Journal of Sound and Vibration 509 (2021) 116243

Contents lists available at ScienceDirect

Journal of Sound and Vibration

journal homepage: www.elsevier.com/locate/jsvi

Hybrid feedback PID-FxLMS algorithm for active vibration control

of cantilever beam with piezoelectric stack actuator
Weiguang Li , Zhichun Yang *, Kui Li , Wei Wang
School of Aeronautics, Northwestern Polytechnical University, Xi′ an, 710072, China

A R T I C L E I N F O A B S T R A C T

Keywords: FxLMS (Filtered-x Least Mean Square) algorithm and PID (Proportional Integral Derivative)
Active vibration control controller have been widely used for AVC (Active Vibration Control). Yet, the convergence rate
Piezoelectric stack actuator and vibration suppression performance are restricted by each other for both the classical FxLMS
FxLMS algorithm
algorithm and the traditional PID controller. A hybrid PID-FxLMS algorithm which combines the
PID controller
Hybrid control algorithm
feedback FxLMS algorithm and PID controller is proposed, and applied to improve the vibration
control efficiency of piezoelectric cantilever beam. In order to conveniently install the piezo­
electric stack on the beam structures, a new piezoelectric stack actuator is designed and installed
at the root of cantilever beam. Meanwhile, a coupled finite element model of the piezoelectric
stack actuator and cantilever beam is established in ABAQUS, and the state-space model of the
coupled system is obtained. Subsequently, based on the obtained state-space model, an AVC
simulation and experimental system is built to verify the effectiveness of the proposed algorithm.
Numerical and experimental results show that convergence rate and vibration suppression per­
formance of the hybrid PID-FxLMS algorithm are much better than that of the classical FxLMS
algorithm or traditional PID controller alone, while the hybrid controller also has strong adapt­
ability and anti-noise ability.

1. Introduction

In the fields of aerospace, civil and mechanical engineering, the problem of undesired vibration of structures caused by external
environment during operation cannot be avoided or ignored. For example, the structural vibration of an aircraft not only affects the
comfort of passengers, but also causes fatigue damage to the airframe structure and increases the maintenance cost of various airborne
equipment [1]. With the development of vibration control technology, the active control of structural vibration using piezoelectric
materials has become one of the research hot spots in this field [2-8].
The piezoelectric effect was discovered by the Curie Brothers in 1880. The direct piezoelectric effect refers to the ability of certain
crystalline materials to generate electric charges proportional to the externally applied force, which can be used for the development of
sensors. According to the inverse piezoelectric effect, when an electric field parallel to the polarization direction is applied, the
piezoelectric materials will produce mechanical deformation, which is gradually developed into actuators. Piezoelectric actuators are
widely used in aerospace equipment for vibration and noise control due to their advantages of wide operating frequency, light weight,
large driving force and high electromechanical conversion efficiency. Davood Asadi and Touraj Farsadi studied the control of a Thin-
Walled Beam (TWB) wing-engine system using piezoelectric actuators to enhance the performance of aeroelastic response. The

* Corresponding author: Zhichun Yang

E-mail address: [email protected] (Z. Yang).

https://doi.org/10.1016/j.jsv.2021.116243
Received 7 July 2020; Received in revised form 19 March 2021; Accepted 25 May 2021
Available online 1 June 2021
0022-460X/© 2021 Elsevier Ltd. All rights reserved.
W. Li et al. Journal of Sound and Vibration 509 (2021) 116243

effectiveness of active control strategy based on piezoelectric composite actuator for suppressing flutter response of composite wing-
engine system was verified by numerical simulation, and the influence of position and fiber angle orientation of piezocomposite
actuator on closed-loop control performance was also studied [9]. Song and Xia introduced a harmonic control method of helicopter
fuselage vibration driven by piezoelectric stack actuators. Numerical and experimental studies under harmonic excitation were carried
out, which achieved good vibration suppression performance [10-11].
Piezoelectric active control technology for structural vibration refers to adopting appropriate active control strategies, issuing
control instructions and driving piezoelectric actuators to apply appropriate force or moment on the structure to be controlled, so as to
achieve the purpose of suppressing or eliminating vibration response. The key point of piezoelectric AVC system is how to design
reasonable control algorithm and controller. The traditional PID controller is widely used because of its simple structure, good stability
and high reliability. However, the disadvantage of the PID controller is of poor capacity in dealing with system uncertainty, i.e.,
parameter variations and external disturbance [12], while the convergence rate and steady-state error are also mutually restricted. At
the same time, the FxLMS algorithm developed based on adaptive filtering technology has become one of the hot spots in AVC al­
gorithm research due to its advantages of good convergence and strong adaptability to system uncertainty. Elliott et al. first introduced
adaptive filtering technology into the field of AVC, and achieved good vibration suppression effect by designing a controller based on
LMS algorithm [13]. Oh.JE et al. used FxLMS algorithm to carry out experimental research on AVC of a piezoelectric cantilever beam,
and verified the feasibility of the FxLMS algorithm [14]. Based on existing studies, FxLMS algorithm is gradually developed and
applied in engineering field. Niu et al. proposed an enhanced FxLMS algorithm based on online identification of secondary paths.
Ground experiments of vertical tail buffeting control under harmonic excitation and random excitation were carried out to verify the
effectiveness of the proposed algorithm [5]. Meng et al. proposed a multi-input multi-output multi-harmonic (MIMOMH) feedforward
adaptive AVC method for the helicopter fuselage structure by using piezoelectric stack actuators, which is composed of the harmonic
coefficients-identification and the MIMOMH feedforward FxLMS algorithm. The effectiveness of the proposed method was verified by
simulation and experiment of AVC [15-16]. However, the classical FxLMS algorithm has the same contradiction between the
convergence rate and vibration suppression performance [17-18]. Therefore, it is an effective approach in the development of AVC to
combine the PID controller with the FxLMS algorithm to build a hybrid controller to complement the merits of both advantages.
A hybrid PID-FxLMS algorithm which combines the feedback FxLMS algorithm and PID controller is proposed and verified in the
AVC experiments of a cantilever beam in this paper. A bending moment actuator using piezoelectric stack is designed and installed at
the root of the beam to generate an active control bending moment. Meanwhile, based on reasonable simplification, a coupled finite
element model of the piezoelectric stack actuator and cantilever beam is established in ABAQUS, and the accuracy of the established
model is verified through experiments. By discarding higher-order modes, a reduced mathematical model similar to the dynamic
characteristics of the original coupled system is extracted to represent the controlled system model, and the state-space equation is also
derived. Based on the obtained state-space model of the coupled system, the AVC simulation and experiments under harmonic
excitation with variable amplitude, phase and frequency or superimposed measuring noise are carried out, in which the effectiveness,
adaptability and anti-noise ability of the proposed PID-FxLMS controller are verified.

