Documente Academic
Documente Profesional
Documente Cultură
Actuator
Zhenyan Wang1,2, Zhen Zhang1, Jianqin Mao1, Kemin Zhou3
1.School of Automation Science and Electrical Engineering, Beijing University of Aeronautics and Astronautics, Beijing 100191,China
2. School of Electronical and Information Engineering, Taiyuan University of Science and Technology, Taiyuan, 030024, China
3. School of Electrical Engineering, Southwest Jiaotong University, Chengdu, 610031, China
E-mail: w9851@126.com
Abstract: Most smart materials used in engineering applications have rate-dependent hysteresis nonlinearity. In this
paper, a Hammerstein-based model is proposed to describe the dynamic characteristics of rate-dependent hysteresis in
piezoelectric actuator. A Bouc-Wen model is used to approximate the static nonlinear characteristic while a linear
dynamic model is constructed to capture the rate-dependent property of the hysteresis. Firstly, Bouc-Wen model
parameters are optimized with particle swarm optimization (PSO) algorithm to model the static hysteresis nonlinearity.
Based on this constructed static hysteresis nonlinear model, a recursive least squares (RLS) algorithm is utilized to
identify the dynamic linear model parameters of Hammerstein model according to the input-output data with rich
frequency information. Finally, the experimental results of applying the proposed method to the modeling of
rate-dependent hysteresis in a piezoelectric actuator are presented with a 100Hz sinusoidal scanning signal. The model
generation capability is verified in the given frequency range from 1Hz to100Hz when the excitation voltage are 40V,
80V, 120V, respectively.
Key Words: Hammerstein; Bouc-Wen; Rate-dependent; Piezoelectric actuator
INTRODUCTION
c
978-1-4577-2074-1/12/$26.00 2012
IEEE
MODEL DESCRIPTION
(1)
1392
(2)
x = dV H ()
(3)
PARAMETER
IDENTIFICATION
OF
HAMMERSTEIN
MODEL
FOR
RATE-DEPENDENT HYSTERESIS
(5)
(Y
exp
YBW i )2 / L
(6)
i =1
J RE =
(Y
exp
i =1
YBW i ) 2 / (Yexp i ) 2
(7)
i =1
(8)
f ( , , , d )
x = dV h
subject to
h = V V h h V h
(9)
1393
20
10Hz
40Hz
70Hz
100Hz
15
(10)
10
displacement(m)
A( z ) y (k ) = B( z ) x(k ) + (k )
A( z 1 ) = 1 + a1 z 1 + " + an z n
1
(11)
B ( z 1 ) = b0 + b1 z 1 + " + bm z m
(13)
k =1
S i k +1 k +1 S i
, i = 0,1,..., L 1
S i+1 = S i
T
1 + k +1 S i k +1
(14)
MODEL
VALIDATION
PIEZOELECTRIC ACTUATOR
IN
-20
20
40
60
voltage(V)
80
100
120
30
Experimental curve
Modeling curve
25
20
15
15
10
5
0
10
5
0
-5
-5
-10
-10
-15
Experimental curve
Modeling curve
25
20
displacement(m)
(12)
-15
displacement(m)
J = [ y (k ) T (k ) ]2
0
-5
-10
-1
y (k ) = T + (k )
0.5
1.5
2.5
time
3.5
4.5
-15
20
40
60
Voltage
80
100
120
(a)
(b)
Fig.5. Bouc-Wen modeling performance. (a) Displacement curve; (b)
Input/output curve.
