Download as pdf or txt
Download as pdf or txt
You are on page 1of 24

Accepted manuscript to appear in IJSSD

Accepted Manuscript
International Journal of Structural Stability and Dynamics
by NATIONAL UNIVERSITY OF SINGAPORE on 06/01/20. Re-use and distribution is strictly not permitted, except for Open Access articles.

Article Title: Optimal vibration control of membrane structures with in-plane


polyvinylidene fluoride actuators

Author(s): Yifan Lu, Qi Shao, Fei Yang, Honghao Yue, Rongqiang Liu

DOI: 10.1142/S0219455420500959

Received: 10 February 2020


Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com

Accepted: 23 May 2020

To be cited as: Yifan Lu et al., Optimal vibration control of membrane structures with
in-plane polyvinylidene fluoride actuators, International Journal of Struc-
tural Stability and Dynamics, doi: 10.1142/S0219455420500959

Link to final version: https://doi.org/10.1142/S0219455420500959

This is an unedited version of the accepted manuscript scheduled for publication. It has been uploaded
in advance for the benefit of our customers. The manuscript will be copyedited, typeset and proofread
before it is released in the final form. As a result, the published copy may differ from the unedited
version. Readers should obtain the final version from the above link when it is published. The authors
are responsible for the content of this Accepted Article.
T
Accepted manuscript to appear in IJSSD

IP
Optimal vibration control of membrane structures with in-plane
polyvinylidene fluoride actuators

CR
Yifan Lu*, Qi Shao, Fei Yang, Honghao Yue*, Rongqiang Liu
State Key Laboratory of Robotics and System, Harbin Institute of Technology, Harbin 150001, China
by NATIONAL UNIVERSITY OF SINGAPORE on 06/01/20. Re-use and distribution is strictly not permitted, except for Open Access articles.

Correspondence: Yifan Lu: yf.lu@hit.edu.cn; Honghao Yue: block@hit.edu.cn

US
Abstract
Different kinds of membrane structures have been proposed for future space exploration and earth
observation. However, due to the low stiffness, high flexibility, and low damping properties, membrane
structures are likely to generate large-amplitude (compared to the thickness) vibrations, which may lead
to the degradation of their working performance. In this work, the governing equations are established at

AN
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com

first, taking into account the modal control force induced by the polyvinylidene fluoride (PVDF)
actuator.The optimal vibration control of the membrane structure is explored subsequently. A square
PVDF actuator is attached on the membrane to achieve the vibration suppression. The influence
ofactuator position and control gains on the vibration control performance are studied. The optimal
criteria for actuator placement and energy allocation are developed. Several case studies are numerically
simulated to demonstrate the validity of the proposed optimization criteria. The analytical results in this
DM

study can serve as guidelines for optimal vibration control of membrane structures. Additionally, the
proposed optimization criteria can be also applied to active control of different flexible structures.

Keywords:Optimal control; polyvinylidene fluoride; membrane structure; direct velocity feedback.

1 Introduction
Due to their merits of lightweight, low manufacturing cost, and high flexibility, membrane structures
TE

can well satisfy the demands of the large scale of future space structures1, 2. In recent years, gossamer
structures have become one of the hot spots in space science research. However, because of the low
stiffness and low damping of the membrane structure, once excited by external loads, the vibration of the
membrane is difficult to attenuate, which may lead to the deterioration of the system working
performance. Consequently, it is imperative to analyze the dynamics of the membranes and to achieve
EP

active control of their vibration.


Sorts of literature have worked on modeling the vibration of the membrane. Zheng et al. obtained the
free vibration frequencies of a rectangular orthotropic membrane by means of the power series formula3.
Liu et al. solved the frequencies with the LindstedtPoincaré Perturbation Method (L-P perturbation) and
the homology perturbation method (HPM) successively, and their accuracies are compared 4, 5. The
C

assumed mode method and the finite element method (FEM) were employed to solve the vibration
frequencies of the membrane by Liu et al., which turned out that the former results were constant, while
the latter results were time-varied6. The fundamental frequency of a prestressed orthotropic membrane
AC

1
T
Accepted manuscript to appear in IJSSD

IP
was acquired by Ahmadi et al. using the variational iteration method (VIM), with the effects of various
parameters discussed7. Also, Liu and Zheng et al. solved the forced vibration frequenciesbased on the
Krylov-Bogolubov-Mitropolsky (KBM) perturbation method and L-P perturbation, respectively8, 9.
Moreover, the effects of the piezoelectric actuators and sensors attached to the membrane on its

CR
frequencies were investigated by Gajbhiye et al.10.
Piezoelectric materialsincluding the lead zirconate titanate (PZT)11, the macro-fiber composites
(MFC)12, 13, and the polyvinylidene fluoride (PVDF)14 have been widely used for vibration control of
by NATIONAL UNIVERSITY OF SINGAPORE on 06/01/20. Re-use and distribution is strictly not permitted, except for Open Access articles.

various kinds of structures.However, compared with MFC and PZT,PVDF is much thinner and can
therefore be easily segmented intovarious shapes to adapt itself to themembrane structures.Different

US
control algorithms such as direct velocity feedback (DVF), positive position feedback (PPF), linear
quadratic regulator (LQR), and H∞have been used to achieve the vibration suppression of the membrane
experimentally with polyvinylidene fluoride actuators15-19. However, the sizes of the aforementioned
membranes were relatively small,no more than 0.05m2. Aiming at controlling the vibration of
membranes with a relatively larger size, more researches have been undertaken. A web-cable

AN
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com

configuration was firstly proposed by Sakamoto et al. to isolate a 10m×10m membrane from
vibration.The tension cables are designed to actuate both in-plane and out-of-plane with linear quadratic
controllers adopted20. The cable controller was studied by Liu et al. through a simple
single-degree-of-freedom system to attenuate the vibration of a 23.4m×5.6m membrane, with optimal
placement of cable actuators explored21. In the meanwhile, a series of researches have been carried out by
Liu et al. to investigate the membrane vibration control based on a model reference adaptive controller
DM

