You are on page 1of 7

ARTICLE IN PRESS

Mechanical Systems
and
Signal Processing
Mechanical Systems and Signal Processing 22 (2008) 1917–1923
www.elsevier.com/locate/jnlabr/ymssp

Modeling and controller design of a shaking table


in an active structural control system
Yang Xua,b,, Hongxing Huab, Junwei Hanc
a
School of Mechanical Engineering, Donghua University, Shanghai 201620, PR China
b
State Key Laboratory of Mechanical System and Vibration, Shanghai Jiao Tong University, Shanghai 200240, PR China
c
School of Mechatronic Engineering, Harbin Institute of Technology, Harbin 150001, PR China
Received 17 June 2007; received in revised form 5 January 2008; accepted 11 February 2008
Available online 10 March 2008

Abstract

The purpose of this paper is to develop an approach to improve the tracking characteristics of a small-scale shaking
table, which is used in an active structural control system. Firstly, the active control experimental system is briefly
described. Then, a system dynamic model is constructed, which integrates hydraulic actuator system and a three-floor test
structure mounted on the shaking table. Due to low natural frequency and small damping, the small-scale shaking table
cannot meet the experimental requirements. Thus, three states control algorithm based on pole-assignment principle is
applied for extending the acceleration bandwidth and improving system damping. Lastly, the experimental results prove
that the new controller is effective.
r 2008 Elsevier Ltd. All rights reserved.

Keywords: Active structural control; Shaking table; Hydraulic actuator; Modeling; Three states control

1. Introduction

Active structural control is the most advanced technology for enhancing structural functionality and safety
against natural hazards such as strong winds and earthquakes. The active mass driver (AMD) benchmark
structural control problem is presented to provide an analytical test bed for the study of various active control
algorithms [1]. Shaking table is one of the most important equipments in an active structural control system.
Generally, a shaking table should have a certain acceleration response bandwidth and enough damping to
ensure reliability and repeatability. The bandwidth of the shaking table is affected not only by the actuators
characteristics itself but also by the test structure mounted on the table [2]. In our experimental system, which
is designed according to AMD benchmark problem, the small-scale shaking table is originally established for
another experiment and its highest bandwidth is only about 17 Hz. It cannot meet the requirements of the
active control experiment, which needs an acceleration bandwidth of 1–30 Hz. To ensure that the shaking table
can operate with the desired performance levels, three states control algorithm is developed to extend the

Corresponding author. Tel.: +86 021 3420 6573x231; fax: +86 021 3420 6006.
E-mail address: xuxuyangyang@yahoo.com.cn (Y. Xu).

0888-3270/$ - see front matter r 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ymssp.2008.02.003
ARTICLE IN PRESS
1918 Y. Xu et al. / Mechanical Systems and Signal Processing 22 (2008) 1917–1923

acceleration bandwidth and improve system damping without changing the original configuration of the
shaking table. Three states control algorithm is a new low-cost shaking table controller method. It includes
two parts: three states feedback control and three states feed-forward control. The three states feedback
control method can make the shaking table system obtain a relatively desired stable performance. The three
states feed-forward control method can further extend frequency bandwidth by canceling the poles that
determine system bandwidth.

2. Experimental setup

The active control experimental system is shown in Fig. 1. The assembly mainly consists of a shaking table,
a hydraulic actuator, a three-floor test structure with an AMD system, a signal modulator and sensors. The
size of the small-scale aluminum shaking table is 900  700  26 mm3 and its weight is 44.3 kg. The actuator
has an effective piston area of 294.5 mm2 and 723 mm dynamic stroke. Hydraulic flow of the actuator is
controlled by a symmetric servo-valve, which has a 100 Hz natural frequency. The three-floor test structure
mounted on the shaking table is 1580 mm in height and has a weight of 304.3 kg, distributed evenly among the
three floors. The first three modes of the test structure are 5.38, 15.68 and 22.53 Hz, respectively, which is
obtained by Finite Element software (ANSYS) and verified by the system identification experiment. The
controller used in this experiment is implemented on a digital computer.

Fig. 1.
ARTICLE IN PRESS
Y. Xu et al. / Mechanical Systems and Signal Processing 22 (2008) 1917–1923 1919

3. Dynamic model

Fig. 2 shows the schematic diagram of a shaking table attached test structure, where ps is supply oil pressure;
p0 is return oil pressure; yp is the horizontal displacement of piston; i is servo-valve input electric current;
A1 and A2 are effective piston areas; P1 and P2 are the pressures in the two chambers; Q1 is the flow into
chamber 1; Q2 is the flow out of the chamber 2; m is platform mass; and mj, kj and yj (j ¼ 1, 2, 3) are the mass,
stiffness and the relative displacement to ground of the jth floor of test structure, respectively. Because
structure damping is very small, it is ignored in this model for analysis convenience [3]. The following three
dynamics equations of the hydraulic actuator system outline the physical features of the shaking table:

