1
and converse piezoelectric eects being exploitable to sense the structural response and
to apply the control action. A wide amount of literature is devoted to the basic problems
of this technology and only a very partial list is hereafter summarized. Theoretical and
mathematical models of piezoelectric materials have been available in the literature from
the beginning of their use, e.g. [1],[2]. The associated analytical solutions, however, are
restricted to simple geometries and boundary conditions; several papers describe more
general approaches, most of them base on nite element approximation, e.g. Tzou and
Tseng [3], Hwang and Park [4] and Wang and Rogers [5], while a simpler but popular
approach is represented by the pin force method, e.g. [6]. Furthermore a number of
applications, referring to numerical and experimental evaluation of piezo active systems,
are available, ranging from the vibration active control of beams and panels, e.g. Crawley
and Anderson [7], Gaudenzi [8], Yang and Lee [9]or Yang and Chiu [10], to the active
control of noise due to vibrations, e.g. Clark and Fuller [11] or Dimitradis, Fuller and
Rogers [12].
In this paper some of the experiences, gathered in testing piezoelectric control systems are
summarised. These activities are aimed at controlling the vibration of a clamped rectan-
gular
at panel, mechanically excited and actuated by piezoceramic patches. Correlations
with numerical predictions are also presented. These are based on a integrated nite el-
ement program capable of describing the dynamic behaviour of the system composed by
a generic structure embedding piezoelectric devices (Ghiringhelli et Al.[22]). It allows a
natural integration of structural and piezoelectric models with actual signal conditioning
circuits and any other electronic devices used to realise the control system.
A completely analogic active control system and simple direct feedback laws have been
used to point out problems related to measurement conditioning circuits. Self sensing
actuators, the usefulness of which is well known (Dosch et Al.[13], Saunders et Al.[14],
Brusa et Al.[15], Akishita et Al.[16], Cole and Clark[17], Anderson and Hagood[18], Tsou
and Hollkamp[19], Ko and Tongue[20], Chang-qing et Al.[21]), have also been tested using
appropriate bridge and conditioning circuits. The theoretical possibilities oered by this
kind of devices are evident but the actual realisation entails non trivial problems and
some of these papers pointed-out diculties in keeping the balance of sensor bridges.
The choice of a fully analogic system is due to two main considerations. First of all the
need of testing and tuning electronic devices suitable to be used in actual implementation
of the control system (the testing of a digital multi-input multi-ouput active control and of
adaptive control laws is in progress). The second reason is related to the diculties that
2
can be envisaged when implementing a digital control the band of which reaches relatively
high frequencies: in this case the availability of well tested analogic control systems could
ease the solution of some problems.
2. INTEGRATED MODELISATION OF PIEZOELECTRIC
In this section the main features of an integrated modeling techniques, based on a nite
element discretisation, are presented (Ghiringhelli et Al.[22]).
2.1 Piezoelectric eect
The direct and inverse piezoelectric eects, i.e. the generation of an electric eld due to
mechanical strains and the ability to induce a strain when subjected to an electric eld,
are described by the linear constitutive laws of piezoelectric materials, in IEEE standard
notation [23]:
h i
fT g = cE fS g , [e]T fE g (1)
h i
fDg = [e] fS g + S fE g (2)
where fT g is the stress vector, fS g is the strain vector, fE g is the electric eld, fDg is the
h i
electric displacement, cE is the elastic matrix at constant electric eld, [e] is the piezo-
n o
electric characteristic matrix and S is the dielectric matrix at constant strain. Indeed
Eqs. 1 and 2 represent both a simplication and a linearisation of generally nonlinear
constitutive equations that characterise the more general electro-magneto-thermo-elastic
problem. These laws show a simultaneous dependence on stress fT g, heat
ux fqg, cur-
rent density fj g, electric displacement fDg and magnetising eld fH g from strain fS g,
temperature T , electric and magnetic elds fE g,fB g (Eringen[2]), i.e.:
fj g = J (fS g ; fE g ; fB g ; T ) ; fH g = H (fS g ; fE g ; fB g ; T )
fqg = Q (fS g ; fE g ; fB g ; T )
fT g = T (fS g ; fE g ; fB g ; T ) ; fDg = D (fS g ; fE g ; fB g ; T )
A completely coupled formulation can be obtained, but usually no dependence of stresses
and electric displacement on magnetic eld and temperature is assumed, i.e. a simple
electrostatic assumption. Small changes, with respect to a reference temperature, are
then considered, as well as frequencies low enough to avoid heating of the piezoelectrics,
3
due to the hysteresis associated to the direct and inverse eects. Finally a linearisation
of the constitutive laws in a reference conguration is made, i.e.:
" # " # " # h i " @ fDg #
= @@ ffTS gg [e] = @@ ffD g = @@ ffET gg
h i
cE Sg [e]T S = @ fE g
leading to Eqs. 1 and 2.