(MRAC). The placement of piezoelectric actuators wasoptimized under the guidance of particle swarm
optimizer (PSO) and genetic algorithm (GA)22, 23. The vibration control performance of distributed
in-plane and boundary actuators on large space membranes was studied by Shi et al. based on the
iterative membrane property modeling approach which proved that the combined actuator configuration
can provide the best control efficiency24.Additionally, wave-based methods have been verified effective
to control the vibration, which holds back wave reflections or the formation of standing waves 25,
26
.Nevertheless, althoughlots of researches have been done on optimal vibration control of membrane
TE

structures, the majority of them are focusing on the layout optimization of the actuators. In most existing

literatures,the optimal placement of the actuators are obtained using the genetic algorithm (GA) based

on energy assessment and other different optimal criteria, including maximizing the controllability
Grammian or other objective functions27-30. However, very few studies have considered the issue from
the perspective of modal actuation capability of the actuator. Moreover, for multi-mode vibration control,
EP

even when the actuator has been arranged on the optimal position, the allocation of the control energy on
each order mode will still affect the control performance, which has rarely been discussed.
In this paper,the optimal vibration control of a 1m×1m rectangular membrane are investigated. The
governing equation of the membrane is obtained by means of Galerkin decomposition method, in
C

conjunction with the actuating behavior of the piezoelectric actuator.The optimal vibration control
strategy for single-mode and multi-mode vibration of the membrane is then proposed and numerical
simulated. For multi-mode vibration, the criteria of the actuator layout optimization and the energy
AC

allocation optimization are set up, based on the degree of the participation and the dynamic
2
T
Accepted manuscript to appear in IJSSD

IP
characteristics of each mode in vibration, respectively.Analytical results indicate that the vibration of the
membrane can be effectively suppressed with this proposed optimal control strategy.
2 Modeling and control of the vibration
A rectangular membrane model with thickness h is studied in this paper, where a and b denote the

CR
length and width of the membrane, as shown in Figure 1. The structure is modeled as an isotropic
membrane attached with a PVDF actuator patch, with uniformly distributed pretensions N0x and N0y in
the directions of x and y, respectively. Since the PVDF is ultrathin(usually 10 to 100μm), lightweight, and
by NATIONAL UNIVERSITY OF SINGAPORE on 06/01/20. Re-use and distribution is strictly not permitted, except for Open Access articles.

covers only a small region of the membrane, its influence on the physical property of the membrane
substrate is neglected for simplicity.

US
y
N0y

AN PVDF
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com

N0x N0x

x
O a
N0y
Figure 1 Model of a rectangular membrane structure with pretension.
DM

2.1 Governing equations and boundary conditions


The transverse displacement wof the membrane is considered as a function of x, y, and t, which is
independent of coordinatez. The transverse normal strain is therefore neglected. Since the stress
components τyz, τxz, and σz are much smaller than the other three components, the deformations they
caused are also ignored consequently, which means the transverse shear deformation would be neglected
as well.According to these assumptions, the strain-displacement relationship of the membrane can be
written as31
TE

2
u 1  w  v 1  w  v u w w
2

x     , y     ,  xy    , (1)
x 2  x  y 2  y  x y x y

where u, v, and w represent the displacement in x, y, and z (transverse) directions, respectively; εx and εy
are strains in x and y directions, and γxy is the shear strain. Assuming plane stress, the constitutive
equations of the membrane can be expressed as31
EP

x 
E
1 2
 x   y  ,  y  1 E 2  y   x  ,  xy  2 1E    xy , (2)

where σx and σy are the principal stresses and τxy is the shear stress; E is Young’s modulus; μ is the Poisson
ratio. Then the in-plane forces per unit length are defined as
C

h/ 2 h/ 2 h/ 2
Nx  
h/ 2
 x d z, Ny  
h/ 2
 y d z, N xy  
h/ 2
xy dz, (3)

where Nx and Ny are in-plane principal forces per unit length in x and y directions; Nxy is the in-plane
AC

3
T
Accepted manuscript to appear in IJSSD

IP
shear force per unit length and h is the thickness of the membrane.
The dynamic equation of transverse motion and the compatibility equation are given by32
2 w 2 w 2 w 2 w
h
t 2
  N x  N 0 x 
x 2
  N y  N 0 y  y 2
 2  N xy  N 0 xy  xy
 p  x, y , t  , (4)

CR
1  2 N x   N y   2 N x 1  N y 2 1+   N xy   2 w   2 w  2 w
2 2 2 2

       , (5)
Eh y 2 Eh y 2 Eh x 2 Eh x 2 Eh xy  xy  x 2 y 2

where ρ is the volume density of the membrane; p(x, y, t) is the external load. By introducing the stress
by NATIONAL UNIVERSITY OF SINGAPORE on 06/01/20. Re-use and distribution is strictly not permitted, except for Open Access articles.

function φ(x, y, t), the in-plane stresses can be expressed as32


 2  2  2

US
x  ,   ,    . (6)
y 2 x 2 xy
y xy

Substituting equation (6) into equations (4) and (5) yields

2 w  2 w 2 w    2 w  2  2  2 w  2 w  2 
h   N +N   h    2   p  x, y , t  , (7)
t 2  x 2 y 2   x y x 2 y 2 xy xy 
0x 0y 2 2

AN
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com

2
1   4  4  4    2 w   2 w  2 w
  +2    . (8)
E  x 4 y 4 x 2 y 2   xy  x 2 y 2

The corresponding boundary conditions can be expressed as follows:


when x=0 or x=a,
w
w  0,  0, N y  0, N xy  0 (9)
t
DM

when y=0 or y=b,


w
w  0,  0, N x  0, N xy  0 (10)
t
2.2 Discretization of the dynamic equations
Based on the modal expansion method, the transverse displacement w(x, y, t) which satisfies the
boundary conditions can be written as

m x n y
w  x, y , t    sin sin q t  , (11)
TE

m , n 1 a b

where q(t) is the modal participating factor. In order to solve the stress function, for single mode
vibration, taking one term from equation (11)and inserting it into equation (8) yields
1   4  4  4  m 2 n 2 4  2m 2n  2
 4  4 +2 2 2   cos x  cos y  q (t ) . (12)
E  x y x y  2a 2 b 2  a b 
EP

