You are on page 1of 7

Mathematical Computation

December 2013, Volume 2, Issue 4, PP.109-115

Analysis the Characteristic of C1, C2 based on


the PSO of Iterative Shift and Trajectory of
Particle
Rongyong Ma, Leilei Yang, Zhichao Zhang #
Institute of civil engineering and architecture, Guangxi University, Nanning, Guangxi, china
#Email:

992704656@qq.com

Abstract
First of all, the fundamental model and operation mechanism of Particle Swarm Optimization(PSO) have been illustrated in this
paper; then the effect of the parameters c1 , c2 on particle behavior and the evolution of the algorithm has been examined through
the particle displacement with iterative, the path analysis and the parameter testing of function. Further, several identified
phenomena and the necessities of random selection in PSO have been discussed including the update space of the solution
constantly collapses, the particles "wandering" and "vibration", the evolution and biodiversity loss of particles. Ultimately, the
causes of premature and the local convergence have been explored. Through the mentioned above, the operation mechanism of the
model and the properties parameters c1 , c2 have been well learnt.
Keywords: Particle Swarm Optimization (PSO); Inertia Weight Factor; Learning Factor; Parameter Characteristics

C1C2

530004
PSO;

c1 , c2 ;
PSO
; PSO
c1 , c2

PSO Kennedy Eberhart 1995

2013ZDX04 51369005
- 109 www.ivypub.org/mc

PSO

1.1 PSO

n Xx1,x2,,xn
xi i i 1, 2,..., n
m j I j ( x j1, x j 2 ,..., x jn )
V j (V j1,V j 2 ,...V jn ) j 1, 2,..., m PSO

V j(k 1) V j(k ) c11 ( pb(jk ) I (jk ) ) c2 2 ( gb(k ) I (jk ) )

(3-1)

I (jk 1) I (jk ) V j(k 1)

(3-2)

V Vmax ,V Vmax V Vmax ,V Vmax

(3-3)

Shi Eberhart

3-1 V j( k )

Vmax Vmax
V j( k ) V j( k 1)
V j(k 1) V j(k ) c11 ( pb(jk ) I (jk ) ) c2 2 ( gb(k ) I (jk ) )

(3-4)

I (jk 1) I (jk ) V j(k 1)

(3-5)

c1 , c2

1, 2 0,1 1 (r11, r12 ,..., r1n ) , 2 (r21, r22 ,..., r2n )


(k )
)
V j(k ) ,V j(k 1) j k k+1 V j(k ) (V j(1k ) ,V j(2k ) ,...,V jn

( k 1)
V j(k 1) (V j(1k 1) ,V j(2k 1) ,...,V jn
)

I (jk ) , I (jk 1) j k k+1 I (jk ) ( x(jk1) , x(jk2) ,..., x(jnk ) )

I (jk 1) ( x(jk11) , x(jk21) ,..., x(jnk 1) )


pb(jk ) k j personal best position in history pbest
j j

gb(k ) k group best position in history gbest

Vmax

1.2 PSO
1
j V j( k 1) V j( k )
pbest c11 ( pb(jk ) I (jk ) )
gbest c2 2 ( gb(k ) I (jk ) )
V j(k 1) I (jk 1) I (jk ) I (jk ) I (jk 1)
V j( k 1) V j( k )

2PSO
- 110 www.ivypub.org/mc

(0) (0)
(0)
(0)
PSO I (0) I1(0) , I 2(0) ,..., I m
I (0)
j ( x j1 , x j 2 ,..., x jn )

(0)
(0)
j 1, 2,..., m V (0) (V1(0) ,V2(0) ,...,Vm(0) )T V j(0) (v(0)
j1 , v j 2 ,..., v jn ) j 1, 2,..., m

3
PSO
x y F(u)
F(x)>F(y) F(u) x x F(u) y
y pbest gbest
4
PSO
pbest gbest

pbest gbest
pbest gbest
GOSGlobal Optimal Solution gbest GOS gbest
gbest
pbest

c1 , c2

(k+1)

(k+1)

1)
0

(k+

1>

>0

gb(k)

)
g

(k+1)

(k+1)

'(k+1)

(k+1)

(k+1)

O(k+1

pb(k)

2>

1)
2

k+

2(

1)

k+

2 2

1(

g(k+

)
+1

1)

p (k

1
1

(k)

Particle current position

(k)

00

00(k+1),

(k+1)

'(k+1)
1'<

1,

2'<

Single particle's displacement in single-step iteration in PSO


(mainly in different c1,c2)
2

Origin

Two-dimensional space

1 c1 , c2
I (k ),V ( K ) I (k 1),V (k 1)
0 V0 (k 1) I0 (k 1) pbest gbest
c1 0, c2 0, 0 c2 0 c1 0 c1 c2 0

c1 ' c1, c2 ' c2 ,1,2 ,00 , '


1
- 111 www.ivypub.org/mc

1 c1 , c2
c1 c2 c1 0
gbest
2 PSO
300 0.5 c1 0.1 c2 2 gbest
gbest 1 Vg (k 1)

2 c1 0.1 , c2 2 PSO

c1 c2 c2 0
c2 0 m PSO
gbest
gbest 3 PSO
300 0.5 c1 2 c2 0.1 28
gbest pbest 1
V p (k 1)

3 c1 2 , c2 0.1 PSO

c1 0, c2 0
1 V00 (k 1) I 00 (k 1) PSO
- 112 www.ivypub.org/mc

c2
c1
(k+1)

(k+1)

1)
0

(k+

gb(k)

pb(k)

1>

>0

O(k+1)

2>

'(k+1)

(k+1)

'

(k+1)

1)

k+

2(

1(

k+
1)

'(k+1)

00

1'<

(k)

Particle current position (k)

1,

2'<

(k+1)

00
00(k+1),

< '; < 1<

; < 2<

00

00

Single particle's displacement orientation in single-step iteration in PSO


(mainly in different c1,c2)
2

Origin

Two-dimensional space

PSO

3.1. PSO
1
PSO pbest gbest GOS

2
PSO

3
PSO
PSO
5

gbest

GOS

GOS

5 PSO
- 113 www.ivypub.org/mc

PSO gbest
gbest
gbest

3.2.
PSO gbest
GOS
PSO
PSO

gbest

gbest

3.3.
1

c2

2
GOS GOS GOS
GOS



2 c1 , c2 c1
c2

3PSO

- 114 www.ivypub.org/mc

REFERENCES
[1]

Kennedy J, Eberhart R. Particle swarm optimization. In: Proceedings of the IEEE international conference on neural networks
(Australia), 1995, 1942-1948. ISBN 0-7803-2768-3/95

[2]

Shi Y, Eberhart R C. A modified particle swarm optimizer [A]. Proc of the IEEE World Congress on Computational Intelligence
[C]. 1998, 69-73

[3]

Charles Elegbede. Structural reliability assessment based on particles swarm optimization. STRUCTURAL SAFETY. 2005, (27):
171-186

1955-

1986

1989-

8 1988 12

Email: 992704656@qq.com

1992 8 1993 7
1994 8 1995 8
Email: mry57@gxu.edu.cn

- 115 www.ivypub.org/mc

You might also like