2.2 Response of a controlled system
Starting from Eqs. 1 and 2 the electro-structural model can be obtained using the inte-
grated nite element approach described in Ghiringhelli et Al.[22], the results of which
are brie
y recalled here. Once the structural displacement d and the electric
ux , that
is the voltage time integral Z
= V dt
are assumed as primary eld unknowns, they are approximated by means of a nite
element approach in terms of the corresponding nodal values fdg and fg.
The use of an appropriate nite element discretization allows to write a generalised Virtual
Work Principle equation and the exploitation of this principle leads to a linear form of
the dynamics equations of the coupled system, i.e.:
" #( ) " #( ) " #( ) ( )
M 0 d + D ,A d_ + K 0 d = f (3)
0 ,CP ,ST 0 _ 0 0 iP
where [M ] ; [D] ; [K ] are the mass, damping and stiness matrices, fdg is the displacements
vector, ff g is the loads vector, fiP g is the nodal currents vector, [A] is the electro-elastic
coupling matrix related to actuators, [S ] is the electro-elastic coupling matrix related to
sensors and [CP ] is the dielectric matrix. In this equation the second row is changed in
sign to enforce the symmetry of the problem. The analytical denition of these terms and
other details on the nite element development can be found in [22] and [24] respectively.
When piezoelectrics are used for control purposes, the electric unknowns are naturally
limited to a relatively small number of voltages, related to controlled electrodes. On the
contrary, Eq. 3 can contain too many structural unknowns to be directly used in the
design of active control systems. Thus it is generally necessary to take a condensation of
the structural part of Eq. 3. Here a modal transformation of the structural part alone
has been adopted, based on the eigenvectors matrix [X ] of the structural problem, ( i.e.
fdg = [X ] fqg ), fqg being the generalised structural dofs., so that we can write the
4
generalised form of Eq. 3:
" T #( ) " #( )
X MX 0 q + X T DX ,X T A q_ +
0 ,CP ,ST X 0 _
" #( ) ( )
+ X 0KX 00 q = XT f
T
iP (4)
The previous modal condensation does not change the structure of the dynamics equation,
in the following the previously dened symbols will take the generalised meaning.
In the second row of Eq. 3 we can immediately recognise the matrix form of the sensing
equation, that can be alternatively derived from the second equation of the piezoelectric
constitutive law. The array arranging the currents due to the piezoelectric eect is given
by: fiP g = [S ]T fq_g .
The assumption of the time integral of the voltage as electric primary unknown, instead
of the most commonly used voltage, allows a natural formulation of the coupled elec-
tric/elastic problem in presence of external circuits. In fact a complete analogy between
structural displacements and electric
uxes is established: networks of electric components
can be easily modelled, as properly connected generalised lumped capacitors (masses), re-
sistors (dampers) and inductors (springs), and an integrated linear problem is obtained
by simply appending the electric model to Eq. 3; by expanding the array of
uxes with
the values at the nodes of the external electric network we can write:
" #( ) " #( )
M 0 q + D ,A q_ +
0 CL , CP ,S GL
T _
" #( ) ( )
+ K0 R0 q =
i
f
+ (5)
L P iL
where [CL], [GL], and [RL] are respectively the lumped symmetric capacitance, conduc-
tivity and reluctance matrices. They are assembled from local matrices, related to each
side of the electric circuit, while iL are current lumped inputs.