According to equation (12), one can assume that the stress function φ takes the following form:
 2m 2n 
   k1 cos x  k2 cos y  k3 x3  k4 x 2  k5 y 3  k6 y 2  k7 x 2 y  k8 xy 2  k9 xy  q 2 (t ) . (13)
 a b 
By substituting equation (13) into equation (12) one can obtain
C

16m4 4 2m 16n 4 4 2n m2 n 2 4  2m 2n 


k1 cos x k2 cos y 2 2 
cos x  cos y . (14)
Ea 4
a Eb 4
b 2a b  a b 
Comparing the coefficients of the trigonometric terms on each side of equation (14), it can be found
AC

4
T
Accepted manuscript to appear in IJSSD

IP
that
Ea 2 n2 Eb2 m2
k1  2 2
, k2  . (15)
32b m 32a 2 n2
Then substituting equation (13) into boundary condition equations(9) and(10), the other coefficients

CR
can be determined as
En2 2 Em2 2
k3  k5  k7  k8  k9  0, k4  2
, k6  . (16)
16b 16a 2
by NATIONAL UNIVERSITY OF SINGAPORE on 06/01/20. Re-use and distribution is strictly not permitted, except for Open Access articles.

Finally, the stress function can be expressed as


 E  a 2 n2 2m x b2 m2 2n y  E 2  n 2 2 m2 2   2
  x, y , t     2 2 cos  2 2 cos   2 x  2 y   q (t ) . (17)

US
 32  b m a a n b  16 b a 
Inserting equations (11) and (17) into the transverse dynamic equation (7) gives
 m2 n2  Eh 4  m4 2 n y n4 m x  3
 hWq   2  N0 x 2
 N 0y 2  Wq   4
sin  4
sin 2 Wq  p , (18)
 a b  4 a b b a 

m x n y
where W  sin sin
AN
. Applying the Galerkin method (i.e., multiplying by W and integrating
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com

a b
over the membrane surface), the discretized equation takes the following form:
d 2 q(t )
 K1q(t )  K3 q3 (t )  P , (19)
dt 2
where K1 and K3 represent the linear and nonlinear components of the membrane frequency,
respectively. P represents the modal excitation force, which can be written as
DM

 2  m2 n2 
K1   2 0x
N  N0 y  ,
h  a b 2

3 E 4  m4 n4 
K3   4  4 , (20)
16  a b 
4
 hab S
P pW dS .

2.3 Actuating characteristics of PVDF actuators


The vibration suppression of the membrane is realized by the attached PVDF actuators. The governing
TE

equation of the membrane considering the control effects can be expressed as33
2w  2w 2w    2 w  2  2  2 w  2 w  2 
h   N 0 x 2 +N 0 y 2   h  2  2 2 
t 2
 x y   x y
2
x y 2
xy xy 
, (21)
  2 M xa  2 M ya 
p  x, y , t    + 
 x 2 y 2 

EP

a
where M xa and M y are bending moments of the actuator in x and y directions, which are given by

M xa  r a d31Yp a  x, y, t  , M ya  r a d32Yp a  x, y, t  , (22)

where ra = h+ha is the bending arm and 


a
 x, y , t  is the imposed control voltage; d31 and d32 are the
C

piezoelectric strain constants; Yp is Young’s modulus of the PVDF actuator. Assume that the actuator

covers a rectangular area from (x1, y1) to (x2, y2), (see Figure 4), then   x, y, t  is given by
a
AC

5
T
Accepted manuscript to appear in IJSSD

IP
 a  x, y, t  = a  t  us  x  x1   us  x  x2   us  y  y1   u s  y  y2   , (23)

where us(∙) is the unit step function: us(x−xi) = 1 when x > xi, and us(x−xi)=0 when x < xi. ϕa(t) is the
time-varying control voltage imposed on the actuator. Assume that the intrinsic modal damping ratio of

CR
the membrane is zero, then the modal control equation takes the following form:
q  K1q  K 3 q 3  P  F c , (24)

where Fc is the modal control force for a specific mode which is expressed as
  2 M xa  2 M ya 
by NATIONAL UNIVERSITY OF SINGAPORE on 06/01/20. Re-use and distribution is strictly not permitted, except for Open Access articles.

4
 hab S  x 2
c
Fmn   +  Wmn dS
y 2 
4 r a Y p a  t 

US

 hab 
b

0 0
a
d 31
   x  x1      x  x2   us  y  y1   us  y  y2   +
  
, (25)
d32    y  y1      y  y2   us  x  x1   us  x  x2   sin
m x
a
sin
n y
b
dxdy

4r aYp a  t    m  2  n  
2
m x n y e
     S e sin
 hab  31  a  32 
d d sin dS
 b   a b

AN
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com

where δ(∙) is the Dirac delta function. δ(x−xi ) = ∞ when x = xi, and δ(x−xi )=0 when x ≠ xi .
Furthermore, for an arbitrary actuator patch covering the membrane surface from (x1, y1) to (x2, y2),
the induced modal control force can be expressed as

Fmnc t   Amn a t  , (26)

where Amn, the modal actuation factor, is defined as34


DM

4Yp r a   m 2  n   y2 x2
2
m x n y
Amn  d31    d 32    y1 x1 sin sin dxdy . (27)
 hab   a   b   a b

Obviously, the value of Amn denotes the modal control capability of an actuator on a specific mode (m,
n) at a given control voltage.
For free vibration of the membrane, P equals to zero in equation(24). The real-time vibration control
is accomplished based on the direct velocity feedback (DVF) control algorithm in this paper. But it is
noteworthy that although only the DVF is used here, the results of this work are also applicable for
TE

other control methods. The control gain Gc, which is given by the control system, is defined as
 a t   Gc q t  . (28)

By inserting equation (28) in equation(26), the modal control force is transformed into
Fmnc t    AmnGc q t  . (29)
EP

Then substituting equation (29) into equation (24) yields


q  Amn G c q  K1q  K 3 q 3  0 . (30)
It is obvious that the effect of the modal control force corresponds to the increase of damping. As an
indication of the control effect, the equivalent damping ratio ζmn induced by the actuator is defined as
Amn G c
C

 mn  , (31)
2mn
where ωmn is the natural frequency of mode (m, n), and the value of ζmn can reflect the control effect of
AC

the actuator on mode (m, n).