2. Bending Moment Piezoelectric Stack Actuator

2.1. Design of Bending Moment Actuator

The piezoelectric material actuators are divided into two types: patch type and stack type. The patch type mainly includes PZT
(Lead-Zirconate-Titanate) ceramic patch, PVDF (Polyvinylidene Fluoride) film which is a polymer, and MFC (Macro Fiber Composite)
patch. The piezoelectric patch actuators are usually pasted on the surface of the structures to be controlled, and the driving voltage are
applied to generate actuating bending moments on the structures. However, their driving moments is often small, requiring multiple
piezoelectric patch actuators to work together. At the same time, the piezoelectric stack actuator is composed of multiple piezoelectric
ceramic patches with the same polarization direction stacked in a manner of opposite polarities. In this configuration, the output
capability of the piezoelectric stack is superimposed, but the required driving voltage is not increased compared with the single-layer
piezoelectric ceramic patch. At present, one of the main research directions of piezoelectric stack actuators is to improve the per­
formance of piezoelectric actuators by designing external mechanisms coupled with the actuators [19-21]. Based on this idea, a new
type piezoelectric stack actuator suitable for AVC is developed in this paper.
Considering that in practical application of AVC for beam or plate structures, the coupling between the piezoelectric stack and the
structure to be controlled is rather difficult. When a driving voltage is applied to the piezoelectric stack, it will generate axial force due
to its inverse piezoelectric effect. However, the vibration mode of the beam or plate structure is usually bending vibration. Therefore, it
is necessary to design a host structure to transform the axial force into actuating bending moment and actuate the structure indirectly.
In this consideration, a new type bending moment piezoelectric stack actuator (BPSA) is designed and fabricated. Using the BPSA, it is
convenient for the coupling of the piezoelectric stack and the structures as well as the maintenance and replacement of the piezo­
electric stack. In the BPSA, two ends of the piezoelectric stack are blocked by metal bending stands, which provide a coupling platform
between the piezoelectric stack and the structures to be controlled, thus converting the axial force into the actuating bending moment
locally acting on the structures.
The BPSA consists of five parts, including a piezoelectric stack, a coupling base, two clamping bending stands, two fastening screws,
and two pre-compression screws. The size of the BPSA is 170 mm × 20 mm × 37 mm, and the total weight is 190 g. The relevant
parameters of the piezoelectric stack (PITM PICMA® P-840.60) are shown in Table 1. In the PBSA, the bending stands are mounted on
the coupling base by fastening screws, and the piezoelectric stack is clamped between the bending stands through pre-compression

2
W. Li et al. Journal of Sound and Vibration 509 (2021) 116243

screws. The ends of the pre-compression screws are ground flat to prevent stress concentration from damaging the piezoelectric stack.
Finally, the coupling base is glued and coupled to the structure to be controlled through epoxy resin (when the surface of the structure
is relatively flat, the clamping bending stands can also be directly glued to the structure, and this method will be used in the subsequent
research of this paper). The prototype of the BPSA is shown in Fig. 1.

2.2. Modeling of the BPSA

When the driving voltage V is applied on the piezoelectric stack and the axial block force p is applied by the pre-compression screws,
the induced deformation along the thickness direction of each piezoelectric wafer of the stack can be obtained by the following
equation [22]:
/
E
δ3 = S33 tp As + d33 V (1)

Where SE33 is the flexible coefficient of piezoelectric material under a constant driving voltage, d33 is the piezoelectric strain constant, t
is the thickness of a single wafer, As is the area of the wafer which is also the cross-section area of the stack.
Each piezoelectric wafer in the piezoelectric stack is pasted together with opposite polarities, which is equivalent to electrical
parallel connection and mechanical series connection. Then the deformation of the entire stack can be described as:
( E / )
δs = nδ3 = n S33 tp As + d33 V (2)

where n is the number of the piezoelectric wafers of the stack. The strain on the cross section of the stack can be obtained:
/
δ n ( E / ) d33 V
εs = s = E
S33 tp As + d33 V = S33 p As + (3)
ls n ⋅ t t

Where ls = n ⋅ t, ls is the length of the stack.

According to Hooke’s law, the main stress in the cross section of the piezoelectric stack can be described as σ s = Eps εs , where Eps is
the elastic modulus of the piezoelectric material. Defining Fa as the internal axial force of the stack, it can be obtained that:
Fa
σs = (4)
As
From Eq. (3) and Eq. (4), the axial internal forces of the piezoelectric stack Fa can be obtained:
( )
d33
E
Fa = Eps εs As = Eps S33 p + As V (5)
t
Since both ends of the piezoelectric stack are blocked by the bending stands, the constraint reaction force can be described as Fs=Fa,
the expression of the bending moment of the BPSA can be expressed as:
( )
d33
E
Ma (t) = Fs h = Eps h S33 p + As V(t) (6)
t

where h is the height of BPSA, which is defined as the distance from the axis of the stack to the bottom of BPSA as shown in Fig. 2.
The proportional relationship among the actuating moment, the installing height of the BPSA, the pre-compression force and the
driving voltage is indicated in Eq. (6) [23].

2.3. Finite element modeling of Piezoelectric Cantilever Beam

To design the AVC system of any structure, the mathematical model of controlled system is required. For complex structures, it is
difficult to theoretically establish an accurate model of the structure to be controlled, however, it is relatively simpler to model such
systems in finite element software [24]. In this paper, a finite element model of piezoelectric cantilever beam is established in
ABAQUS. By discarding higher-order modes, a reduced mathematical model similar to the dynamic characteristics of the original
structure to be controlled is extracted to represent the system model, thereby reducing the calculation time.

Table 1
Parameters of piezoelectric stack
Parameter Value

Length × width × height/(mm × mm × mm) 122 × 12 × 12

Maximum dynamic displacement/µm 90
Maximum output force/N 1000
Maximum operation voltage/V 100
Operating temperature/◦ C -20~80
Elastic compliance coefficient/(m2/N) 16.1 × 10− 12
Piezoelectric charge coefficient/(C/N) 4 × 10− 10

3
W. Li et al. Journal of Sound and Vibration 509 (2021) 116243

Fig. 1. The prototype of BPSA. (a) Photo. (b) Design diagram (1- Coupling base, 2- Clamping bending stands, 3- Fastening screws, 4- Pre-
compression screws, 5- Piezoelectric stack).

Fig. 2. The working principle diagram of BPSA.