2.3 Direct feedback with co-located control
A co-located control can be easily implemented provided the two elements of a pair of
piezoelectrics are used to sense and actuate respectively. In fact, if the host structure is
thin enough, we can assume the measurements on the top surface to be fully equivalent
to those on the bottom, one once its sign is properly assumed: in this way a co-located
control is obtained, in the limits of two identical and perfectly placed devices. The benets
of a co-located control are well known, in particular a direct velocity feedback scarcely
5
suer spillover problems even if the actuator dynamics must be carefully evaluated to grant
unconditional stability, e.g. [26], [27]; a feedback on the position, i.e. the positive position
feedback [32], also showed interesting properties of insensitivity to spillover furthermore
being not destabilized by nite actuator dynamics.
The measurement of the piezoelectric charge has been frequently used as the input of a
direct feed-back; the conditioning circuit needed for this job can be simply built using a
voltage follower: the circuit is a high pass lter that behaves like an ideal displacement
transducer in the frequency range of interest by measuring the piezoelectric charge. The
corresponding sensor equation is:
fVS g = [CP ],1 [S ]T fqg (6)
While the control voltage is dened as:
fVC g = , [G] fVS g (7)
In Eq. the control voltage has been dened by means of the gain matrix, [G]; it can be
fully populated but, in the applications here presented, only a diagonal form has been
tested.
[M ] fqg + [C ] fq_g + [K ] + [A] [G] [CP ],1 [S ]T fqg = ff g (8)
This kind of control is usually referred to as a position feed-back, e.g. [8] and [29],
because the sensing voltage, in Eq. 6, is related to the structural displacement through
the array of modal amplitudes. Indeed this measurement senses the surface strain, i.e. the
local curvature of the panel, and a priori it cannot be linearly related to the transversal
displacement at the piezoelectric location. So in the following it will be referred to as
strain feedback.
By the use of a measurement of the piezoelectric current, a strain rate feedback can be
implemented: the second row of Eq. 3 is straightforwardly exploited for this:
fVS g = [R] [S ]T fq_g (9)
leading to the following controlled system dynamics equation:
[M ] fqg + [C ] + [A ] [G] [R] [S ]T fq_g + [K ] fqg = ff g (10)
In these equations the matrix [R] is diagonal and is representative of the resistive branches
along with the piezoelectric currenties fiP g
ow. In analogy with the previous discussion
6
this control is known as a velocity feedback, e.g. [8],[28], [25]. It has been also tested,
using a current to voltage conversion circuit, but this conditioning circuit failed: high
frequency instabilities occurred as the control gain was increased (analoguous problems
have been already outlined in the literature, e.g. Akishita et Al.[16] and in Hong el Al.
[25]).
It is well known that a colocated control (both in position and in velocity) is stable. But
practical applications lead to imperfect systems and some of the possible causes can be
cited: the actual frequency properties of each element of the control system, e.g. sen-
sor conditioning circuits and the driving devices, and the imperfect co-location, due to a
wrong position of sensors and actuators. The use of piezoelectric patches worsen these
problems. In fact piezos represent high capacitive loads for driving devices, in this way
limiting their frequency properties. Furthermore they are more sensitive to high frequency
modes, when used to sense, as well as they have more authority on them, when used to
actuate, in this way entailing problems in an ecient ltering. At last the piezos of a pair
can show some dierences, e.g. in sizes, piezoelectric and dielectric properties, possibly
worsening the aforementioned sensitivity problems.
