Professional Documents
Culture Documents
Model Formulation: MZ CZ KZ KZ My
Model Formulation: MZ CZ KZ KZ My
The model of the single-degree-of-freedom (SDOF) VI-EH system is a nonlinear differential equation
(NDE) and the equation of motion of the moving levitated magnet inside the dual-purpose device is
given as
Z Y
where z x y , x X sin(t ) , y Y sin(t ) , z Z sin(t ) and Ta then the Vibration
Y
Isolator Energy Harvester (VI-EH) model becomes,
The dynamic model of the VI-EH system as described in (2) is representative of the duffing-equation.
Such equations have previously been solved using several mathematical tools like the method of
multiple scales [1], direct numerical integration [2] , nonlinear normal forms (NLNF) [3] and harmonic
balance method [4],[5]. In this study, the output frequency response function (OFRF) will be employed
for the analysis, design and optimisation of the VI-EH system.
The average input mechanical power of the VI-EH system due to the base acceleration is given as
1
Pin mY 2 3 (3)
However the average output power (power harvested) across the load resistance of the harvesting
circuit is given as
1 k Z
2
Pout t RL (4)
2 Rc RL
where kt BNl
Considering the average output power, Pout of the VI-EH system when subjected to a harmonic base
excitation with an average input mechanical power of Pin , the energy conversion efficiency e of the
VI-EH system is given as;
Pout
e (5)
Pin
The output response of system (3), Z is a function of the excitation frequency, and the nonlinear
parameter, k3 . Therefore considering (4) and (5), it can be deduced that the average output power, Pout
and energy conversion efficiency, e are both functions of these parameters. Though the VI-EH system
has a dual-purpose function, its primary function is vibration isolation where the absolute displacement
transmissibility is less than unity i.e. Ta 1 while the secondary function is energy harvesting. This
implies the frequency range of interest is n 2 which comprises the region of isolation hence
energy harvesting can only happen within this region. The objective here is to maximise e within the
isolation range Ta 1 . The next section discusses the OFRF concept and thereafter the OFRF
1
polynomials of the relative displacement transmissibility, absolute displacement transmissibility and
subsequently the average output power and energy conversion efficiency is derived.
where L is the order of the derivative and M is the maximum degree of nonlinearity in terms of the
system input and output, y (t ) and z (t ) . The system output response of (2) can be represented by a
polynomial function in terms of the system parameters as
m1 mSN
Z ( j ) ( j1 ,, jSN ) ( j )1j1 SjNSN (7)
j1 0 jSN 0
where Z ( j ) represents the output spectrum of (6), ( j1 ,, jSN ) ( j ) are complex-valued frequency
functions (also called ‘OFRF coefficients’) of the system linear parameters and system input while
1j1 SjNSN is a set of monomials (OFRF structure) in the OFRF representation of the output spectrum.
Let the set of monomials in the OFRF representation of the nth-order output spectrum of (3) be denoted
as M and the frequency function vector be denoted as ( j ) , therefore the OFRF polynomial is
obtained as
Z ( j ) M ( j )T (8)
where
2
L n1 n( m p ) L
Mn c0,n (l1 ,, ln ) c p,(m- p) (l1 ,, lm ) Mn(m p), p
l1 ,,ln 0 m p1 p1 l1 ,,ln 0
(10)
n L
c p ,0 (l1 ,, lm ) Mn, p
p2 l1 ,,ln 0
N
Then the set of monomials is obtained as M Mn
n 1
System Analysis
In this section, the OFRF method is used in the analytical study of system (2). System (2) is a particular
instance of (6) for L 2 and M 3 with system parameters obtained as c10 (2) m , c10 (1) c1 ,
c10 (0) k1 , c30 (000) k3 , c01 (0) m 2Y , else cp ,m p 0 . Applying the algorithm for obtaining the
OFRF structure (monomials) as presented in (10) and (11) to system (2) up to 7th-order i.e. N=15 , yields
the following monomials;
N
M M n 1, k3 , k32 , k33 , k34 , k35 , k36 , k37 (12)
n1
The frequency function vector ( j ) is computed by applying the method described in proposition 4
outlined in [6]. Fifteen simulation studies were conducted where system (2) was excited with same base
input but the nonlinear stiffness parameter, k3 takes different values from a training set
k3 0 : 0.1:1.4 106 . The output spectra of the system are evaluated from the 15 sets of numerically
simulated (Runge-Kutta method) system outputs to obtain the results Z ( j ) |k (i ) for i 1, 2,3......,15
3
and the estimates of the OFRF coefficients are evaluated using the least square method as
Z ( j )
0 ( j ) k3 (1)
1 ( j ) Z ( j ) k3 (2)
1
( j ) T T
(13)
( j ) Z ( j )
6 k3 (14)
( j )
7 Z ( j ) k (15)
3
where
3
1 k3 (1) k32 (1) k36 (1) k37 (1)
1 k (2) k 2 (2) k36 (2) k37 (2)
3 3
(14)
2
1 k3 (15) k3 (15) k36 (15) k37 (15)
Therefore, the OFRF of system (2), when subjected to a specific input, is obtained as
Z ( j , k3 ) 0 ( j ) 1 ( j )k3 2 ( j )k32 3 ( j )k33 6 ( j )k36 7 ( j )k37
R (15)
Z ( j , k3 ) r ( j )k3r where R 7
r 0
where (j ) are the frequency functions dependent on the system input and linear characteristic
parameters, c1 and k1 .