The constitutive equations of a general piezoelectric material are:

T = cE S − eE
(7)
D = eT S + εS E

where T and S are the stress (N/m2) matrix and strain matrix, respectively, D and E are the electric displacement (C/m2) and the
electric field (V/m), respectively, cE is the matrix of elastic constants under constant electric field (Hooke’s tensor), e is the piezo­
electric constants matrix (C/m2) and εS is the permittivity matrix under constant strain.
The finite element formulation of a piezoelectric continuum can be derived from Hamilton’s principle. According to Hamilton’s
principle:
∫ t2
[δL + δWnc ]dt = 0 (8)
t1

Among them, L = T∗ + We∗ is the Lagrangian of the coupled system, and δWnc is the virtual work of nonconservative external forces
and applied currents. When integrating along any possible virtual displacement path from t1 to t2, the corresponding initial and final
state are known quantities.
∫
1
T∗ = ρẋT ẋdΩ (9)
2 Ω

is the kinetic energy (ρ is the density) and

∫
1 ( T T )
We∗ = 2S e E + ET εT E − ST cS dΩ (10)
2 Ω

is the potential energy of the conservative system [25].

Therefore, the Lagrangian of a structure involving a finite number of discrete piezoelectric actuators can be written in the general
form:
1 1 1
L = T ∗ + We∗ = ẋT Mẋ − xT Kxx x + φT Kφφ φ + φT Kφx x (11)
2 2 2
In these equations, M is the mass matrix, Kxx is the stiffness matrix (including the mechanical part of the actuators with short
circuited electrical boundary conditions), Kφφ is the matrix of capacitance of the transducers (for fixed displacements), Kφx is the
coupling matrix of piezoelectric properties, relating the mechanical and electrical variables, x is the physical coordinate displacement
of the structure, and φ is the electric potential of the piezoelectric material.
The piezoelectric AVC system contains two parts: the structure to be controlled and the piezoelectric actuator. The structure to be
controlled is an elastic deformable body, and its DOF (Degree of Freedom) is its generalized displacement; while the DOF of

4
W. Li et al. Journal of Sound and Vibration 509 (2021) 116243

piezoelectric actuator is in addition to displacement and also includes the electric displacement information, which respectively
corresponding to the displacement vector and the electric potential of the node in finite element model. Therefore, the DOF of the
coupled system can be divided into two parts: the regular displacement DOF, the piezoelectric displacement and potential DOF.
For the structure to be controlled, the element mass matrix and stiffness matrix are as follows:
⎡ ⎤ ⎡ ⎤
m1 k11 ⋯ k1n
Mi = ⎣ ⋱ ⎦, Ki = ⎣ ⋱ ⋮ ⎦ (12)
mn SYS. knn

For piezoelectric actuators:

⎡ ⎤ ⎡ ⎤
m1 k11 ⋯ k1m k1φ
⎢ ⋱ ⎥ ⎢ ⋱ ⋮ ⋮ ⎥
e
Mi = ⎣⎢ ⎥ e ⎢
⎦, Ki = ⎣
⎥ (13)
mm kmm kmφ ⎦
0 SYS. kφφ

The element DOF are:

⎧ ⎫
⎧ ⎫ ⎪ x1 ⎪
⎨ x1 ⎬ ⎪
⎨ ⎪ ⎬
⋮
⋮ , (14)
⎩ ⎭ ⎪ ⎪ x ⎪
xn ⎩ ⎪ m
⎭
φ

After the element matrix is assembled into a total matrix, the displacement DOF are classified and combined, and the node potential
DOF corresponding to the piezoelectric actuators are integrated under the total DOF column vector. The results are as follows:
ẍs
⎡ ⎤⎧
Ms ⎨ ⎫ ⎡ ẍe ⎤⎧ ⎫ ⎧ ⎫
⎣ Me ⎦ ⎬ Kss Kse Ksφ ⎨ xs ⎬ ⎨ 0 ⎬ (15)
⎩ φ̈ + ⎣ Kee Keφ ⎦ xe = 0
0
⎭ ⎩ ⎭ ⎩ ⎭
SYS. Kφφ φ 0

where Ms is the mass matrix of the structure to be controlled, and Me is the piezoelectric actuator mass matrix; Kss and Kee are the
displacement freedom stiffness matrices of these; Ksφ and Keφ are the influence matrices of the piezoelectric material on the
displacement of the structure to be controlled and the piezoelectric actuator; Kφφ is the electric potential stiffness matrix, which is
understood as the equivalent capacitance of the piezoelectric material from the electrical point of view.
Eq. (15) can be expressed as:
[ ]{ } [ ẍ ]{ } { }
Mxx 0
φ̈ +
Kxx Kxφ x
=
0 (16)
0 0 Kxφ Kφφ φ 0

The resulting dynamic equations obtained from Eq. (11) read [25]:

Mẍ + Kxx x − KTφx φ = f (17)

Kφx x + Kφφ φ = Q (18)

Where f is the vector of applied force, and Q is the vector of electric charges appearing on the electrodes.
Assuming that the piezoelectric actuator has no applied force, f = 0, and the piezoelectric actuator is usually driven by an external
control voltage in actual working conditions, while the electric potential φ at each node of the piezoelectric material can be obtained by
interpolation. Then Eq. (17) can be written as:

Mẍ + Kxx x = KTφx φ (19)

From Eq. (19), taking the system acceleration response vector ẍ as the output, the state-space equation of the piezoelectric AVC
system can be derived as:
⎧ [ ]
⎪ [ ] 0
⎪
⎪ 0 I
⎨ ẋ = x + φ
− M− 1 Kxx 0 M− 1 KTφx (20)
⎪
⎪ [ ]
⎪
⎩ y = − M− 1 Kxx 0 x + M− 1 KT φ
φx

If the system damping is considered, Eq. (20) can be further written as:

5
W. Li et al. Journal of Sound and Vibration 509 (2021) 116243

⎧ [ ]
⎪ [ ] 0
⎪
⎪ 0 I
⎨ ẋ = x+ φ
− M− 1 Kxx − M− 1 Cxx M− 1 KTφx (21)
⎪
⎪ [ ]
⎪
⎩ y = − M− 1 Kxx − M− 1 Cxx x + M− 1 KTφx φ

Similarly, if the system displacement response vector x is taken as the output, the output equation can be derived as:
y = [ 1 0 ]x (22)
So far, the electromechanical coupling phenomenon in AVC system is transformed into the transmission characteristics of the
structure to be controlled, and the state-space model of the coupled system is obtained for subsequent AVC system.

3. Hybrid PID-FxLMS Controller Design

Traditional PID controller is widely used in the field of AVC due to its simple structure, good stability, high reliability and
convenient parameter adjustment. According to the residual vibration response of the controlled system, PID controller uses pro­
portional, integral and differential algorithms to calculate the required control and drive the actuator to suppress the vibration
response.
The classical FxLMS algorithm based on FIR (Finite Impulse Response) filter has the advantages of good convergence, strong
tracking capability, and is also suitable for suppressing the vibration response of the systems with uncertainties. The FxLMS algorithm
can be divided into two types: feedforward FxLMS algorithm and feedback FxLMS algorithm, mainly including FIR digital filter and
LMS adaptive algorithm. The coefficients of digital filter are adjusted continuously by using the adaptive algorithm to improve the
filter performance.

