and Parisa Sadat Tabatabaei Kooshki
Department of Industrial Management, Science and Research Branch, Islamic Azad University, Tehran, Iran.
ABSTRACT
In this paper, one of the artificial intelligence techniques called networks for prediction of
reliability of the tire is neuro-fuzzy network. At first key indices which are effective on
reliability of tire are identified and then some rules have been designed and applied for
system training with use of experts’ views. ANFIS system implements Takagi –Sugno
fuzzy system as a system and inference and deduction of the system are mixed and an
efficient tool is provided for simulating a nonlinear mapping. In the performed simulations,
some 4-input networks are different with membership functions and have been studied in
different iterations for prediction and finally the model which had high predictability was
selected as optimal model.
© 2011 Elixir All rights reserved.
ARTICLE INFO
Article history:
Received: 3 October 2011;
Received in revised form:
25 November 2011;
Accepted: 12 December 2011;
Keywords
Reliability,
Fuzzy system,
ANFIS neural –fuzzy system.
Elixir Mgmt. Arts 41 (2011) 5877-5881
Management Arts
Available online at www.elixirpublishers.com (Elixir International Journal)