Then instabilities are expected, especially with high gains, but they arise after a good
eectivenes has been achieved, during tests on strain feedback control, while, using the
strain-rate feedback, the instability onset is very early, then preventing any positive re-
sults. This fact is worth studying in depth by exploiting the analysis of the properties of
the control matrix, in relation to stability theorems. In fact in the ideal case it is sym-
metric and positive semi-denite, but this is no more true in practical applications: when
the actual properties of devices are introduced, independent sensing and control coupling
matrices should be used. It is apparent that it can now be no more symmetric and its
symmetric part, fundamental for the stability of the control, no more positive denite.
3. EXPERIMENTAL SET-UP
The structure under test is a
at panel (600x400x2mm), made of 2027 Aluminium Alloy,
clamped on each side to a steel frame (Fig. 2). The main structure is elastically sus-
pended to a support frame and it is excited by an electro-mechanical shaker (B&K Type
4810). To limit the dynamic response of the frame torsional mode (
= 450Hz), the shaker
9
has been placed on a symmetry plane (Fig. 2). Eight pairs of piezeceramic patches have
been used, made of PZ21A (See Tab.1), produced by Ferroperm (DK), the dimensions of
which are 50x50x0:5mm. The elements of each couple are bonded, by means of epoxy
resin, in corresponding locations on the two surfaces of the panel. These locations have
been chosen to control selected low frequency modes with an odd number of half waves
along the short side, due to the previously mentioned reason. To obtain better perfor-
mances the piezoelectrics have been intuitively placed as far as possible from areas where
the modal curvature of the selected modes vanishes, as depicted in Fig. 3. In Table 2 a
comparison is made between experimental and Finite Element natural frequencies of the
panel equipped with piezoelectric devices. No adjustements were made to improve this
comparison because the goal of the test was also to verify the robustness of the adopted
design techniques against model uncertainities.
10
easier. The power stage and the previously described gains regulations have been used.
The modal lter implementation allows the signal to be amplied by a 5dB factor at
the corner frequency together with a 180deg change in phase across this frequency. It is
important to note that while the bridge allows to sense a strain rate, the 90deg change
in phase, produced by the modal lter at the corner frequency, transforms it in a strain
measurement.
11
4.2 The single self-sensing actuator
The use of a single piezoelectric device as a self-sensing actuator, combined with the modal
lter, gave satisfactory results. Also in this test case the results shown are related to a
xed frequency excitation and measurements are related to the output of an accelerome-
ter. The lters have been tuned by setting the working frequency to match specic modes.
Simulations have been performed by including the power amplier transfer function and
the bridge electric model. A 5% unbalance has been introduced and experimental gains
have been used. In fact this set of testing activities was carried out before the simulations,
because of a frequent operative overlapping, and simulations were in some case driven by
the experimental observations, e.g. in relation to the adopted control gains. In partic-
ular, during the tests the maximum stable gain has been achieved by a trial and error
procedure and these values have been used in simulations afterward. In Tables 4 and 5
the dynamic response, both in controlled and uncontrolled case are presented for several
excitation frequencies. The eectiveness of the control is apparent also in this case, but
the correspondence between tests and numerical predictions is not satisfactory, as it can
be veried in Figure 10 and Table 6, where some of these experimental and numerical
results are compared in terms of acceleration reduction. In Figure 10 the rst two bars
refer to the performances measured during the tests and predicted by the simulation: it
is possible to appreciate that a lower authority is often predicted by the simulation.
Based on these data we noted that the control eectiveness was often underestimated
by the numerical predictions. Furthermore we found that the experimental gains do not
mark the stability limit of the numerical model. So the same procedure has been applied
to the numerical simulations and the largest possible gains have been assessed: by this
approach the correlation improved signicantly. The results of this procedure are shown
in Table 6, and depicted by the third bar in Figure 10: an overestimate of the eectiveness
on the second mode is present also in this case, conrming the remarks of the previous
section, but the behaviour of the other modes is satisfactorily predicted as well.