6
T
Accepted manuscript to appear in IJSSD

IP
3Optimal vibration control: actuator placement
In this section, based on the actuating characteristics of the PVDF actuator, the criterion of optimal
actuating position for single-mode vibration is established. An average modal actuation factor is then
proposed to serve as the criterion of the placement optimization of the actuator for multi-mode vibration

CR
control. Finally, numerical simulations for single-mode and multi-mode vibration control of a 1m×1m
rectangular membrane with a piece of PVDF actuator located on different positions are carried out to
demonstrate the control efficiency.
by NATIONAL UNIVERSITY OF SINGAPORE on 06/01/20. Re-use and distribution is strictly not permitted, except for Open Access articles.

3.1Optimization criterion
In order to explore the influence of the actuator position on the control effect, the membrane model is

US
divided into twenty-five identical actuating areas, as shown in Figure 2. The properties of the PVDF
actuator and the membrane are presented in Table 1. By substituting position coordinates x1, x2, y1, y2 into
equation (27), the modal actuation factor Amn of each actuating unit of different modes can be obtained, as
listed in Table 2. More intuitively, the modal actuation effectiveness distributions on the membrane are
also illustrated in Figure 3.

AN
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com

Table 1 Geometric and material properties of the membrane and PVDF actuator
Properties PVDF Membrane
2 9
Young's modulus (N/m ) Yp=2×10 Y=3.5×109
Poisson ratio μ=0.4
Density (kg/m ) 3
ρ=1400
−3
Thickness (m) a
h =0.1×10 h=0.25×10−3
DM

Length and width (m) a=1, b=1


−12
Piezoelectric constants (C/N) d31=d32=21×10
Initial stress (KPa) σ0x=σ0y=50

b
1 2 3 4 5

6 7 8 9 10
TE

11 12 13 14 15
y2
16 17 18 19 20
y1
21 22 23 24 25
x
O
EP

x1 x2 a

Figure 2 The segmented membrane model with a polyvinylidene fluoride actuator.


C
AC

7
T
Accepted manuscript to appear in IJSSD

IP
CR
by NATIONAL UNIVERSITY OF SINGAPORE on 06/01/20. Re-use and distribution is strictly not permitted, except for Open Access articles.

US
AN
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com

Figure 3 The modal actuation effectiveness distribution on the membrane.


Table 2 The modal actuation factor Amn of an actuator at different locations.

Amn Mode Mode


DM

(×10−3) (1,1) (1,2) (2,1) (2,2) (1,1) (1,2) (2,1) (2,2)


#1 0.012 −0.055 0.055 −0.160 #14 0.104 0 −0.290 0
#2 0.032 −0.145 0.090 −0.260 #15 0.040 0 −0.179 0
#3 0.040 −0.179 0 0 #16 0.032 0.090 0.145 0.260
#4 0.032 −0.145 −0.090 0.260 #17 0.084 0.235 0.235 0.420
#5 0.012 −0.055 −0.055 0.160 #18 0.104 0.290 0 0
#6 0.032 −0.090 0.145 −0.260 #19 0.084 0.235 −0.235 −0.420
#7 0.084 −0.235 0.235 −0.420 #20 0.032 0.090 −0.145 −0.260
#8 0.104 −0.290 0 0 #21 0.012 0.055 0.055 0.160
TE

#9 0.084 −0.235 −0.235 0.420 #22 0.032 0.145 0.090 0.260


#10 0.032 −0.090 −0.145 0.260 #23 0.040 0.179 0 0
#11 0.040 0 0.179 0 #24 0.032 0.145 −0.090 −0.260
#12 0.104 0 0.290 0 #25 0.012 0.055 −0.055 −0.160
#13 0.128 0 0 0
EP

From Figure 3 one can find that for different modes at different positions, the modal actuation factor
is different. However, the distribution of the magnitude of Amn is similar to the mode shape of the
membrane. This indicates that the actuator placed on the maximum amplitude of the membrane can
provide the best control effect; while the actuator placed on the nodal line of the membrane will have
no control effect.It is obvious that the actuator should be placed at the optimal position so as to attenuate
C

the vibration faster with the same control voltage, namely, to achieve a higher equivalent damping ratio
ζmn. According to equation(31), at a given control gain Gc, the larger the value of Amn is, the greater the
AC

equivalent damping ratio will be, which leads to a better control effect.It should be noted from Table
8
T
Accepted manuscript to appear in IJSSD

IP
3that at some positions, the value of Amn is negative, which means that the vibration will be amplified but
not suppressed if the input control gain is positive. However, since the sign of Gccan be easily changed
through a logic circuit by the controller, one should only need to concern the absolute value of Amn.
3.2Numerical simulations of single-mode vibration control

CR
For single-mode vibration control, the optimal actuator placement is easy to be determined. One may
just need to find the maximum absolute value of Amn of a specific mode from Table 2, the corresponding
number of the actuator will be the optimal one. In this section, the first four modes of the membrane are
by NATIONAL UNIVERSITY OF SINGAPORE on 06/01/20. Re-use and distribution is strictly not permitted, except for Open Access articles.

considered independently. According to Table 2, the optimal actuator for mode (1,1) is #13; the optimal
actuator for mode (1,2) is #18 (or #8); the optimal actuator for mode (2,1) is #12 (or #14); and the optimal

US
actuator for mode (2,2) is #9 (or #7, #17, #19).The input control voltage of the actuator is limited within
±500V to ensure that the PVDF can work safely and reliably.The control gain Gc is chosen appropriately
to guarantee the voltage limitation. For the four modes, Gc is set to be 18500, 10000, 10000, and 7600,
respectively. For each mode, the initial vibration amplitude A0is set to be 0.001m. The time responses of
different modes of the membrane with optimal and non-optimal actuator control are plotted in Figure

AN
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com

4.The dot line denotes the free vibration of the membrane without control effect (the intrinsic modal
damping ratio is assumed zero). The input control voltages are also presented to ensure that the voltage
imposed on the actuator is not beyond 500V. The non-optimal actuators employed in these cases are #7,
#22, #16, and #5, respectively. It can be seen from Figure 4 that for each mode, the optimally positioned
actuator can provide a better control performance than the non-optimal one. The equivalent damping
ratio induced by the actuator is collected in Table 3. One can find that with the same voltage limitation,
DM

ζmn of the optimally positioned actuator is apparently larger than that of the non-optimal one andthe
maximumζmn decreases with the increase of the mode order. For mode (1,2) and mode (2,1), the optimal
ζmn is identical, which is due to the symmetry of the mode shape.
TE
C EP
AC

9
T
Accepted manuscript to appear in IJSSD

IP
CR
by NATIONAL UNIVERSITY OF SINGAPORE on 06/01/20. Re-use and distribution is strictly not permitted, except for Open Access articles.