3.1. Feedback FxLMS Algorithm

In the feedforward FxLMS control process, it is necessary to know the reference signal related to the external disturbance in advance
[26]. However, in actual vibration control, it is generally difficult to obtain the external disturbance in real time. Therefore, for this
case, it is necessary to construct a feedback FxLMS algorithm based on the reference signal self-extraction method [27].
The feedback FxLMS algorithm only needs the residual vibration response signal to drive the control system. The residual vibration
response signal of closed-loop system is extracted and used to construct a reference signal based on the structure and algorithm of the
controller. This signal has a correlation with the external excitation signal, and then it is input to the controller, so that the additional
reference signal is not required. The feedback FxLMS algorithm structure based on the reference signal reconstruction is shown in
Fig. 3.
In Fig. 3, plant represents the controlled object model which is called the primary path, x(k) is the reference signal at time k. d(k) is
the open-loop vibration response of the structure at time k, y(k) is the control voltage signal at time k, the path H(z) from the actuator to
the sensor is called the secondary path, H(z)
̂ is the model for the secondary path, and ̂ y (k) is the structural vibration response caused by
the control signal passing through the secondary path, and e(k) is the residual vibration response of the structure. If the secondary path
model is accurate enough, then ̂ d(k) is a better estimate of d(k). For a linear structure system, the external excitation signal is linearly
related to the open-loop vibration response of the structure d(k). The reference signal ̂ d(k) reconstructed from the residual vibration
response e(k) and the response of secondary path ̂ y (k) is also linearly related to the external excitation signal.
In summary, the feedback FxLMS algorithm based on reference signal reconstruction can be summarized as follows:

̂ ̂ T (k)y(k)
d(k) = e(k) + H (23)

y(k) = XT (k)W(k) (24)

Fig. 3. Feedback FxLMS algorithm based on reference signal reconstruction

6
W. Li et al. Journal of Sound and Vibration 509 (2021) 116243

e(k) = ̂
d(k) − y(k) (25)

W(k + 1) = W(k) + 2μe(k)X(k) (26)

where X(k) is the filter input signal vector with length L,W(k) is the weight coefficients of the adaptive filter at time k, μ is the step size.
Eq. (23) represents the process of reference signal reconstruction.

3.2. Hybrid feedback PID-FxLMS Control Algorithm

On site vibration control applications, the traditional PID controller has a fast convergence rate, but its vibration suppression
performance is often poor. However, FxLMS algorithm will generate the corresponding deviation due to the inevitable interference at
the input of the filter. The greater the interference, the greater the resulting imbalance. Reducing the step size can reduce the steady-
state imbalance of the algorithm, and further improve the suppression performance of structural vibration response, while a smaller
step size will reduce the convergence rate. However, when the step size is increased, the convergence rate of the algorithm is
accelerated, while the convergence accuracy is obviously reduced, which may even lead to divergence of the control system. Therefore,
the FxLMS algorithm conflicts with the adjustment of step size in terms of convergence rate and convergence accuracy. Considering
that the PID controller and the FxLMS controller are quite different in form, however, when judging from the input and output signals
of the controllers, they both take the error signal as the input, and the control output is obtained through calculation. Therefore, a
hybrid feedback PID-FxLMS algorithm combining the feedback FxLMS algorithm and the PID controller is proposed in the current
study, which aims to give full play to the respective advantages of the two algorithms and improve the AVC effect.
The diagram of hybrid feedback PID-FxLMS algorithm is shown in Fig. 4, where ULMS (k) and UPID (k) are the control voltage signals
output by FxLMS controller and PID controller at time k, respectively, which are summed to obtain a total control signal U(k) to drive
the actuator, and U(k)
̂ is the structural vibration response caused by the total control signal passing through the secondary path.
The hybrid PID-FxLMS algorithm is summarized as follows:

̂ ̂ T (k)U(k)
d(k) = e(k) + H (27)

ULMS (k) = XT (k)W(k) (28)

1 Ns
UPID (k) = Kp ⋅ e(k) + Ki ⋅ e(k) ⋅ + Kd ⋅ e(k) ⋅ (29)
s s+N

U(k) = ULMS (k) + UPID (k) (30)

e(k) = ̂
d(k) − U(k) (31)

W(k + 1) = W(k) + 2μe(k)X(k) (32)

where Kp , Ki , Kd and N are proportional coefficient, integral coefficient, differential coefficient and filter coefficient of PID controller,
respectively, and μ is the step size of hybrid FxLMS algorithm.

4. AVC Simulation of Piezoelectric Cantilever Beam

4.1. Finite element analysis of piezoelectric cantilever beam model

In this paper, an aluminum alloy cantilever beam is taken as the research object. The relevant parameters of the cantilever beam are
shown in Table 2. Based on the maximum modal strain energy criterion [28], that is, the placement of the actuator at the location

Fig. 4. Hybrid feedback PID-FxLMS algorithm

7
W. Li et al. Journal of Sound and Vibration 509 (2021) 116243

Table 2
Parameters of cantilever beam
Parameter Value

Length × width × height/(mm × mm × mm) 900 × 20 × 5

density/(Kg/m3) 2700
Elastic Modulus/GPa 72
Poisson’s ratio 0.33

where the target modal strain energy of the object to be controlled is maximum can effectively suppress the vibration response of this
mode [29]. The BPSA is installed at a distance of 50 mm from the root of the beam to establish the piezoelectric cantilever beam model.
The parameters of piezoelectric stack are shown in Table 3.
With the aid of ABAQUS software, a finite element model of the piezoelectric cantilever beam as shown in Fig. 5 is established.
Among them, the piezoelectric stack actuator adopts the C3D8E piezoelectric hexahedral element (8 nodes, each node has 3 trans­
lational DOF and 1 piezoelectric related DOF), which is equivalent to a uniform piezoelectric cylinder model. 122 grids are divided in
the axial direction, so the thickness of each layer is 1 mm. The cantilever beam and the coupling structures of piezoelectric actuator are
modeled by C3D8R linear hexahedral element (8 nodes, each node has 3 translational DOF). It is assumed that the piezoelectric
actuator and the cantilever beam are well coupled, and the influence of the epoxy resin layer can be ignored. The clamping bending
stands of actuator and the cantilever beam are connected by tie. Then, the junction between the left end of the actuator and the bending
stand is connected by MPC Beam, while the right end is connected by MPC Pin. Meanwhile, by constraining all mechanical DOF of the
root section of the cantilever beam, the fixed-end boundary condition is realized. Modal analysis adopts the Linear perturbation,
Frequency Lanczos method to solve the natural frequency and formation of the piezoelectric cantilever beam. Static analysis calculates
the static displacement response of the tip of the piezoelectric cantilever beam. The electrode at one end of the piezoelectric stack
actuator is set to 0 potential, while the other end is applied with a progressive potential of 122 × 0~100 V as the electrical boundary,
that is, 0~100 V driving voltage is applied to both ends of the actuator. Frequency response analysis uses Steady-state dynamics, Modal
method to calculate the frequency response function of primary path and secondary path of the piezoelectric cantilever beam model,
and the first and second modal damping ratios are set as 1.6% and 0.7%, respectively.