(1) Valve flow equation


QL ¼ K q xv  K c PL (1)
3 3
where QL ¼ (Q1+Q2)/2 is load flow (m /s), Kq is flow gain ((m /s)/m), xv is valve spool position (m), Kc is
flow-pressure coefficient of servo-valve ((m3/s)/Pa) and PL ¼ P1P2 is the differential pressure between the
actuator chambers (Pa).
(2) Flow continuity equation of the actuator
dyp V t dPL
QL ¼ A þ C tc PL þ (2)
dt 4be dt
where Vt is total volume of the actuator two chambers (m3), be is effective bulk modulus (Pa) and Ctc is the
total leakage coefficient of actuator (m3/s)/Pa).
(3) Force balance equation
d 2 yp
APL ¼ m þ k1 ðyp  y1 Þ (3)
dt2

d 2 y1
k1 ðyp  y1 Þ  k2 ðy1  y2 Þ ¼ m1 (4)
dt2

d 2 y2
k2 ðy1  y2 Þ  k3 ðy2  y3 Þ ¼ m2 (5)
dt2

d 2 y3
k3 ðy2  y3 Þ ¼ m3 (6)
dt2

Fig. 2.
ARTICLE IN PRESS
1920 Y. Xu et al. / Mechanical Systems and Signal Processing 22 (2008) 1917–1923

Perform Laplace transformation to Eqs. (1)–(6) and defining the state vector as X_ ¼ [y_ p y€ p yp(3) yp(4) yp(5)
yp yp(7) yp(8) yp(9)]T, the state space equation and measurement equation are given by
(6)

2 3 2 3
0 1 0 0 0 0 0 0 0 0
60 0 1 0 0 0 0 0 0 7 607
6 7 6 7
6 7 6 7
60 0 0 1 0 0 0 0 0 7 607
6 7 6 7
60 0 0 0 1 0 0 0 7
0 7 607
6 6 7
6 7 6 7
_ 6 0 0 0 0 0 1 0 0 0 7 6 7
X ¼6 7X þ 6 0 7Q0 (7)
60 0 0 0 0 0 1 0 0 7 6 7
6 7 6 70
6 7 6 7
60 0 0 0 0 0 0 1 0 7 607
6 7 6 7
60 0 0 0 0 0 0 0 7
1 5 6 7
4 405
0 a1 a2 a3 a4 a5 a6 a7 a8 1
h i
yp ¼ b0 0 b1 0 b2 0 b3 0 0 X (8)

where Q0 ¼ Kqxv is no-load flow (m3/s). Factors in Eqs. (7) and (8) are directly listed as follows:

4beA2 k1 k2 k3
a1 ¼ (9)
V t mm1 m2 m3

4beK ce ðm þ m1 þ m2 þ m3 Þk1 k2 k3
a2 ¼ (10)
V t mm1 m2 m3
where Kce ¼ Kc+Ctc is the total flow-pressure coefficient (m3/s)/Pa).

ðm þ m1 þ m2 þ m3 Þk1 k2 k3 þ ðm1 k2 k3 þ m2 k1 k3 þ m3 k1 k2 þ m3 k1 k3 þ m2 k3 k2 þ m3 k2 k3 Þ
a3 ¼ (11)
V t mm1 m2 m3

ð4beK ce ðmm1 k2 k3 þ mm2 k1 k3 þ mm3 k1 k2 þ mm3 k1 k3 þ mm2 k3 k2 þ mm3 k2 k3


þm1 m2 k2 k3 þ m1 m3 k1 k2 þ m1 m3 k1 k3 þ m2 m3 k1 k2 ÞÞ
a4 ¼ (12)
V t mm1 m2 m3

ððmm1 m2 k3 þ mm2 k1 k3 þ mm3 k1 k2 þ mm3 k1 k3 þ mm2 k2 k3 þ mm3 k2 k3 þ m1 m2 k2 k3


þm1 m3 k1 k2 þ m1 m3 k1 k3 þ m2 m3 k1 k2 Þ þ ðm1 m2 k3 þ m1 m3 k2 þ m1 m3 k3 þ m2 m3 k1 þ m2 m3 k2 ÞÞ
a5 ¼
mm1 m2 m3
(13)

4beK ce ðmm1 m2 k3 þ mm1 m3 k2 þ mm1 m3 k3 þ mm2 m3 k1 þ mm2 m3 k2 þ m1 m2 m3 k1 Þ


a6 ¼ (14)
V t mm1 m2 m3

ðmm1 m2 k3 þ mm1 m3 k2 þ mm1 m3 k3 þ mm2 m3 k1 þ mm2 m3 k2 þ m1 m2 m3 k1 Þ þ m1 m2 m3


a7 ¼ (15)
mm1 m2 m3

4beK ce
a8 ¼ (16)
Vt

4beAk1 k2 k3
b0 ¼ (17)
V t mm1 m2 m3
ARTICLE IN PRESS
Y. Xu et al. / Mechanical Systems and Signal Processing 22 (2008) 1917–1923 1921

Fig. 3.