4.3 Double self-sensing actuator
The basic exploitation of the availability of a self-sensing actuator consists in using a
couple of opposite elements for actuation purposes, so that a better performance, in re-
lation to the previous cases, is expected. Actually the bridge balancing resulted more
dicult in this case and lower gains were found to be usable at some frequencies. This
12
can be justied with the considerations previously presented: due to the system layout,
the capacitance driven by the power amplier is doubled (two patches are controlled), in
this way reducing its dynamic range and exacerbating the sensitivity problems on higest
modes.
Also in this case the results of harmonic tests at several frequencies are presented. Fig-
ure 11 and Table 7 compare experimental measurements and simulations. Again a satis-
factory agreement has been found once simulation gains have been scaled to the maximum
allowable to avoid instabilities. The use of a pair of self-sensing actuators led in general
to better performances but at the same time both simulations and experiments showed
lower maximum gains can be used in some cases.
4.4 Acoustic
5. CONCLUDING REMARKS
The research activities demonstrated once again the possibilities oered by piezoelectric
patches in suppressing vibrations, even by using very simple control laws, e.g. a direct
strain feedback.
The inecient behaviour of velocity sensors should be further investigated: in fact this
kind of feedback shows several typical advantages but due to the specicity of the co-
located piezoelectric control it could lose part of its appeal. Nonetheless the availability
of references successfully exploiting it cannot be disregarded, even if on this subject the
literature is contradictory.
The use of a self-sensing actuator, together with a modal lter of the input signal leads
to results comparable to those obtained using separate sensor and actuator devices, but
with the saving of a piezo patch. The use of a pair of self-sensing actuators improved,
but not dramatically, the eectiveness of the active control, due to some diculties in
balancing the bridge and because lower maximum gains can be used at some frequencies.
The formulation adopted to simulate the dynamic behaviour of the coupled system and
to design the active control shows a satisfactory agreement with experimental results, so
that the related performances can be reasonably predicted. The availability of validated
models, conditioning circuits and power ampliers allows one to continue with the design
and implementation of a digital control system.
Acknowledgements
The author is grateful to Mr.Mariofelice Zanardi for his helpful contribution to the work.
This research was supported in part by ENEL-CRA under Grant No.R23TC0009 and by
the National Council of Researches - CNR under Grant No.97.00891.PF34.
References
[1] Crawley E.F. and deLuis J, 1987 Use of piezoelectric actuators as elements of intel-
ligent structures, AIAA J., V.25, 1373-1385
[2] Eringen A.C. and G.A.Maugin, 1989, "Electrodynamics of Continua", Vol. 1 and 2,
Springer- Verlag.
14
[3] Tzou H.S. and Tseng C.T., 1990, Distributed piezoelectric sensor/actuator design
for dynamic measurement control of distributed parameter systems: a piezoelectric
nite element approach, J. of Sound and Vibration, V.138, 17-34
[4] Hwang W.S. and Park H.C., 1993, Finite element modeling of piezoelectric sensors
and actuators, AIAA J., V.31 ,No.5, May 1993, 930-937
[5] Wang B.T. and Rogers C.A., 1991, Modeling of nite-length spatially-distributed
induced strain actuators for laminate beams and plates, J.of Intell. Mater. Syst. &
Struct., V.2, 38-58
[6] Chaudhry Z. and Rogers C.A., 1994, The Pin-Force Model Revisitated, J.of Intell.