4.2. Experimental analysis of piezoelectric cantilever beam model

This section verifies the established finite element model of the piezoelectric cantilever beam through experiment. The location of
the acceleration sensor R (at the tip of the beam), force sensor F (on the top of the exciter rod) and the BPSA are shown in Fig. 6. Where
Δ is the excitation position of the exciter (at a distance of 25 mm from the root of the beam), which is used to exert the external
excitation. The details of the equipment used in the experiment are shown in Table 4.

4.2.1. Frequency response characteristic experiment

According to the FxLMS algorithm, the primary path describes the dynamic characteristics between the external disturbance and
the open-loop vibration response of the structure d(k). At the same time, in order to suppress the vibration response of observation
point R, the secondary path is used to generate opposite vibration response ̂y (k) or U(k)
̂ to reduce them. Therefore, the accuracy of
primary path and secondary path models directly affect the control effect and system stability.

4.2.1.1. Primary path. First, the dSPACE signal generator module outputs a sinusoidal frequency sweep signal, which is amplified by
the exciter power amplifier, and then applied to the electromagnetic exciter to excite the piezoelectric cantilever beam with the
frequency range of 0.1~50 Hz. The acceleration sensor and displacement sensor are used to collect the acceleration and displacement
response of the observation point R, and the excitation force measured by the force sensor is recorded at the same time. Taking the
excitation force as the input, and the acceleration and displacement response as the output, the frequency response analyses are carried
out in MATLAB and compared with the steady-state analysis results of the finite element model in Section 4.1. As shown in the Fig. 7.

4.2.1.2. Secondary path. Secondly, the same signal is amplified by the piezoelectric power amplifier, and then applied to the BPSA.
Taking the driving voltage of the piezoelectric actuator as the input, and the acceleration and displacement response as the output.
Then, the frequency response analyses are performed and compared with the steady-state analysis results of the finite element model. It

Table 3
Parameters of piezoelectric stack
Parameter Value

Length × width × height/(mm × mm × mm) 122 × 12 × 12

density/(Kg/m3) 5200
Elastic Modulus/GPa 5
Poisson’s ratio 0.3
piezoelectric strain constant d33/(C/N) 1.12e-8

8
W. Li et al. Journal of Sound and Vibration 509 (2021) 116243

Fig. 5. Finite element model of piezoelectric cantilever beam

Fig. 6. Diagram of experimental system of AVC for the cantilever beam

Table 4
The equipment used in the experiment
Equipment name Equipment type Remarks

Acceleration sensor PCB 333B30 Sensitivity: 100 mV/g

Force sensor PCB 208C02 Sensitivity: 11.28 mV/N
Displacement sensor system KEYENCE150 Sensitivity: 1000 mV/m
Power supply LPS-350 Output voltage: 0~24 V
Acceleration signal conditioner PCB 482C Output range: -10~10 V
Real-time control system dSPACEMicroLabBox Input and output range: -10~10 V
Output range: 0~120 V
Piezoelectric power amplifier PA-V-M4
Gain: 1~12
Exciter power amplifier MB Modal SL500VCF Frequency Range: 10 Hz~20 KHz
Electromagnetic exciter MB Modal 50 Frequency Range: 5000 Hz

Fig. 7. Primary path frequency response characteristic comparison results. (a) Acceleration. (b) Displacement.

9
W. Li et al. Journal of Sound and Vibration 509 (2021) 116243

is shown in the Fig. 8.

It can be seen from the above comparison results that the first and second flexural modes of the piezoelectric cantilever beam are
6.6 Hz and 31.7 Hz, respectively. Then, the dynamic characteristics of the finite element model are in good agreement with the actual
model, which can be used for subsequent controller design.

4.2.2. Static experiment

In the static experiment, the driving voltage as shown in the Fig. 9 is applied to the BPSA, and the displacement response of point R
is collected by the displacement sensor. The experiment obtains the static deflection values of the cantilever beam under 11 working
conditions including 0 V, 10 V, ..., 90 V, 100 V, etc. Fig. 10 shows the comparison curves between the static analysis of the piezoelectric
cantilever beam model and the experiment results. The simulation results are basically consistent with the experiment, and the error is
less than 5.3 %.

4.3. AVC Simulation of Piezoelectric Cantilever Beam

In this section, the advantages of the proposed hybrid feedback PID-FxLMS algorithm over the feedback FxLMS algorithm or the
traditional PID controller is verified through AVC simulation.

4.3.1. AVC Simulation under Harmonic Excitation

The external excitation signal adopts a 6.5 Hz sinusoidal signal, which is near the first-order bending natural frequency of the
piezoelectric cantilever beam. And the state-space model of the secondary path is obtained from Section 2.3. When the vibration
response of the structure reaches a stable state at 10 s, the controller is turned on, and the vibration control effects of each controller are
compared and studied. The PID controller parameters Kp , Ki , Kd and N are obtained by particle swarm optimization algorithm [30] and
set to 0.6209, 4.0918, 0.0073 and 1000, respectively. The feedback FxLMS controller takes the order of the adaptive filter as 32 and the
step size as 1E-6, and the feedback PID-FxLMS controller parameters are the same as the above settings.
In Fig. 11(a) and 11(b), numerical results show that the PID controller converges with 5 seconds, while the final vibration response
is decreased by 56.6%. At the same time, the feedback FxLMS controller based on reference signal reconstruction still achieves 93.6%
vibration suppression performance when accurate reference signals cannot be obtained. However, its convergence rate is slow in the
initial stage of the control process, which cannot achieve the purpose of rapidly reducing the structural vibration level. Meanwhile, due
to the error in the secondary path model, the residual vibration response cannot be eliminated. By superposing the control voltage of
the PID controller and the feedback FxLMS controller, a hybrid feedback PID-FxLMS controller is constructed. It can be seen from
Fig. 11(c) and 11(d) that the output voltage of the hybrid controller is larger than the PID controller in the whole control process and
much larger than the feedback FxLMS controller in the initial stage. The time required for each controller when the vibration sup­
pression performance reaches 20%, 40%, 60%, and 80% for the first time and the steady-state vibration suppression performance of
each controller are shown in Table 5.
As shown in Table 5, the time required for the hybrid controller to reach the same vibration suppression performance is basically
the same as that of the PID controller, but the final vibration suppression performance is 37.5% higher than that of the PID controller.
At the same time, when the hybrid controller is compared with the FxLMS controller, it takes less time for the hybrid controller to
achieve the same vibration suppression performance. This is particularly obvious at the beginning of the control process, and the final
vibration suppression performance is basically the same as the FxLMS controller. In other words, the hybrid controller combines the
advantages of faster convergence rate of the PID controller and higher steady-state vibration suppression performance of the FxLMS
controller, hence to achieve better control effect. Thus, the advantages of the hybrid controller proposed in this paper compared to the
PID controller or the FxLMS controller are verified.

