Studies on Fuzzy C-Means Based on Ant Colony Algorithm
Wang Guicheng, Yin Xuejiao, Pang Yujun, Zhang Min, Zhao Wendan, Zhang Zhansheng
Shenyang Institute of Chemical Technology, Shenyang, 110142, China
wguic@sohu.com
Abstract—A fault identification with fuzzy C-Mean clustering
algorithm based on improved ant colony algorithm (ACA) is
presented to avoid local optimization in iterative process of
fuzzy C-Mean (FCM) clustering algorithm and the difficulty in
fault classification. In the algorithm, the problem of fault
identification is translated to a constrained optimized
clustering problem. Using heuristic search of colony can find
good solutions. And according to the information content of
cluster center, it could merger surrounding data together to
cause clustering identification. The algorithm may identify
fuzzy clustering numbers and initial clustering center. It can
also prevent data classification from causing some errors.
Thus, applying in fault diagnosis shows validity of computing
and credibility of identification results.
Keywords- Ant Colony Algorithm; FCM; Clustering; Fault
Diagnosis; Fault Identification
I. INTRODUCTION
FCM has been extensively applied in many areas,
including automatic control, pattern recognition, and
decision analysis. However, FCM clustering algorithm must
define sample clustering numbers. In addition, whether the
result of samples clustering is reasonable or not belongs to a
clustering validity problem. So it can make local
optimization in iterative process of FCM clustering
algorithm. The ant colony algorithm is a new bionic
simulated evolutionary algorithm [1]. The main
characteristics are positive feedback, distributed
computation, and the use of a constructive greedy heuristic.
Positive feedback accounts for rapid discovery of good
solutions. Distributed computation avoids premature
convergence. And the greedy heuristic helps find acceptable
solutions in the stages of the search process. This paper
presents a new method of fault identification classified
which is based on AC-FCM clustering algorithm. This
method utilizes the strong ability of ant colony algorithm and
avoids the sensitivity to local optimization of the FCM
algorithm. Through comparing FCM with AC-FCM, it
illustrates this method valid.
II. A
NT COLONY ALGORITHM
The natural metaphor on which ant algorithms are based
is that of ant colonies. Real ants are capable of finding the
shortest path from a food source to their nest without using
visual cues by exploiting pheromone information. While
walking, ants deposit pheromone on the ground and follow,
in probability, pheromone previously deposited by other
ants.
Artificial ants simulate the behavior of real ants who find
shortest paths from feeding sources to back. We decide to
use the well-known traveling salesman problem (TSP) as an
example to illustrate the ant colony model. Given a set of n
towns, the TSP can be stated as the problem (TSP) as the
problem of finding a minimal length closed tour that visits
each town once. Let b
i
(t) (i = 1, . . . , n) be the number of
ants in town i at time t and let
∑
=
=
n
i
i
tbm
1
)(
be the total number
of ants [2]. Let
)(t
ij
τ
be the intensity of the pheromone trail
on edge (i, j) at time t. Generally speaking, we set the
intensity of trail to 0 at time 0. In fact, the choice is
influenced by the intensity of pheromone trails left by
preceding ants. A higher level of pheromone on one path
gives ants a stronger stimulus and thus a higher probability
to choose this path. We define tabu
k
(k=1,2,…,m) as the
dynamically growing vector which contains the tabu list of
the kth ant, and tabu
k
(s) as the sth element of the list (i.e., the
sth town visited by the kth ant in the current tour).
We define the transition probability from town
i to town
j
for the kth ant as
⎪
⎪
⎩
⎪
⎪
⎨
⎧
∈
⋅
⋅
=
∑
⊂
otherwise
allowedjif
tt
tt
tp
k
alloweds
isis
ikij
k
ij
k
,
,
0
)]([)]([
)]([)]([
)(
βα
βα
ητ
ητ
(1)
where allowed
k
={C-tabu
k
} is the set of cities that remain to
be visited by kth ant positioned on next city and
where
and
are parameters that control the relative
importance of trail versus visibility [3]. We call
visibility
(t)η
ij
=1/d
ij
, d
ij
is the Euclidean distance.
In ant colony algorithm, the global updating rule is
implemented as follows. Once all ants have built their tours,
pheromone is updated on all edges according to
)()()1()( ttnt
ijijij
Δ+⋅−=+
(2)
∑
=
Δ=Δ
m
k
k
ijij
tt
1
)()(
ττ
(3)
where
is a coefficient such that (1-
) represents the
evaporation of trail between time
t
and t+n. And
where
)(t
k
ij
τ
Δ
is the quantity per unit of length of trail
substance laid on edge
(i,j) by
the
kth ant between time t and
t+n. It is given by
⎪
⎩
⎪
⎨
⎧
+
=Δ
otherwise,
n)time t and (between t
its tour(i,j) in uses edge kth ant if
,
L
Q
(t)τ
k
k
ij
0
(4)
2010 International Conference on Measuring Technology and Mechatronics Automation
978-0-7695-3962-1/10 $26.00 © 2010 IEEE
DOI 10.1109/ICMTMA.2010.384
515