Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

A Vibration Absorption Method

1
Yushu Bian
School of Mechanical Engineering for Alleviating Impact of the
and Automation,
Beihang University,
Beijing 100191, China
Flexible Robotic Arm
e-mail: bian_bys@buaa.edu.cn
Impact may excite intense vibration responses of the flexible robotic arm and thus deteri-
Zhihui Gao orate its working performance. A vibration absorption method is put forward to alleviate
School of Mechanical Engineering impact influence of the flexible robotic arm. To dissipate the impact vibration energy, a
and Automation, slider mass–spring–dashpot mechanism is used as a vibration absorber and attached to
Beihang University, the flexible robotic arm. Internal resonance is sufficiently utilized to provide a bridge for
Beijing 100191, China the transfer of impact vibration energy between the flexible link and the absorber via non-
linear coupling. In the presence of damping of the absorber, the impact vibration energy
of the flexible link can be effectively migrated to and dissipated by the absorber. Numeri-
Ming Fan cal simulations and virtual prototype simulations verify its effectiveness and feasibility in
The Second of Corps of
alleviating impact vibration of the flexible robotic arm undergoing a collision.
Engineers GEH PLA,
[DOI: 10.1115/1.4035409]
Beijing 100036, China
Keywords: flexible robotic arm, impact, vibration

1 Introduction components. However, their flexible cantilever beams were mod-


eled as a rigid beam connected by a torsion spring. Recently, the
Impact is a complex physical phenomenon, which occurs when
above research has been extended to the distributed flexible beam.
two or more bodies collide with each other. When a robotic arm
Pai et al. [13] used a higher-order internal resonance absorber to
performs such tasks as the capture and assembly, the impact
reduce vibration of an aluminum plate. Yaman and Sen [14]
unavoidably arises. Usually, impact may excite intense vibration
absorbed vibration of a cantilever beam with a tip mass and pen-
responses of the flexible robotic arm and thus degrade its end-
dulum. Hui et al. [15] suggested the internally resonant energy
effector accuracy [1].
transfer from the symmetrical to antisymmetrical mode to reduce
In addition to a number of methods on impact control for the
vibration. But all of these models involved the flexible structures
rigid robotic arm [2,3], studies on reducing impact vibration of the
without rigid motion. Since the flexible robotic arm is a typical
flexible robotic arm have drawn much attention [4–6]. Many of
nonlinear dynamic system possessing both flexural structural
these control algorithms need to know accurate collision position
deformation and rigid joint motion, it usually exhibits much more
and collision time in advance, but it is difficult in some actual
complex dynamic behaviors than the flexible structure. As a
environments like outer space. Furthermore, because of the struc-
result, whether internal resonance can be successfully employed
tural flexibility, the collision detection time becomes longer. As a
to overcome vibration problem of the flexible robotic arm requires
result, control methods based on the collision detection probably
an in-depth study. Its stability and effectiveness should be proved
face challenges due to the delay of reaction strategy.
and verified in detail. To the best of our knowledge, there is little
In recent years, various active control schemes have been sug-
research on alleviating impact influence of the flexible robotic
gested to suppress vibration of the flexible robotic arm [7,8]. In
arm via internal resonance.
addition, numerous control methods based on smart materials,
In this paper, a vibration absorption method is put forward to
such as piezoelectric ceramic and shape memory alloy, have been
alleviate impact influence of the flexible robotic arm. To dissipate
investigated [9]. Compared with these vibration suppression meth-
the impact vibration energy, a slider mass–spring–dashpot mecha-
ods, since impact often excites strong postimpact vibration
nism is used as a vibration absorber and attached to the flexible
responses, vibration absorption methods based on transfer and dis-
robotic arm. Internal resonance is sufficiently utilized to provide a
sipation of vibration energy are especially effective to control
bridge for the transfer of impact vibration energy between the
strong vibration. Besides, since strong vibration can unavoidably
flexible link and the absorber via nonlinear coupling. In the pres-
excite distinct nonlinear effects, many methods based on linear
ence of damping of the absorber, the impact vibration energy of
simplified model may cause fundamental mistakes.
the flexible link can be effectively migrated to and dissipated by
In fact, although the nonlinearity can increase the analysis diffi-
the absorber.
culty, modal interaction caused by the nonlinearity can be
expected to migrate and dissipate strong vibration energy. Internal
resonance is such an important nonlinear principle. Golnaraghi
[10] tried to use internal resonance to decrease structural vibration 2 System Model
of a flexible cantilever beam. Afterward, Tuer et al. [11] suggested A model is developed for alleviating the impact influence based
two methods to control structural vibration of a flexible cantilever on migration and absorption of vibration energy, as shown in Fig.
beam based on internal resonance. Oueini and Golnaraghi [12] 1. It is a plane system and consists of a rigid link (link 1), a flexi-
built an internal resonance controller with analog electronic ble link (link 2), and a vibration absorber. Driven by the torques
s1 and s2 , links 1 and 2 can rotate around joints O1 and O2 , which
rotation angles are denoted by q1 and q2 , respectively. Link 1 is a
1
Corresponding author. uniform beam with the length l1 and mass m1 . Link 2 is a uniform
Contributed by the Design Engineering Division of ASME for publication in the
JOURNAL OF COMPUTATIONAL AND NONLINEAR DYNAMICS. Manuscript received May 31,
Euler–Bernoulli beam with the length l2 , the mass m2 , the rectan-
2016; final manuscript received November 23, 2016; published online January 20, gle cross section of height h and width b, the constant flexural
2017. Assoc. Editor: Hiroshi Yabuno. rigidity E2 I2 , and mass per length q2 . A load mB is attached to the