US
Figure 4Single-mode vibration control effects with optimal and non-optimal actuators.
Table 3 The equivalent damping ratio of single-mode vibration with different actuator positions.

AN
Optimal position Non-optimal position
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com

Mode
Position ζmn Position ζmn
(1,1) #13 4.24% #7 2.77%
(1,2) #18 2.98% #22 1.49%
(2,1) #12 2.98% #16 1.49%
(2,2) #9 2.54% #5 0.97%
DM

3.3Numerical simulations of multi-mode vibration control


In the above sub-section it has proved that for single-mode vibration control, the optimal actuator
placement can be realized according to the modal actuation factor Amn of each mode. However, in
practical applications, there may be multiple modes that participate in the vibration. In this situation,
the optimal placement of the actuator will be different. For example, if the vibration of the membrane
includes both the first order and the second order, although actuator #13 is the optimal actuator of mode
(1,1), it has no control effect on mode (1,2) or mode (2,1). Therefore, #13 is obviously not the optimal
TE

actuator for this case since the total vibration cannot be suppressed completely due to the second mode.
In this spirit, a new indicator named the average modal actuation factor is proposed, which is used to
determine the optimal position of the actuator for multi-mode vibration. Being different from the
single-mode modal actuation factor Amn, the average modal actuation factor A is defined as

A Amn
EP

0 mn

A= m, n 1 , (32)
A
m , n 1
0 mn

where A0mn denotes the initial vibration amplitude of mode (m,n).


C
AC

10
T
Accepted manuscript to appear in IJSSD

IP
Table 4 The average modal actuation factor A of an actuator at different locations.

A (×10−3) A (×10−3)
#1 0.071 #14 0.099
#2 0.132 #15 0.055

CR
#3 0.055 #16 0.132
#4 0.132 #17 0.243
#5 0.071 #18 0.099
by NATIONAL UNIVERSITY OF SINGAPORE on 06/01/20. Re-use and distribution is strictly not permitted, except for Open Access articles.

#6 0.132 #19 0.243


#7 0.243 #20 0.132

US
#8 0.099 #21 0.071
#9 0.243 #22 0.132
#10 0.132 #23 0.055
#11 0.055 #24 0.132
#12 0.099 #25 0.071
#13 0.032

AN
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com

Equation (32) indicates that when judging the optimal placement of the actuator for a multi-mode
vibration, the degree of participation of each mode in the total vibration should be considered, as well as
the modal actuation factor of each mode.In this case, the initial vibration amplitudes of all modes are
defined as A0 = [0.0006, 0.0006, 0.0006, 0.0006] m, which means that the first four modes are
participating in the vibration and the degrees of participation of all modes are the same. One may note
DM

that in the single-mode case, the generalized coordinate qmnis directly used to reflect the vibration of the
membrane. However, for multi-mode vibration, an average displacement w is further defined as
1
w  qmn t Wmn  x, y  dxdy ,
ab S m, n 1
(33)

where a and b are the length and width of the membrane, which is used to mirror the overall vibration of
the membrane.The average modal actuation factor A of an actuator is obtained based on equation (32)
and is listed in Table 4. It can be found that the optimal position of the actuator for this case is #17 (or #7,
TE

#9, #19), where the value of A reaches a maximum 0.243×10−3.Note that Table 4 is only used for
selecting the optimal actuator, the ‘real’ modal actuation factor used in the control system should refer to
Table 2.The control gain of each mode should also be chosen carefully to ensure the total voltage
limitation. For the optimal actuator #17, the control gain vector is chosen as Gc=[12540, 4487, 4487,
2508]. Each entry of the vector denotes the control gain of each mode, respectively. The time responses
EP

of each mode and the total average displacement of the membrane are shown in Figure 7. It can be found
that all participating modes are controlled effectively and the total vibration of the membrane can be
suppressed within eight seconds. Also one can see clearly nonlinearity from Figure 5 that with the
attenuation of the vibration, the period of vibration of the membrane becomes larger.
Three other cases with non-optimal actuators are also numerically simulated. The non-optimal
C

actuators are chosen as #13, #12, and #16, respectively. The control gain vector for actuator #13 is Gc =
[31813, 0, 0, 0]; for actuator #12 is Gc = [21235, 0, 7597, 0]; for actuator #16 is Gc = [14761, 5281, 3264,
1825]. The reason why Gc is set like this will be explained in the next section. The time responses of each
AC

11
T
Accepted manuscript to appear in IJSSD

IP
mode and the total average displacement of the membrane are shown in Figure 6to Figure 8. The
equivalent damping ratio induced by the actuator of each case is collected in Table 5.
Table 5 The equivalent damping ratio of multi-mode vibration with different actuator positions.

ζmn Optimal position Non-optimal position

CR
Mode #17 #13 #12 #16
(1,1) 1.94% 7.50% 4.05% 0.87%
(1,2) 1.18% 0 0 0.53%
by NATIONAL UNIVERSITY OF SINGAPORE on 06/01/20. Re-use and distribution is strictly not permitted, except for Open Access articles.