4beAðm1 k2 k3 þ m2 k1 k3 þ m3 k1 k2 þ m3 k1 k3 þ m2 k2 k3 þ m3 k2 k3 Þ
b1 ¼ (18)
V t mm1 m2 m3

4beAðm1 m2 k3 þ m1 m3 k2 þ m1 m3 k3 þ m2 m3 k1 þ m2 m3 k2 Þ
b2 ¼ (19)
V t mm1 m2 m3
4beA
b3 ¼ (20)
V tm
Considering servo-valve as a proportional element, let system gain K ¼ 0.1, measurement frequency response
of the shaking table with 1–30 Hz white noise input is shown in Fig. 3. It is noted that there are two modes about
5, 15 Hz and the system bandwidth is about 15 Hz. The value of system bandwidth is so low that the small-scale
shaking table with only the position feedback controller cannot meet bandwidth requirements. Thus, the three
states controller design method of the shaking table must be investigated to extend the system bandwidth.

4. Three states controller design

The three states control strategy is a controller method based on pole-assignment theory. A desired system
performance can firstly be obtained by designing three states feedback factors [4,5]. Plentiful experimental
results validate that system damping can be improved by adding acceleration feedback factor Kaf. System
bandwidth can be extended by increasing the velocity feedback factor Kvf while system damping will be
reduced at the same time. System gain can be increased by adjusting the displacement feedback factor Kdf.
Here, Kaf ¼ 0.08, Kvf ¼ 0 and Kdf ¼ 1.1. Zeros and poles distribution plot of actuator system with three states
feedback control is shown in Fig. 4. Following arrowhead orientation, system natural frequency improved but
damping did not.
However, only three states feedback control is not enough to reach the required bandwidth. If conjugated
poles pairs (s+99.6745.4i), which determine system bandwidth, is canceled (Fig. 4), system frequency
bandwidth will be further improved to 127 Hz in theory. Let that feed-forward polynomial T(s) equal the
following equation, system bandwidth can be extended and meet our experimental requirements.
TðsÞ ¼ ðs þ 99:6 þ 45:4iÞðs þ 99:6  45:4iÞ ¼ s2 þ 199:2s þ 11; 981 (21)

5. Experiment

To demonstrate the effectiveness of three states controller, experimental validation is performed on the
small-scale shaking table with the test structure. An exciting system with 1–40 Hz white noise input signal,
frequency response of acceleration bandwidth is obtained as shown in Fig. 5. Comparison between the output
ARTICLE IN PRESS
1922 Y. Xu et al. / Mechanical Systems and Signal Processing 22 (2008) 1917–1923

Fig. 4.

Fig. 5.

acceleration responses versus the input acceleration signal is shown in Fig. 6. From Figs. 5 and 6, it is noted
that the system acceleration bandwidth can already increase up to 30 Hz and the input signal has a good
matching with the output signal within 1–30 Hz bandwidth. The natural frequencies of the structural system
are about 5, 15 and 21 Hz, respectively. Coherence function coh(ejw) is an important criterion to evaluate
similarity of the two random signals. If the coherence function value is more than 0.8, signal repeatability is
usually thought as credible in engineering. Fig. 7 shows the coherence function of desired acceleration
response and actual actuator acceleration response. It is clear that the coherence function value is more than
0.8 and the shaking table tracking motion that the shaking table reproduces is feasible and reliable.

6. Conclusion

This paper presented a new low-cost shaking table controller design method to extend system acceleration
bandwidth. Experimental results show that the input signal and output signal has a good matching within
ARTICLE IN PRESS
Y. Xu et al. / Mechanical Systems and Signal Processing 22 (2008) 1917–1923 1923

Fig. 6.

Fig. 7.

1–30 Hz bandwidth, which meets our active structure experiment requirement. It proves that this new
controller based on the three states strategy is very effective. In addition, the first three modes of the test
structure coincided with experimental results, which proved that the constructed mathematic model is correct
and credible.

References

[1] B.F. Spencer, S.J. Dyke, H.S. Deoskar, Benchmark problems in structural control—part I: active mass driver system, Earthquake
Engineering and Structural Dynamics (1998) 1127–2247.
[2] J. Kuehn, D. Epp, W.N. Patten, High-fidelity control of a seismic shake table, Earthquake Engineering and Structural Dynamics
(1999) 1235–1254.
[3] H.R. Li, Controlling Electrohydraulic System, National Defence Industry Press, 1990.
[4] M. Tarokh, Linearity of the coefficients of the closed-loop characteristic polynomial, International Journal of Control (1987)
1383–1385.
[5] M. Tarokh, On output feedback stabilization and pole assignment, International Journal of Control (1980) 399–408.

You might also like