Mater. Syst. & Struct., Vol.5, May, pp.347-354
[7] Crawley E.F. and Anderson E.H., 1990 Detail models of piezoeletric actuation of
beams, J.of Intell. Mater. Syst. & Struct., V.XXX, 4-25
[8] Gaudenzi P., R. Barboni, R.Carbonaro and S. Accetalla "Direct position and velocity
feedback control on an active beam with PZT sensors and actuators", Proceedings
of the International Forum on Aeroelasticity and Structural Dynamics , 17-20 June
1997, Rome
[9] Yang S.M. and Lee Y.J. 1993 Vibration suppression with optimal sensor/actuator
location and feedback gain Smart Mater. & Struct. Vol.2 232-239
[10] Yang S.M. and Chiu J.W. 1993 Smart structures - vibration of composites with
piezoelectric materials Composite Structures, 25, 381-386
[11] Clark L.R. and Fuller C.R., 1992 Experiments on Active Control of Structurally
Radiated Sound Using Multiple Piezoceramic Actuator J.of Acoustical Society of
America, 91(6), 3313-3320
[12] Dimitradis E.K., Fuller C.R. and Rogers C.A., 1991 Piezoelectric actuators for dis-
tributed vibration excitation of thin plates, J.of Vib.Acoust., V.113, 100-107
[13] Dosch J.J., D.J. Inman, E. Garcia, 1992 "A self-sensing piezoelectric actuator for
collocated control", Journal of Intelligent Material System and Structures, Vol.3,
pp.659-667, Jan.
15
[14] Saunders W.R., D.G. Cole and H.H. Robertshaw "The impact of piezoelectric senso-
riactuators on active structural acoustic control", Proceedings of the SPIE, Vol.1917,
No.1,p.578-586
[15] Brusa E., S. Carabelli, A. Tonoli, 1996 " Self-Sensing collocated structures with dis-
tribuited piezoeletric transducers", Mechatronics Laboratory Politecnico di Torino,
ICAST 1996, pp.84-94, Roma, Italy
[16] Akishita S., Y. Mitani and H. Miyaguchi, 1994 "Sound transmission control through
rectangular plate by using piezoelectric ceramics as actuators and sensors", Journal
of Intelligent Material System and Structures, Vol 5, May
[17] Cole D.G., Clark R.L., 1994 "Adaptive compensation of piezoelectric sensoriactua-
tors", Journal of Intelligent Material System and Structures, Vol 5, pp.665-672
[18] Anderson E.H., Hagood N.W., 1992 "Self-sensing piezoelectric actuation: analysis
and application to controlled structures", AIAA/ASME/ASCE/AHS/ASC Struc-
tures, Structural Dynamics and Material Conf., Dallas TX, pp.2141-2155
[19] Tsou H.S., Hollkamp J.J, 1997, "Collocated independent modal control with self-
sensing orthogonal piezoelectric actuators", AIAA-94-1737-CP
[20] Ko B., Tongue B.H., 1995, "Acoustic control using a self-sensing actuator", J. of
Sound and Vibration, Vol.187(1),pp.145-165
[21] Chang-qing C., W.Xiao-ming, S.Ya-peng, 1996, "Finite element approach of vibration
control using self-sensing piezoelectric actuators", Computers & Structures, Vol.60,
No.3, pp.505-512.
[22] Ghiringhelli G.L., M. Lanz, P. Mantegazza, 1993. " Numerical modelling and exper-
imental testing of distribuited piezoeletric actuators", Int. Forum of Aeroelasticity
and Structure Dynamics, Vol.1, pp.277-292, Strasbourg
[23] ANSI/IEEE Standard notation for piezoelectricity, 1988
[24] Ghiringhelli G.L., " Numerical modelling and experimental testing of distribuited
piezoeletric actuators", l'Aerotecnica Missili e Spazio, 1995
16
[25] Hong S.Y., V.V. Varadan and V.K.Varadan, 1991 "Comparison of analog and digital
strategies for automatic vibration control of lightweight space structures", SPIE, Vol.