Fig. 8. Secondary path frequency response characteristic comparison results. (a) Acceleration. (b) Displacement.

10
W. Li et al. Journal of Sound and Vibration 509 (2021) 116243

Fig. 9. Driving voltage of the BPSA

Fig. 10. Contrast curves of static experiment

4.3.2. AVC Simulation under Variable Harmonic Excitation

The structural vibration environment of the practical situation is more complicated, usually manifested as changes in amplitude,
phase or frequency of the exciting force. Therefore, in order to further investigate the adaptability of the proposed algorithm to
variable external disturbances, the AVC under harmonic excitation with variable amplitude, phase or frequency are analyzed in this
section.
The system starts to use the 6.5 Hz sinusoidal signal as external excitation input, and the controller is turned on at the 10th second.
When the vibration response is suppressed, the amplitude of the external excitation increases to 1.5 times the original amplitude at the
100th second, the phase increases by π/6 at the 110th second, and the frequency is increased by 2% (that is, 6.63 Hz) at the 120th
second. The controller parameters are consistent with those in Section 4.3.1, and the displacement response of observation point R is
shown in Fig. 12(a), while the output voltage of the feedback PID-FxLMS controller is shown in Fig. 12(b). From Fig. 12, it can be seen
that when the amplitude, phase or frequency of the external excitation is changed, the structural vibration response increases rapidly.
At this time, the controller quickly adjusts the output voltage according to the changes of the vibration response and the response
decays again, which proves that the hybrid controller has strong adaptability.

4.3.3. AVC Simulation under Harmonic Excitation with Measuring Noise

As we all know, the real structural vibration response is generally accompanied by noise, while the signal acquisition system is
inevitably interfered by measuring noise, which greatly reduces the performance of the control system. In severe cases, it may even
cause the control to diverge. Thus, the anti-noise ability of the controller must be checked, and the AVC under harmonic excitation with
measuring noise are studied in this section.
The 6.5 Hz sine signal is used as the external excitation input, while the zero-mean Gaussian white noise is superimposed on the
residual signal of the vibration response collected by the sensor, and the variance of the superimposed noise σ2 is set to 0.25. The
controller is still turned on at the 10th second, and the controller parameters are consistent with those in Section 4.3.1. The
displacement response of point R is shown in Fig. 13(a), and the comparison of PSD curves are shown in Fig. 13(c). From Fig. 13, the

11
W. Li et al. Journal of Sound and Vibration 509 (2021) 116243

Fig. 11. Comparison of simulation results of the feedback PID-FxLMS controller. (a) Displacement response at point R. (b) Peak displacement
response. (c) and (d) Output voltage.

Table 5
AVC simulation comparison results
Time 20% 40% 50% 60% 80% Steady-state performance

Performance
Controller
Hybrid PID-FxLMS 0.73 s 0.89 s 1.05 s 6.89 s 32.28 s 94.1%
PID 0.73 s 0.89 s 3.82 s - - 56.6%
FxLMS 5.04 s 9.04 s 11.81 s 15.66 s 38.14 s 93.6%

Fig. 12. Adaptability examination results of the feedback PID-FxLMS controller. (a) Displacement response at point R. (b) Output voltage of the
feedback PID-FxLMS controller.

12
W. Li et al. Journal of Sound and Vibration 509 (2021) 116243

Fig. 13. Anti-noise ability test results of the feedback PID-FxLMS controller. (a) Displacement response at point R with measuring noise. (b) Output
voltage of the feedback PID-FxLMS controller. (c) Displacement response PSD contrast curves.

displacement response is finally decreased by 89.5%. Meanwhile, the PSD peak value is reduced by 48.8 dB and the RMS (Root Mean
Square) of the displacement response is decreased by 93.5%, which indicates that the hybrid controller has strong anti-noise ability.

5. AVC Experiment of Piezoelectric Cantilever Beam

This section verifies the effectiveness of the proposed hybrid feedback PID-FxLMS algorithm through AVC experiments. As shown
in Fig. 14, the cantilever beam AVC experimental system consists of the experimental model, the external excitation device and the
active control system. The external excitation device is used to excite the cantilever beam structure, and the residual vibration response
of the beam structure is obtained by the displacement sensor. Then, enter it into the vibration controller through the dSPACE input
board, and the control voltage calculated by the controller is output through the dSPACE output board, which is amplified by the
power amplifier to drive the BPSA. Thus, realizing the active control of structural vibration response. The detailed information of the
equipment used is shown in Table 4.

5.1. AVC Experiment under Harmonic Excitation

The experimental conditions of AVC are the same as the simulation in Section 4.3.1, and the AVC experiment results are shown in
Fig. 15. The “output voltage” of the controllers mentioned in this section is the actual driving voltage applied to the piezoelectric
actuator after the control voltage signals of the control algorithms are passed through the piezoelectric power amplifier. As can be seen
from Fig. 15(a) and 15(b), the similar conclusions as the simulation are verified. Compared with the PID controller, the feedback PID-
FxLMS controller has better performance in the whole control process, and finally the displacement response suppression performance
is increased by 36.2%, which reaches 93.9%. When compared with the feedback FxLMS controller, the convergence rate in the first 20
seconds is obviously accelerated, and the structural vibration response at the initial stage is effectively decreased, while the final
suppression performance is increased by 1.4%. However, due to the inevitable time delay phenomenon in the control system, the error
suppression effect of the controller always lags behind the change of the residual vibration response signal. It has a greater impact on
the tracking rate of the PID controller, resulting in a slightly weaker control effect of the hybrid controller than the feedback FxLMS
controller during the 29th second to the 108th second. The output voltage of each controller is shown in Fig. 15(c). As we can see, the
output voltage of the hybrid controller is much larger than the FxLMS controller in the initial stage of control process, which quickly
reduces the structural vibration level, and is smaller than the FxLMS controller in final stage, reducing the control energy required.
Comparison of displacement response PSD curves at point R are shown in Fig. 15(d), and the PSD peak value of the first-order bending

13
W. Li et al. Journal of Sound and Vibration 509 (2021) 116243

Fig. 14. Set-up of AVC experiment of the cantilever beam

Fig. 15. Comparison of experimental results of the feedback PID-FxLMS controller. (a) Displacement response at point R. (b) Peak displacement
response. (c) Output voltage. (d) Displacement response PSD contrast curves.

mode are decreased by 14.9 dB, 45.2 dB and 49.1 dB, respectively.
Fig. 16 shows the output voltage composition of the feedback PID-FxLMS controller. It can be seen that in the initial stage of control
process, the output voltage of the PID controller is significantly large, which greatly improves the convergence rate of the hybrid
controller. As the output voltage of the feedback FxLMS controller increases, the convergence accuracy of the hybrid controller is
further improved, which enhances the final vibration suppression performance and adaptability of the hybrid controller.