Journal of Computational and Nonlinear Dynamics JULY 2017, Vol. 12 / 041006-1


C 2017 by ASME
Copyright V

Downloaded From: http://computationalnonlinear.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jcnddm/936012/ on 02/10/2017 Terms of Use: http://www.as


w12 ¼ m2 l22 =3 þ m2 l1 l2 cos q2  B4 l1 sin q2 u1
þ mB ðl22  l1 u1l2 sin q2 u1 þ l1 l2 cos q2 Þ
þ m½sa ðu1r u1 þ sÞ þ ðsa þ l1 sin q2 Þðu1r u1 þ s þ sa Þ
þ ðr þ l1 cos q2 Þðr  sa b ~ u Þ
r 1

w13 ¼ B3 þ B4 l1 cos q2  m2 l1 B1l2 sin q2 u1


þ mB ½u1l2 ðl1 cos q2 þ l2 Þ  l1 B1l2 sin q2 u1 
~ ðu1r u þ sÞ þ ðsa þ l1 sin q2 Þðb
þ m½sa b ~ sþb~ sa þ B1r u Þ
r 1 r r 1
Fig. 1 Flexible robotic arm with a vibration absorber ~ 2 u Þ  u1r b
þ ðr þ l1 cos q2 Þðu1r  sa b ~ sa u
r 1 r 1

tip of link 2. The vibration absorber is used to absorb the impact


~ l1 sin q2 u þ r þ l1 cos q2 Þ
w14 ¼ mðb
vibration and cantilevered onto the flexible link at x2 ¼ r. The r 1
absorber is a slider mass–spring–dashpot mechanism with mass
m, stiffness k, and damping c. The displacement of mass m is w15 ¼ m2 l1 l2 sin q2 =2 þ mB l1 l2 sin q2
expressed as s and the equilibrium position is expressed as sa .
þ ml1 ½cos q2 ðsa þ l1 sin q2 Þ þ sin q2 ðr þ l1 cos q2 Þ
Only the deflection d of link 2 about y2 axis is considered in
present work. According to the assumed-modes theory, the deflec-
tion d is composed of multiple modes. Without loss of generality, w16 ¼ ½m2 l1 l2 sin q2 =2 þ mB l1 l2 sin q2
only the fundamental mode of the flexible link is studied because þ ml1 ðsa cos q2  r sin q2 Þ
of its most contribution to the vibration response in common
cases. Then, the deflection can be written as
w17 ¼ 2B4 l1 sin q2 þ 2mB l1 u1l2 sin q2  2mu1r ðsa þ l1 sin q2 Þ
dðx2 ; tÞ ¼ u1 ðx2 Þu1 ðtÞ (1)
w18 ¼ m2 l1 B1l2 sin q2  mB l1 B1l2 sin q2 þ mB1r ðsa þ l1 sin q2 Þ
where u1 ðtÞ and u1 ðx2 Þ are the fundamental modal generalized ~ 2 ðr þ l1 cos q2 Þ
 msa b r
elastic coordinate and mode shape of the flexible link, respec-
tively. The angle of the tangent of the flexible link at x2 ¼ r
~ u ðtÞ, where w19 ¼ 2mðsa þ l1 sin q2 Þ
with respect to x2 axis is denoted by a  b r 1
~
b r ¼ ðdu1 =dx2 Þjx2 ¼r . The axial displacement resulting from the ~ ðsa þ l1 sin q2 Þ
w20 ¼ 2mb r
transverse bending of the flexible link can be written as
ð x2  2 s1 is the torque of the joint 1
1@dð#; tÞ 1
hðx2 Þ ¼  d# ¼  B1 u21 (2)
0 2 @# 2 w21 ¼ m2 l22 =3 þ m2 l1 l2 cos q2 =2  B4 l1 sin q2 u1
Ðx þ mB ðl22  l1 u1l2 sin q2 u1 þ l1 l2 cos q2 Þ
where # is a dummy variable and B1 ¼ 0 2 f½du1 ð#Þ=d#g2 d#.
Using Kane’s method, the dynamic equations of the system are þ m½sa ð2u1r u1 þ 2s þ sa Þ  l1 sin q2 ðu1r u1 þ s þ sa Þ
derived ~ u l1 cos q2
þ rl1 cos q2 þ r 2  sa b r 1