(2,1) 1.18% 0 2.47% 0.53%


(2,2) 0.92% 0 0 0.42%

US
From Figure 6 one can find that when actuator #13 is employed, the first mode can be suppressed
rapidly, almost within one second. However, from the sub-figures one can also find that the input control
voltages for mode (1,2), (2,1) and (2,2) are always zero, which means that actuator #13 has no control
effect on the other modes. Therefore, total vibration control performance is quite poor. Table 6 also
indicates that when actuator #13 is used, although ζ11 is 7.50%, the equivalent damping ratios of other

AN
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com

modes are all zero. Similarly, from Figure 7 it can be observed that when actuator #12 is employed, only
mode (1,1) and (2,1) can be controlled; while mode (1,2) and (2,2) still exist after the vibration
suppression. Regarding actuator #16, Figure 8 shows that #16 is effective for all modes, but the overall
control performance is not as good as actuator #17. According to Table 5, also, one can find that the
induced equivalent damping ratio of each mode of actuator #16 is always less than that of actuator #17
(the optimal one). In summary, the control capability of the actuators on this combination of multi-mode
DM

vibration can be concluded as: #17 > #16 > #12 > #13. However, it should be noted that if the degree of
participation of each mode in the total vibration changes, the optimal position of the actuator may change
as well.The case studies have proved that the position of the actuator has a strong influence on the
vibration control performance. Therefore, to achieve an effective vibration control of the membrane
structure, the optimal placement of the actuator is of great significance.
TE
C EP
AC

12
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com
by NATIONAL UNIVERSITY OF SINGAPORE on 06/01/20. Re-use and distribution is strictly not permitted, except for Open Access articles.

AC
C EP
TE

13
DM
AN
Accepted manuscript to appear in IJSSD

US
Figure 5Multi-mode vibration control effects with optimal actuator #17.
CR
IP T
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com
by NATIONAL UNIVERSITY OF SINGAPORE on 06/01/20. Re-use and distribution is strictly not permitted, except for Open Access articles.

AC
C EP
TE

14
DM
AN
Accepted manuscript to appear in IJSSD

US
Figure 6 Multi-mode vibration control effects with non-optimal actuator #13.
CR
IP T
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com
by NATIONAL UNIVERSITY OF SINGAPORE on 06/01/20. Re-use and distribution is strictly not permitted, except for Open Access articles.

AC
C EP
TE

15
DM
AN
Accepted manuscript to appear in IJSSD

US
Figure 7 Multi-mode vibration control effects with non-optimal actuator #12.
CR
IP T
T
Accepted manuscript to appear in IJSSD

IP
CR
by NATIONAL UNIVERSITY OF SINGAPORE on 06/01/20. Re-use and distribution is strictly not permitted, except for Open Access articles.

US
AN
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com

DM

Figure 8 Multi-mode vibration control effects with non-optimal actuator #16.


TE

4 Optimal vibration control: energy allocation


In section 3, the optimal displacement of the actuator is explored. It has beendemonstrated that the
optimally positioned actuator can provide much better control performance than the non-optimal one.
For single-mode vibration control, when the optimal position of the actuator has been identified, the only
control gain can be determined by the input control voltagelimitation (in this case, ±500V). Therefore,
EP

for single-mode vibration control, the optimal control gain is unique. However, for multi-mode vibration
control, the control gain vector contains multiple control gains corresponding to each participating mode.
This means that the control gain vector that satisfies the total input voltage limitation is not unique. Since
the control gain vector represents the allocation of the control energy on each vibration mode, one can
imagine that those modes with relatively larger energy of vibration may need more input control energy
C

to achieve the vibration suppression. Therefore, even when the actuator has been placed at the optimal
position, the control energy should be optimally allocated to each mode, i.e., the unique optimal control
AC

gain vector should be identified.


16
T
Accepted manuscript to appear in IJSSD

IP
4.1 Optimization criterion
In order to achieve the best control performance, it is assumed that all the vibration modes should
attenuate simultaneously, namely, in the same amount of time, the percentage of attenuation of all modes
are the same. This can not only ensure the vibration attenuation velocity but also maximize the energy

CR
utilization efficiency. In view of this thought, an optimization criterion for determining the optimal
control gain vector for multi-mode vibration control is proposed in this section.
According to the simultaneous attenuation, the two controlled modes should be suppressed by the
by NATIONAL UNIVERSITY OF SINGAPORE on 06/01/20. Re-use and distribution is strictly not permitted, except for Open Access articles.

same percentageafter the same time T. Assume that the frequencies of the two modes are ωp and ωq. At
the initial moment, the two modes are at their first period. After a period of T, the two modes are at their

US
a+1 and b+1 period, respectively; and their vibration amplitudes are both suppressed to 1/k of the initial
amplitudes, namely,
2 2
T a b , (34)
p q

A1p A1q
 k,
AN (35)
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com

Aap1 Abq1
where A denotes the vibration amplitude; the superscript and the subscript stand for the mode number and
the current period number, respectively.
Additionally, the damping ratios of the two modes can be expressed as
1 A1p 1 Aq
p= ln , q = ln q1 . (36)
2 a p
Aa 1 2 b Ab 1
DM

According to equation (34)~(36), the frequency and damping ratio of the two modes should satisfy the
following relation:
 p b q
  , (37)
 q a p
or more concisely,
 p  p   q q . (38)
TE

Therefore, the optimization criterion can be expressed as: for all participating modes in the vibration, the
product of the equivalent damping ratio and the vibration frequency of each mode should be a constant.
4.2 Numerical simulations
The multi-mode vibration control with initial amplitude A0 = [0.0006, 0.0006, 0.0006, 0.0006] m is
studied in this section, which remains the same with the previous cases. The actuator is placed at the
EP

optimal position #17. Based on the proposed energy allocationoptimization criterion, the optimal control
gain vector is obtained as Gc_opt = [12540, 4487, 4487, 2508]. One may note that this control gain vector
is the same as that used in the case of Figure 5. Recall that in Section 3.3, we have directly given the
value of the control gain vector of each case without explanation. Actually, all the control gain vectors
were obtained based on this optimization criterion. However, in order to demonstrate the superiority of
C

the optimal control gains over the non-optimal ones, three other cases with non-optimal control gains are
also numerical simulated as a comparison. The non-optimal control gain vectors are chosen as Gc_1 =
[2000, 7500, 7500, 2000], Gc_2 = [18000, 3000, 3000, 3000] and Gc_3 = [2500, 2500, 2500, 9500].For
AC

17
T
Accepted manuscript to appear in IJSSD

IP
the non-optimal cases, the time responses of each vibration mode and the total average displacement of
the membrane are shown in Figure 9to Figure 11. Since the optimal case is just the same with Figure 7,
there is no need to present it again in this section. The induced equivalent damping ratio of each mode of
different control gains arelisted in Table 6.

CR
Table 6 The equivalent damping ratio of multi-mode vibration with different control gains.