14
W. Li et al. Journal of Sound and Vibration 509 (2021) 116243

Fig. 16. Output voltage composition of the feedback PID-FxLMS controller

Based on the above experimental results, the hybrid feedback PID-FxLMS algorithm proposed by combining feedback FxLMS al­
gorithm with traditional PID controller gives full play to the respective advantages of the two algorithms, and improves the AVC effect
of the piezoelectric cantilever beam.

5.2. AVC Experiment under Variable Harmonic Excitation

In this section, the adaptability of the proposed feedback PID-FxLMS controller is tested. Firstly, the same condition as the
simulation in Section 4.3.2 is verified. The displacement response of observation point R under variable harmonic excitation is shown
in Fig. 17(a), and the output voltage of the feedback PID-FxLMS controller is shown in Fig. 17(b). From Fig. 17, it can be seen that
whether the amplitude or phase changes, the feedback PID-FxLMS controller can quickly converge within 3 seconds. Meanwhile, the
increase of the excitation amplitude leads to a corresponding increase in the output voltage of the controller, due to the increase in
required control energy. When the phase changes, the output voltage begins to increase slightly, but it quickly decreases to the
previous voltage level, for the phase change does not change the required control energy. For the case of frequency changes, the
feedback PID-FxLMS controller can still maintain a good control effect, and the vibration suppression performance is gradually
improved with the adaptive parameter adjustment of the controller.
In order to further verify the adaptive performance of the proposed controller, the amplitude of external excitation is increased by
1.8 times, the phase is increased by π/4, and the frequency is increased by 3%, that is, 6.695Hz. The other experimental conditions
remain the same. The experimental results are shown in Fig. 18. As can be seen that the hybrid controller still maintains good vibration
suppression performance, which indicates that the proposed hybrid feedback PID-FxLMS algorithm has strong adaptability.

5.3. AVC Experiment under Harmonic Excitation with Measuring Noise

In order to verify the anti-noise ability of the proposed feedback PID-FxLMS controller, this section superimposes the zero-mean
Gaussian white noise on the residual response collected by the displacement sensor to simulate measuring noise. The variance of
the superimposed noise σ 2 is set to 0.25, 1, and 4, respectively, and the other conditions are consistent with those in Section 4.3.3 in the
AVC experiments.
The displacement response of point R with measuring noise is shown in Fig. 19, and the comparison of displacement response PSD
curves are shown in Fig. 20. It can be seen that, for the case with measuring noise, the feedback PID-FxLMS controller still has a good
control effect on harmonic vibration response. The peak value of the PSD curves is decreased by 48.2 dB, 48.3 dB and 47.9 dB, and the
RMS of the displacement response are decreased by 93.2%, 92.9% and 92.1%, respectively, which indicates that the proposed hybrid

Fig. 17. Adaptability experimental results of the feedback PID-FxLMS controller (the amplitude of external excitation increases by 1.5 times, phase
changes by π/6, frequency increases by 2%, that is, 6.63 Hz). (a) Displacement response at point R. (b) Output voltage of the hybrid controller.

15
W. Li et al. Journal of Sound and Vibration 509 (2021) 116243

Fig. 18. Adaptability experimental results of the feedback PID-FxLMS controller (the amplitude of external excitation increases by 1.8 times, phase
changes by π/4, frequency increases by 3%, that is, 6.695 Hz). (a) Displacement response at point R. (b) Output voltage of the hybrid controller.

Fig. 19. Anti-noise ability experimental results of the feedback PID-FxLMS controller: displacement response at point R with measuring noise.

Fig. 20. Anti-noise ability experimental results of the feedback PID-FxLMS controller: displacement response PSD contrast curves with
measuring noise.

feedback PID-FxLMS algorithm has strong anti-noise ability.

6. Conclusion

1) In this paper, a bending moment piezoelectric stack actuator (BPSA) is designed, which greatly facilitates the coupling of piezo­
electric stack with beam or plate structures as well as the maintenance and replacement of piezoelectric stack. At the same time, a
coupled finite element model of the piezoelectric stack actuator and cantilever beam is established in ABAQUS. By discarding
higher-order modes, a reduced mathematical model similar to the dynamic characteristics of the original controlled system is
extracted to represent the coupled system, and the accuracy of the established model is verified through experiments. Based on the
obtained state-space model, the AVC simulation and experiments are carried out and the results show that BPSA can be used as an
efficient actuator in the field of AVC.
2) A hybrid PID-FxLMS algorithm is proposed, which combines the feedback FxLMS algorithm based on reference signal recon­
struction and the traditional PID controller. Subsequently, an AVC simulation and experimental system of a piezoelectric cantilever
beam is built to verify the effectiveness of the proposed algorithm. AVC experimental results under harmonic excitation show that

16
W. Li et al. Journal of Sound and Vibration 509 (2021) 116243

within the first 20 seconds of the control process, the hybrid feedback PID-FxLMS controller has a much faster convergence rate
than the classical feedback FxLMS controller, which effectively suppresses the vibration response at the initial stage. Meanwhile,
the vibration suppression performance of the displacement response finally reaches 93.9%, and the PSD peak value of the first-
order bending mode are decreased by 49.1 dB. Compared with the traditional PID controller, the hybrid controller has better
control performance in the entire control process, and the final control effect is improved by 36.2%. Therefore, the experimental
research verifies the superiority of the hybrid feedback PID-FxLMS algorithm. At the same time, for the case of AVC experiment
under variable harmonic excitation, it can be seen that the hybrid controller has strong adaptability, which means that when the
amplitude, phase or frequency of the external excitation changes to a certain extent, the structural vibration response can be
quickly attenuated by adjusting the control voltage accordingly. When the zero-mean Gaussian white noise is superimposed on the
residual response of the structure to be controlled, while the variance of the superimposed noise σ2 is set to 0.25, 1, and 4,
respectively, the RMS of the displacement response are decreased by 93.2%, 92.9% and 92.1%, which indicates that the proposed
hybrid feedback PID-FxLMS algorithm also has strong anti-noise ability. Therefore, the hybrid PID-FxLMS algorithm proposed in
this study can be effectively applied to the field of AVC with uncertainty.