Using the algorithm proposed in [12] , the square magnitude of the output response of (15) can be
computed as
R R
Z ( j, k3 ) Z ( j, k3 ) r ( j )k3r r ( j )k3r
2
Z ( j , k3 )
r 0 r 0
t
Z ( j , k3 ) 00* k3t t*
2
t 1 0
(16)
Z ( j , k3 )
2
0 1k3 k k k
2 13 14
2 3 13 3 14 3
R
r ( )k3r where R 14
2
Z ( j , k3 )
r0
kt RL R
2
r ( )k3r
(18)
2mRL 2Y Rc RL r 0
2
R kt RL
2
( ) r ( )k3 where ( ) =
r
2
r 0 2mRL 2Y Rc RL
4
From (17) and (18), the following representations for Pout and e can be made.
R
P ( , k )
out 3
r 0
r ( ) k3r where r ( ) ( ) r ( )
(19)
R
e ( , k3 ) r ( )k3r where r ( ) ( ) r ( )
r 0
Eq. (19) shows the OFRF approximations for the average output power, Pout and the energy conversion
efficiency, e of the VI-EH system respectively. The OFRF for the absolute displacement
transmissibility, Ta , can also be derived as in (13) thus
R
Ta ( , k3 ) r ( ) k3r (20)
r 0
where r ( ) , r ( ) and r ( ) are frequency functions dependent on the system input and linear
characteristic parameters.
f1 (c1 , k1 , k3 ) : k1 24.87 0
(21)
f 2 (c1 , k1 , k3 ) : c1 2.7 0
The OFRF polynomials for the system performance indices of interest Ta , Pout ,e are obtained subject
to the bound constraint k3 1.4 106 Nm -3 . Given the derived OFRF polynomial approximations, the
optimisation problem can be formulated accordingly;
max e (c1 , k1 , k3 )
c1 , k1 , k3
f (c , k , k ) : k 24.87 0
1 1 1 3 1
f (c , k , k ) : c 2.7 0
2 1 1 3 1 (22)
f 3 (c1 , k1 , k3 ) : k3 1.4 10 0
s.t. 6
R
f 4 (c1 , k1 , k3 ) : r ( ) k3r 1 0
r 0
In this study, four levels of base acceleration inputs, 0.25g, 0.5g, 0.75g and 1g , are considered. For
each input, the maximum e obtainable by the VI-EH system, subject to the set of constraints on the
system, is computed using a MATLAB optimisation algorithm ‘fmincon’ and the results presented in
table 1.
5
Results and discussions
Table 1 shows the maximum energy conversion efficiency attainable subject to the system constraints.
Base excitation level Absolute disp trans () Max. Energy Nonlinear stiffness
conversion efficiency value ()
()
6
References
[1] T. D. Equation and N. Oscillators, The Duffing Equation. .
[2] A. Erturk and D. J. Inman, “Broadband piezoelectric power generation on high-energy orbits
of the bistable Duffing oscillator with electromechanical coupling,” J. Sound Vib., vol. 330,
no. 10, pp. 2339–2353, 2011.
[3] A. Cammarano, S. Neild, S. Burrow, D. Wagg, and D. Inman, “Optimum resistive loads for
vibration-based electromagnetic energy harvesters with a stiffening nonlinearity,” J. Intell.
Mater. Syst. Struct., vol. 25, no. 14, pp. 1757–1770, 2014.
[4] S. Souayeh and N. Kacem, “Computational models for large amplitude nonlinear vibrations of
electrostatically actuated carbon nanotube-based mass sensors,” Sensors Actuators, A Phys.,
vol. 208, pp. 10–20, 2014.
[5] A. Jallouli, N. Kacem, G. Bourbon, P. Le Moal, V. Walter, and J. Lardies, “Pull-in instability
tuning in imperfect nonlinear circular microplates under electrostatic actuation,” Phys. Lett. A,
vol. 380, no. 46, pp. 3886–3890, 2016.
[6] Z. Q. Lang, S. a. Billings, R. Yue, and J. Li, “Output frequency response function of nonlinear
Volterra systems,” Automatica, vol. 43, pp. 805–816, 2007.
[7] Z.-Q. Lang and S. . Billings, “Output frequencies of a class of nonlinear systems,” Int. J.
Control, vol. 67, no. 5, pp. 713–730, 1997.
[8] Z.-Q. LANG and S. A. BILLINGS, “Output frequency characteristics of nonlinear systems,”
Int. J. Control, vol. 64, no. 6, pp. 1049–1067, Feb. 2007.
[9] X. J. Jing, Z. Q. Lang, and S. A. Billings, “Output frequency properties of nonlinear systems,”
Int. J. Non. Linear. Mech., vol. 45, no. 7, pp. 681–690, 2010.
[10] J. C. Peyton Jones and K. Choudhary, “Output frequency response characteristics of nonlinear
systems. Part II: overlapping effects and commensurate multi-tone excitations,” Int. J. Control,
vol. 85, no. 9, pp. 1279–1292, 2012.
[11] Z. Q. Lang, P. F. Guo, and I. Takewaki, “Output frequency response function based design of
additional nonlinear viscous dampers for vibration control of multi-degree-of-freedom
systems,” J. Sound Vib., vol. 332, no. 19, pp. 4461–4481, 2013.
[12] X. J. Jing, Z. Q. Lang, and S. A. Billings, “Output frequency response function-based analysis
for nonlinear Volterra systems,” Mech. Syst. Signal Process., vol. 22, no. 1, pp. 102–120,
2008.
[13] Y. Zhu and Z. Q. Lang, “Design of Nonlinear Systems in the Frequency Domain: An Output
Frequency Response Function-Based Approach,” IEEE Trans. Control Syst. Technol., pp. 1–
14, 2017.