g
20
15
Experimental curve
Modeling curve
Experimental curve
Modeling curve
10
Experimental curve
Modeling curve
15
-2
displacement(m)
displacement(m)
displacement(m)
10
-5
5
0
-5
-10
-4
-10
-15
(a)
(b)
Fig.3. Experimental setup. (a) Diagram; (b) Photograph;
1394
-6
10
15
20
25
voltage
30
35
40
45
-15
10
20
30
40
50
voltage
60
70
80
90
-20
20
(a)
(b)
Fig.6. Model performance. (a) 40V; (b) 80V; (c) 120V;
40
60
80
voltage
(c)
100
120
140
-2
-6
-4
10
15
20
25
voltage(V)
30
35
-6
40
Experimental curve
Model curve
10
10
displacement(m)
-5
-10
10
40
20
30
40
50
voltage(V)
60
70
80
10
20
(a) 10Hz
60
70
80
60
70
80
5
displacement(m)
displacement(m)
40
50
voltage(V)
Experimental curve
Model curve
10
30
(b) 20Hz
Experimental curve
Model curve
10
-5
-5
-10
-10
-5
10
20
30
40
50
voltage(V)
60
70
80
10
20
(c) 40Hz
0.3
0.4
0.5
time(s)
0.6
0.7
0.8
0.9
1
5
displacement(m)
-5
-5
-10
-10
10
20
30
40
50
voltage(V)
60
70
80
10
20
30
40
50
voltage(V)
60
70
80
(e) 80Hz
(f) 100Hz
Fig. 9 Model checking with 80V excitation voltage
20
20
Experimental curve
Model curve
15
displacement(m)
10
10
5
0
-5
-10
5
0
-5
-10
-15
-20
Experimental curve
Model curve
15
-15
20
40
60
voltage(V)
80
100
-20
120
20
(a) 10Hz
6
40
50
voltage(V)
Experimental curve
Model curve
10
displacement(m)
0.2
displacement(m)
0.1
30
(d) 60Hz
Experimental curve
Model curve
10
40
60
voltage(V)
80
100
120
80
100
120
(b) 20Hz
Experimental curve
Model curve
20
Experimental curve
Model curve
20
Experimental curve
Model curve
15
-2
10
displacement(m)
-2
displacement(m)
displacement(m)
5
0
-5
-10
-4
Experimental curve
Model curve
15
10
displacement(m)
35
-10
0
-15
5
0
-5
-10
-4
-15
0
10
15
20
25
voltage(V)
30
35
-6
40
(a) 10Hz
10
15
20
25
voltage(V)
30
35
40
-20
-15
20
40
(b) 20Hz
60
voltage(V)
80
100
-20
120
Experimental curve
Model curve
15
10
-2
displacement(m)
5
0
-5
-10
-4
60
voltage(V)
displacement(m)
displacement(m)
-2
40
20
Experimental curve
Model curve
15
10
20
(d) 60Hz
20
Experimental curve
Model curve
(c) 40Hz
Experimental curve
Model curve
4
displacement(m)
30
-5
5
0
-5
-10
-4
-15
-6
20
25
voltage(V)
-10
-6
15
Experimental curve
Model curve
10
Experimental curve
Model curve
Error curve
10
-2
(e) 80Hz
(f) 100Hz
Fig. 8 Model checking with 40V excitation voltage
displacement(m)
displacement(m)
-4
15
Experimental curve
Model curve
displacement(m)
G=
6
Experimental curve
Model curve
displacement(m)
10
15
20
25
voltage(V)
30
(c) 40Hz
35
40
-6
10
15
20
25
voltage(V)
(d) 60Hz
30
35
40
-20
-15
20
40
60
voltage(V)
80
100
120
-20
20
40
60
voltage(V)
80
100
120
(e) 80Hz
(f) 100Hz
Fig. 10 Model checking with 120V excitation voltage
1395
80V
120V
f (HZ)
10
20
40
60
80
100
RMSE
0.3263
0.3500
0.3329
0.3157
0.3035
0.3237
RE
f (HZ)
RMSE
RE
f (HZ)
RMESE
0.1019
0.1132
0.1090
0.1065
0.1033
0.1129
10
20
40
60
80
100
0.4237
0.4234
0.4533
0.4174
0.4803
0.5778
0.0559
0.0568
0.0623
0.0589
0.0687
0.0842
10
20
40
60
80
100
0.7290
0.6499
0.6326
0.7404
0.9354
1.3115
RE
0.0585
0.0529
0.0534
0.0635
0.0830
0.1200
[12]
[13]
Conclusions
REFERENCES
[14]
[15]
[16]
[17]
[18]
[19]
1396
[11]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]