ζmn Optimal control gains Non-optimal control gains


c
Mode G _opt Gc_1 Gc_2 Gc_3
by NATIONAL UNIVERSITY OF SINGAPORE on 06/01/20. Re-use and distribution is strictly not permitted, except for Open Access articles.

(1,1) 1.94% 0.31% 2.78% 0.39%


(1,2) 1.18% 1.97% 0.79% 0.66%

US
(2,1) 1.18% 1.97% 0.79% 0.66%
(2,2) 0.92% 0.74% 1.10% 3.49%

By comparing Figure 9~11 with Figure 5 one can find that the total control performance of the
non-optimal control gains is obviously not as good as the optimal ones. Specifically, for non-optimal
Gc_1, one can see that the second and the third terms of the control gain vector are relatively larger, which

AN
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com

indicates that more energy is allocated to the second and the third modes. Therefore, it can be found from
Figure 9 that although mode (1,2) and mode (2,1) have been suppressed completely at 4s, mode (1,1)
still exhibits a relatively large vibration amplitude at 8s. This leads to overall poor control performance.
Similarly, for non-optimal Gc_2, more energy has been allocated to mode (1,1). In this case, the control
effect is better than that of Gc_1, but small amplitude vibrations of mode (1,2) and mode (2,1) still exist at
8s. For non-optimal Gc_3, most of the control energy is allocated to the fourth mode. One can see from
DM

Figure 11 that mode (2,2)is quickly suppressed within 2s due to the large input control voltage. However,
the input voltages of other modes are quite small, especially mode (1,1). It can be imagined that such a
small input voltage cannot provide a satisfactory control effect on mode (1,1). As expected, the overall
control performance is even worse than that of Gc_1. Look back to Figure 5, one can find that the
vibration amplitudes of all modes are suppressed simultaneously and have nearly been completely
controlled at 6s. This evidently demonstrated that the optimal allocation of the control energy is of great
significance for effective vibration control of the membrane structures.
TE
C EP
AC

18
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com
by NATIONAL UNIVERSITY OF SINGAPORE on 06/01/20. Re-use and distribution is strictly not permitted, except for Open Access articles.

AC
C EP
TE

19
DM
AN
Accepted manuscript to appear in IJSSD

US
Figure 9Multi-mode vibration control effects with non-optimal control gainGc_1.
CR
IP T
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com
by NATIONAL UNIVERSITY OF SINGAPORE on 06/01/20. Re-use and distribution is strictly not permitted, except for Open Access articles.

AC
C EP
TE

20
DM
AN
Accepted manuscript to appear in IJSSD

US
Figure 10 Multi-mode vibration control effects with non-optimal control gainGc_2.
CR
IP T
T
Accepted manuscript to appear in IJSSD

IP
CR
by NATIONAL UNIVERSITY OF SINGAPORE on 06/01/20. Re-use and distribution is strictly not permitted, except for Open Access articles.

US
AN
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com

DM

Figure 11 Multi-mode vibration control effects with non-optimal control gainGc_3.


5 Conclusions
TE

In this study, the dynamic equations of the vibration of the membrane are derived and solved first, and
the actuating characteristics of the PVDF actuator and control strategy are investigated successively.The
optimal vibration control of the membrane is then explored. The actuator placement optimization method
and the energy allocation optimization criterion are studied respectively. The simulation results indicate
EP

that both the aforementioned two factors have a strong influence on the vibration control performance. If
the actuator is not placed at the appropriate position, some vibration components may not be suppressed
by the actuator no matter how large input voltage is applied. However, even if the actuator is placed at the
optimal position, if the control gains are set unreasonable, the vibration cannot be suppressed effectively.
The approach introduced in this work can guarantee the best control performance formembrane vibration;
C

also, it can be extended to other flexible or gossamer structures.


Acknowledgments
AC

This research is supported by grant from the National Natural Science Foundation of China (51835002),
21
T
Accepted manuscript to appear in IJSSD

IP
the China Postdoctoral Science Foundation (2019M661289), the Heilongjiang ProvincialPostdoctoral
Science Foundation (LBH-Z19150) and the Self-Planned Task of the State Key Laboratory of Robotics
and System (SKLRS201808B).
Conflict of interest

CR
The authors declare that they have no conflict of interest.
References
1. L. Dell’Elce, G. Kerschen, Probabilistic Assessment of Lifetime of Low-Earth-Orbit Spacecraft:
by NATIONAL UNIVERSITY OF SINGAPORE on 06/01/20. Re-use and distribution is strictly not permitted, except for Open Access articles.

Uncertainty Propagation and Sensitivity Analysis, J. Guid. Control. Dyn. 38 (2015) 886–899.
2. S.K. Jeon, J.N. Footdale, Scaling and Optimization of a Modular Origami Solar Array, in: 2018

US
AIAA Spacecr. Struct. Conf., American Institute of Aeronautics and Astronautics, Reston, Virginia,
2018: pp. 1–15.
3. Z. Zhou-Lian, L. Chang-Jiang, H. Xiao-Ting, C. Shan-Lin, Free Vibration Analysis of Rectangular
Orthotropic Membranes in Large Deflection, Math. Probl. Eng. 2009 (2009) 1–9.
4. L. Chang-jiang, Z. Zhou-lian, H. Xiao-ting, S. Jun-yi, S. Wei-ju, X. Yun-ping, L. Jun, L-P

AN
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com

Perturbation Solution of Nonlinear Free Vibration of Prestressed Orthotropic Membrane in Large


Amplitude, Math. Probl. Eng. 2010 (2010) 1–17.
5. C.J. Liu, Z.L. Zheng, X.Y. Yang, J.J. Guo, Geometric Nonlinear Vibration Analysis for
Pretensioned Rectangular Orthotropic Membrane, Int. Appl. Mech. 54 (2018) 104–119.
6. X. Liu, G. Cai, F. Peng, H. Zhang, L. Lv, Nonlinear Vibration Analysis Of A Membrane Based On
Large Deflection Theory, J. Vib. Control. 24 (2018) 2418–2429.
DM

