Received date: 2011−09−06; Accepted date: 2012−04−17

Corresponding author: CHEN Zhi-mei, Professor, PhD; Tel: +86−351−6998245; E-mail: zhimeichen400@163。com

J。 Cent。 South Univ。 (2012) 19: 2774−2781 2775

uncertainties  of crane  model  parameters,  a new   PSO-

xDl   x

sin x



Dx  x   1  f



sin x5   f

based fuzzy neural network sliding mode control

2 M   4 5

M   2 M   x M l

(FNNSMC)  method  was proposed。  The neural network

x3   x4

was adopted to approximate the uncertainties of  system,

xg cos x



x  (x  )2 Dx sin x5 x2 

4 5 3     6 M

and   the   PSO   was   used   to   optimize the parameters

sin2 x 1

sin x

sin2 x 1

  5 5 5

of sliding mode controller, so the convergence speed of parameters  is  fast  and  it  can  quickly reach  the sliding

(

M

x5   x6



)Dl x4 

m M

fx (   M



) fl m

surface and improve the system robustness。  Considering

xg sin x5  2x4 x6  cos x5  D  x 

6

the  friction  of  system,  it  is  needless  to approximately

x3 Mx3

decouple or linearize the model, and the controller can

sin x5 cos x5  D x  cos x5  f



sin x5 cos x5   f

(2)

accurately position the trolley as well as suppress the

Mx l   4 Mx x

Mx l

payload swing even in the presence of parameters uncertainties and external disturbance。

2Model of bridge crane

The model of the bridge crane system is shown in Fig。 1。 The trolley and the load can be regarded as   point

3 3 3

3Sliding mode control

Define the tracking error e(t) as

e1  xd  x xd x

e e3 ld  l ld  l 

(3)

mass, and the motion of load was always on the X−Y