CRediT authorship contribution statement

Weiguang Li: Conceptualization, Methodology, Software, Validation, Formal analysis, Investigation, Writing - original draft,
Visualization. Zhichun Yang: Resources, Writing - review & editing, Supervision, Project administration, Funding acquisition. Kui Li:
Methodology, Software. Wei Wang: Data curation, Writing - review & editing.

Declaration of Competing Interest

None.

References

[1] RG. Loewy, Helicopter vibrations: a technological perspective, Journal of the American Helicopter Society 29 (4) (1984) 4–30 (27).
[2] A Yousefi-Koma, Sasiadek JZ, G. Vukovich, LQG control of flexible structures using piezoelements, in: AIAA Guidance, Navigation and Control Conference,
Scottsdale, AZ, 1994, pp. 893–900.
[3] CMA Vasques, J.D. Rodrigues, Active vibration control of smart piezoelectric beams: Comparison of classical and optimal feedback control strategies, Computers
& Structures 84 (22) (2006) 1402–1414.
[4] V Sethi, G Song, Multimodal Vibration Control of a Flexible Structure Using Piezoceramic Sensor and Actuator, Journal of Intelligent Material Systems and
Structures 19 (5) (2008) 573–582.
[5] WC Niu, CZ Zou, B Li, et al., Adaptive vibration suppression of time-varying structures with enhanced FxLMS algorithm, Mechanical Systems and Signal
Processing 118 (2019) 93–107.
[6] Wenchao Niu, Bin Li, Tao Xin, Wei Wang, Vibration active control of structure with parameter perturbation using fractional order positive position feedback
controller, Journal of Sound and Vibration 430 (2018) 101–114.
[7] Osama Abdeljaber, Onur Avci, Daniel J. Inman, Active vibration control of flexible cantilever plates using piezoelectric materials and artificial neural networks,
Journal of Sound and Vibration 363 (2016) 33–53.
[8] Mohammed Kassem, Zhichun Yang, Yingsong Gu, Wei Wang, Ehab Safwat. (2020) Active dynamic vibration absorber for flutter suppression, Journal of Sound
and Vibration 469 (2000), 115110.
[9] Davood Asadi, Touraj Farsadi, Active flutter control of thin walled wing-engine system using piezoelectric actuators, Aerospace Science and Technology 102
(2020), 105853.
[10] L. Song, P. Xia, Active Control of Helicopter Structural Response Using Piezoelectric Stack Actuators, Journal of Aircraft 50 (2) (2013) 659–663.
[11] L. Song, P. Xia, A Harmonic Synchronous Identification Updating Method for Active Control of Helicopter Structural Response Driven by Piezoelectric Stack
Actuators, Journal of the American Helicopter Society 60 (3) (2015), 032013.
[12] Tai NT, KK. Ahn, Adaptive proportional–integral–derivative tuning sliding mode control for a shape memory alloy actuator, Smart Materials and Structures 20
(5) (2011), 055010.
[13] S. Elliott, I. Stothers, P. Nelson, A multiple error LMS algorithm and its application to the active control of sound and vibration, in: IEEE Transactions on
Acoustics, Speech, and Signal Processing 35, 1987, pp. 1423–1434.
[14] J. Oh, S. Park, J. Hong, et al., Active vibration control of flexible cantilever beam using piezo actuator and Filtered-X LMS algorithm, KSME International
Journal 12 (1998) 665–671.
[15] D. Meng, P. Xia, L. Song, MIMOMH Feed-Forward Adaptive Vibration Control of Helicopter Fuselage by Using Piezoelectric Stack Actuators, Journal of
Vibration and Control 24 (23) (2018) 5534–5545.
[16] D. Meng, P. Xia, L. Song, Experimental Study on Piezoelectric Stack Actuator Driven Active Vibration Control of Helicopter Floor Structure, Journal of Aircraft
57 (2) (2020).
[17] T Aboulnasr, K Mayyas, A robust variable step-size LMS-type algorithm: analysis and simulations, IEEE Transactions on Signal Processing 45 (3) (1997)
631–639.
[18] Weiguang Li, Wei Wang, et al., Error Signal Differential Term Feedback Enhanced Variable Step Size FxLMS Algorithm for Piezoelectric Active Vibration
Control, Shock and Vibration (2020), 8832467.
[19] P Shivashankar, S Gopalakrishnan, 2020) Design, modeling and testing of d33-mode surface-bondable multilayer piezoelectric actuator, Smart Materials and
Structures 29 (4) (2000).
[20] N. Wan, J. Wen, Y. Hu, et al., A parasitic type piezoelectric actuator with an asymmetrical flexure hinge mechanism, Microsyst Technol 26 (2020) 917–924.
[21] Zheng Li, Liang Zhao, Xuze Yu, A novel stick-slip piezoelectric actuator based on two-stage flexible hinge structure, Review of Scientific Instruments 91 (5)
(2020), 055006.
[22] K Zheng, SZ Yan, SZ Wen, et al., Influence of preload on the piezoelectric for piezoelectric stack actuator, Piezoelectric & Acoustooptics 25 (5) (2003) 363–365.
[23] W Wang, ZC Yang, A Compact Piezoelectric Stack Actuator and Its Simulation in Vibration Control, Tsinghua Science & Technology 14 (2) (2009) 43–48.
[24] S.M. Khot, N.P. Yelve, R. Tomar, et al., Active vibration control of cantilever beam by using PID based output feedback controller, Journal of Vibration and
Control 18 (3) (2012) 366–372.
[25] A. Preumont, Vibration Control of Active Structures: An Introduction, Springer, Berlin, Germany, 2011.
[26] B. Widrow, M.E. Hoff, Adaptive Switching Circuits, IRE Wescon Convention Record 4 (1960) 96–104.

17
W. Li et al. Journal of Sound and Vibration 509 (2021) 116243

[27] SW Gao, Huang QZ, ZY. Gao, Active vibration control algorithm using reference signal self-extraction, Journal of Vibration, Measurement and Diagnosis 30 (5)
(2010) 514–518.
[28] E F Crawley, J. De Luis, Use of piezoelectric actuators as elements of intelligent structures, AIAA journal 25 (10) (1987) 1373–1385.
[29] L Bin, L Yugang, Y Xuegang, et al., Maximal modal force rule for optimal placement of point piezoelectric actuators for plates, Journal of intelligent material
systems and structures 11 (7) (2000) 512–515.
[30] Zhang Jiajun, Optimization Parameters of PID Controller Parameters Based on Particle Swarm Optimization, Computer Simulation 27 (10) (2010) 191–193,
222.

18