7. M. Ahmadi, G. Hashemi, M. Jamali, M.R. Dehkordi, S.A. Razavian, Nonlinear free vibration
analysis of pre-stressed membranes, Proc. Inst. Mech. Eng. Part K J. Multi-Body Dyn. 231 (2017)
346–356.
8. C.-J. Liu, Z.-L. Zheng, X.-Y. Yang, H. Zhao, Nonlinear Damped Vibration of Pre-Stressed
Orthotropic Membrane Structure Under Impact Loading, Int. J. Struct. Stab. Dyn. 14 (2014) 1350055.
9. Z. Zheng, W. Song, C. Liu, X. He, Study on Dynamic Response of Rectangular Orthotropic, 26
(2012) 1467–1479. doi:10.1163/156856111X618335.
TE

10. S.C. Gajbhiye, S.H. Upadhyay, S.P. Harsha, Nonlinear vibration analysis of piezo-actuated flat thin
membrane, J. Vib. Control. 21 (2015) 1162–1170.
11. S. Carra, M. Amabili, R. Ohayon and P.M. Hutin, Active vibration control of a thin rectangular plate
in air or in contact with water in presence of tonal primary disturbance, Aerosp. Sci. Technol. 12
(2008) 54-61.
EP

12. G. Ferrari, M. Amabili, Active vibration control of a sandwich plate by non-collocated positive
position feedback, J. Sound Vib. 342 (2015) 44–56.
13. A. Zippo, G. Ferrari, M. Amabili, M. Barbieri, F. Pellicano, Active vibration control of a composite
sandwich plate, Compos. Struct. 128 (2015) 100–114.
14. Y. Lu, Q. Shao, M. Amabili, H. Yue, H. Guo, Nonlinear vibration control effects of membrane
C

structures with in-plane PVDF actuators: A parametric study, Int J Non Linear Mech. 122 (2020)
103466.
AC

15. Y.-F. Lu, H.-H. Yue, Z.-Q. Deng, H.-S. Tzou, Adaptive Shape Control for Thermal Deformation of
22
T
Accepted manuscript to appear in IJSSD

IP
Membrane Mirror with In-plane PVDF Actuators, Chinese J. Mech. Eng. 31 (2018) 9.
16. E.J. Ruggiero, D.J. Inman, Modeling and vibration control of an active membrane mirror, Smart
Mater. Struct. 18 095027.
17. Y. Lu, H. Yue, Z. Deng, H. Tzou, Experiments on quasi-static and dynamic control of a PVDF

CR
laminated membrane-like mirror, in: Dyn. Vib. Control, vol. 4B, American Society of Mechanical
Engineers, 2016, p. V04BT05A052.
18. Y. Lu, M. Amabili, J. Wang, F. Yang, H. Yue, Y. Xu, H. Tzou, Active vibration control of a
by NATIONAL UNIVERSITY OF SINGAPORE on 06/01/20. Re-use and distribution is strictly not permitted, except for Open Access articles.

polyvinylidene fluoride laminated membrane plate mirror, J. Vib. Control. (2019).


19. Y. Zhang, T. Hiruta, I. Kajiwara, N. Hosoya, Active vibration suppression of membrane structures

US
and evaluation with a non-contact laser excitation vibration test, J. Vib. Control. 23 (2017) 1681–1692.
20. H. Sakamoto, K.C. Park, Y. Miyazaki, Distributed and Localized Active Vibration Isolation in
Membrane Structures, J. Spacecr. Rockets. 43 (2006) 1107–1116.
21. X. Liu, H. Zhang, L. Lv, F. Peng, G. Cai, Vibration control of a membrane antenna structure using
cable actuators, J. Franklin Inst. 355 (2018) 2424–2435.

AN
Int. J. Str. Stab. Dyn. Downloaded from www.worldscientific.com

22. X. Liu, G. Cai, F. Peng, H. Zhang, Dynamic model and active vibration control of a membrane
antenna structure, J. Vib. Control. 24 (2018) 4282–4296.
23. X. Liu, G. Cai, F. Peng, H. Zhang, Active control of large-amplitude vibration of a membrane
structure, Nonlinear Dyn. 93 (2018) 629–642.
24. H. Shi, C. Wang, L. Liu, Z. Gao, Y. Xie, An active control strategy to suppress nonlinear vibrations
of large space membranes, Acta Astronaut. 155 (2019) 80–89.
DM

25. L. Sirota, Y. Halevi, Wave based vibration control of membranes, in: 2014 Am. Control Conf.,
IEEE, 2014: pp. 2729–2734.
26. X. Liu, H. Zhang, L. Lv, F. Peng, G. Cai, Wave based active vibration control of a membrane
antenna structure, Meccanica. 53 (2018) 2793–2805.
27. X. Liu, G. Cai, F. Peng, H. Zhang, Piezoelectric Actuator Placement Optimization and Active
Vibration Control of a Membrane Structure, Acta Mech. Solida Sin. 31 (2018) 66–79.
28. F. Peng, A. Ng, Y.R. Hu, Actuator placement optimization and adaptive vibration control of plate
TE

smart structures, J. Intell. Mater. Syst. Struct. 16 (2005) 263–271.


29. D. Chhabra, G. Bhushan, P. Chandna, Optimal placement of piezoelectric actuators on plate
structures for active vibration control via modified control matrix and singular value decomposition
approach using modified heuristic genetic algorithm, Mech. Adv. Mater. Struct. 23 (2016) 272–280.
30. K.R. Kumar, S. Narayanan, The optimal location of piezoelectric actuators and sensors for vibration
EP

control of plates, Smart Mater. Struct. 16 (2007) 2680–2691.


31. M. Amabili, Nonlinear Mechanics of Shells and Plates: Composite, Soft and Biological Materials,
Cambridge University Press, 2018, New York, USA.
32. M. Amabili, Nonlinear vibrations and stability of shells and plates, Cambridge University Press, Cambridge, 2008.
33. Y. Lu, H. Yue, Z. Deng, H. Tzou, Distributed sensing signal analysis of deformable
C

plate/membrane mirrors, Mech. Syst. Signal Process. 96 (2017) 393–424.


34. Y. Lu, H. Yue, Z. Deng, H. Tzou, Distributed microscopic actuation analysis of deformable plate
membrane mirrors, Mech. Syst. Signal Process. 100 (2018) 57–84.
AC

23

You might also like