1489 Structure Sensing and Control, pp.75-83
[26] Aubrun J.N., "Theory of the control of structures by low-authority controllers",
Journal of Guidance and control, Vol.3, No.5, 1980, 444-451
[27] Chen C.L., "Direct output feedback control for large space structures", Dynamics
Laboratory Report DYNL-82-1, California Institute of Technology, 1982
[28] Sullivan J.M., S.E. Burke and J.E> Hubbard, 1995 "Experimental demonstration of
active broadband vibration suppression of a rectangular plate using gain-weighted,
shaped distributed transducers", AIAA-95-1119-CP, American Institute of Aeronau-
tics ans Astronautica, pp. 3379-3389
[29] Denoyer K.K., Kwak M.K. "Dynamic Modelling and Vibration Suppression of a
Slewing Structure utilizing Piezoelectric Sensors and Actuators", Journal of Sound
and Vibration, Vol.189, N.1, pp.13-31, 1996
[30] Hagood N.W., Anderson E.H., 1991, " Simultaneous Sensing and Actuation using
piezoelectric materials", SPIE Active and Adaptive Optical Components
[31] Masarati P., 1995 "Travi piezoelettiche: modellazione ed analisi", Politecnico di Mi-
lano, Thesis
[32] Fanson J.L. and T.K.Caughey, 1987 "Positive position feedback control for large
space structures", AIAA Paper 87-0902
17
Table 1: PZ21A piezoceramic properties
Kg=m3 7915
TC C 500
s11 m =N 16:94 10,12
E 2
18
Table 3: Comparison of acceleration reduction
Experiment Simulation
MODE ANC [g] AC [g] % ANC [g] AC [g] %
Pair A
I 18.3 1.5 91.8 12.4 0.98 92.0
IV 34.6 13.9 59.9 45.1 3.02 91.1
VIII 33.3 17.1 48.6 22.2 12.1 45.5
Pair B
I 18.25 2.35 87.1 12.4 1.5 87.7
II 43.7 18.9 56.9 46.3 3.4 91.4
IX 9.3 1.3 86.3 11.9 1.34 88.1
19
Table 5: Numerical prediction of modal control eciency
Pair f [Hz] ANC [g] AC [g] % G
A 87.44 1.24 0.61 50.54 680
A 87.44 1.24 0.38 68.91 900
B 137.86 4.63 1.97 57.39 460
B 137.86 4.63 1.46 68.51 550
C 87.44 1.24 0.54 55.99 840
C 87.44 1.24 0.43 65.21 950
C 137.86 4.63 2.96 36.09 408
C 137.86 4.63 1.62 64.92 600
D 467.21 7.15 3.9 45.5 155
20
Table 7: Comparison of acceleration reduction: Numerical Vs. Experimental (Double self-
sensing actuator)
Experiment Simulation
Pair Mode G ANC [g] AC [g] % G ANC [g] AC [g] %
A I 1120 6.33 1.21 80.8 800 1.24 0.33 73.5
B I 1060 6.50 3.59 44.8 950 1.24 0.45 63.6
B II 260 8.76 4.16 52.57 260 4.63 2.41 47.9
C I 700 4.83 1.24 74.25 850 1.24 0.35 71.7
C II 600 12.10 4.45 63.22 450 4.63 1.73 62.6
C VIII 60 11.65 6.17 47.01 100 7.12 4.88 31.40
D VIII 104 10.00 3.59 64.08 110 7.15 4.03 43.7
21
C V1
VC Z1
Z2
CP V2
22
1th Mode 2th Mode
Voltage
Follower PA-85
LM-362
Sensor R pol
Actuator Polarisation
direction
23
A [dB]
Phase
Frequency [Hz]
Vcontrol C
V1 V2 LM-362
Filter
Piezoelectric
Vcontrol
INA118
Vcontrol
24
Single active device (A)
T
Pair A
100
Test
80
Sim.
% 60
40
20
0
I II VIII
Pair B
100
80
% 60
40
20
0
I II IX
Mode
80
70 Test
60 Gspe
50 Gmax
40
%
30
20
10
0
I,A II,B I,C II,C VIII,D
Mode & Piezo #
26
90
80 Experiment
70 Simulation
%
60
50
40
30
20
10
0
I I II I II VIII VIII
(A) (B) (B) (C) (C) (C) (D)
Mode & Piezo #
27