Hybrid fuzzy support vector classifier machine and modified genetic algorithm
for automatic car assembly fault diagnosis
Qi Wu
*
Jiangsu Key Laboratory for Design and Manufacture of Micro-Nano Biomedical Instruments, Southeast University, Nanjing 211189, China
article info
Keywords:
Fuzzy
m
-support vector classifier machine
Triangular fuzzy number
Genetic algorithm
Fault diagnosis
abstract
This paper presents a new version of fuzzy support vector machine to diagnose automatic car assembly
fault diagnosis, the input and output variables are described as fuzzy numbers and the metric on fuzzy
number space is defined. Then by combining the fuzzy theory with
v
-support vector machine, the fuzzy
v
-support vector classifier machine (Fv-SVCM) is proposed. A fault diagnosis method based on Fv-SVCM
and its relevant parameter-choosing algorithm is put forward. The results of the application in car assembly
diagnosis confirm the feasibility and the validity of the diagnosis method. Compared with the fuzzy neural
network (FNN) model, Fv-SVCM method requires fewer samples and has better estimating precision.
Ó 2010 Elsevier Ltd. All rights reserved.
1. Introduction
Since the maintenance has significant impacts in car manufac-
turing industry, it has received a deep attention from the expert
and practical maintenance. According to study, maintenance costs
are a major part of the total operating costs of all manufacturing
and production plants, which can make or break a business.
Depending on the specific industry, maintenance costs can repre-
sent from 15% to 40% of the costs of goods produced. In fact, these
costs are associated with maintenance labour and materials and
are likely to go even higher in the future with the addition of fac-
tory automation through the development of new technologies.
Nowadays, the development of maintenance strategy is sup-
ported by computer technology both in hardware and software.
A recent developed method is using artificial intelligent (AI) tech-
niques as tool for maintenance routine. Based on the idea perform-
ing an excellent and easy maintenance program; it leads the
practical maintenance to create an intelligent maintenance system.
Intelligent maintenance consists of parts (hardware and software),
which are possible for the system to do maintenance routine in
such a way like human being. Application of expert system (ES)
as a branch of AI in maintenance is one of solution. The basic idea
of ES is simply that expertise, which is the vast body of task-spe-
cific knowledge, is transferred from a human to a computer. This
knowledge is then stored in the computer and users call upon
the computer for specific advice as needed. The computer can
make inferences and arrive at a specific conclusion. Then, like hu-
man consultant, it gives advice and explains, if necessary, the logic
behind the advice (Demetgul, Tansel, & Taskin, 2009; Wu & Liu,
2009).
Support vector machine (SVM) is a relatively new computa-
tional learning method based on the statistical learning theory
and can serve as ES. Some seminal papers introduced below to
show the development of SVM that originally came from statistical
learning theory (SLT) developed by Vapnik (1995). SVM is based on
Vapnik–Chervonenkis theory (VC-theory) that recently emerged as
a general mathematical framework for estimating (learning)
dependencies from finite samples. This theory combines funda-
mental concepts and principles related to learning, well-defined
formulation, and self-consistent mathematical theory. Moreover,
conceptual framework of VC-theory can be used for improved
understanding various learning method developed in statistics,
neural networks, fuzzy systems, signal processing, etc. A major
conceptual contribution of VC-theory is revisiting the problem
statement appropriate for modern learning method that makes a
clear distinction between the problem formulation and solution
approach used to solve the problem.
As time goes by, SVM becomes famous and popular in machine
learning community due to the excellence of generalization ability
than the traditional method such as neural network. Therefore,
SVM have been successfully applied to a number of applications
ranging from face detection, verification, and recognition, object
detection and recognition, handwritten character and digit recogni-
tion, text detection and categorization, speech and speaker verifica-
tion, recognition, information and image retrieval, prediction and so
on Widodo and Yang (2007a, 2007b), Yang, Zhang, and Zhu (2007),
Yuan and Chu (2007), Fei, Miao, and Liu (2009), Xiang, Zhou, An,
Peng, and Yang (2008), Widodo and Yang (2008), Camci and Chin-
nam (2008), Juang, Sun, and Chen (2009), Hotta (2008), Wu (2010),
Wu (2009, 2003), Wu and Law (2010), Li, Lord, Zhang, and Xie
0957-4174/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.eswa.2010.07.052
⇑
Tel.: +86 25 51166581; fax: +86 25 511665260.
E-mail addresses: wuqivr@hotmail.com, hmwuqi@polyu.edu.hk
Expert Systems with Applications 38 (2011) 1457–1463
Contents lists available at ScienceDirect
Expert Systems with Applications
journal homepage: www.elsevier.com/locate/eswa