w11 q€1 þ w12 q€2 þ w13 u


€ 1 þ w14 s€ ~ u Þ  mr 2
w22 ¼ m2 l22 =3 þ mB l2 þ msa ð2u1r u1 þ 2s þ sa  2r b r 1
¼ w15 q_ 21 þ w16 ðq_ 1 þ q_ 2 Þ2 þ w17 ðq_ 1 þ q_ 2 Þu_ 1
 w18 u_ 2 w19 ðq_ 1 þ q_ 2 Þs_  w20 u_ 1 s_ þ s1 (3) ~ ð2s þ sa  u1r u Þ þ sa B1r u þ ru1r
w23 ¼ B3 þ mB u1l2 l2 þ m½sa b r 1 1

w21 q€1 þ w22 q€2 þ w23 u


€ 1 þ w24 s€ w24 ¼ mr
¼ w25 q_ 21  w26 ðq_ 1 þ q_ 2 Þu_ 1  w27 u_ 2
w25 ¼ m2 l1 l2 sin q2 þ B4 l1 cos q2 u1 þ mB l1 l2 sin q2
 w28 ðq_ 1 þ q_ 2 Þs_  w29 u_ 1 s_ þ s2 (4)
þ ml1 ðsa cos q2 þ r sin q2 Þ
w31 q€1 þ w32 q€2 þ w33 u
€ 1 þ w34 s€ þ c11 u_ 1 þ k11 u1
w26 ¼ 2msa u1r
¼ w35 q_ 21 þ w36 ðq_ 1 þ q_ 2 Þ2 þ w37 u_ 21 þ w38 u_ 1 s_ (5)
~ Þ
w27 ¼ msa ðB1r  b r
w41 q€1 þ w42 q€2 þ w43 u
€ 1 þ w44 s€ þ c22 s_ þ k22 s
w28 ¼ 2msa
¼ w45 u_ 21 þ w46 ðq_ 1 þ q_ 2 Þ2  w47 q_ 21 (6)
~
w29 ¼ 2msa b r
where
s2 is the torque of the joint 2
w11 ¼ m1 l21 =3 þ m2 l22 =3 þ m2 l1 ðl1 þ l2 cos q2 Þ  2B4 l1 sin q2 u1 ~ 2 rÞ  l1 ðb
w31 ¼ m½sa ðB1r  b ~ 2 sa cos q2 þ B1r sin q2 Þu
r r 1
þ mB ðl21 þ l22  2l1 u1l2 sin q2 u1 þ 2l1 l2 cos q2 Þ þ mb~ ð2sa  l1 sin q2 Þs þ mðb ~ s2 þ ru1r Þ þ B3
r r a
þ m½ðu1r u1 þ sa þ l1 sin q2 Þðsa  l1 sin q2 Þ þ 2sa s þ ml1 ðu1r cos q2  b~ sa sin q2 Þ þ l1 B4 cos q2
r
~ u Þ
þ ðr þ l1 cos q2 Þðr þ l1 cos q2  2sa b þ mB u1l2 ðl1 cos q2 þ l2 Þ
r 1

041006-2 / Vol. 12, JULY 2017 Transactions of the ASME

Downloaded From: http://computationalnonlinear.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jcnddm/936012/ on 02/10/2017 Terms of Use: http://www.as


~ 2 rÞu þ 2mb
w32 ¼ msa ðB1r  b ~ sa s þ mðb
~ s2 þ ru1r Þ w31 q€1 þ w32 q€2 þ w33 u
€ 1 þ w34 €s þ c11 u_ 1 þ k11 u1
r 1 r r a
þ B3 þ mB u1l2 l2 ¼ w35 q_ 21 þ w36 ðq_ 1 þ q_ 2 Þ2 þ w37 u_ 21 þw38 u_ 1 s_ þ fc3 (10)

~ 2 s2 þ u2 Þ þ B2 þ mB u2
w33 ¼ mðb w41 q€1 þ w42 q€2 þ w43 u
€ 1 þ w44 s€ þ c22 s_ þ k22 s
r a 1r 1l 2

~ sa ðB1r  b
þ 2mb ~ u1r Þu þ ~ 2 sa s
2mb ¼ w45 u_ 21 þ w46 ðq_ 1 þ q_ 2 Þ2  w47 q_ 21 þ fc4 (11)
r r 1 r

w34 ¼ mu1r Considering Eq. (7) and taking the integral of the dynamics
Eqs. (8)–(11) from t to t þ Dt, one obtains
~ sa cos q2 þ u1r sin q2 Þ þ B4 sin q2 
w35 ¼ l1 ½mðb r
ð tþDt
~ r  u1r Þ
w36 ¼ msa ðb w11 Dq_ 1 þ w12 Dq_ 2 þ w13 Du_ 1 þ w14 Ds_ ¼ vp1 Fc dt (12)
r
t
~ sa ðB1r  b
w37 ¼ mb ~ u1r Þ
r r ð tþDt
w21 Dq_ 1 þ w22 Dq_ 2 þ w23 Du_ 1 þ w24 Ds_ ¼ vp2 Fc dt (13)
w38 ¼ ~ 2 sa
2m3 b t
r

~ sin q2 u
w41 ¼ mðr þ l1 cos q2 Þ  ml1 b ð tþDt
r 1
w31 Dq_ 1 þ w32 Dq_ 2 þ w33 Du_ 1 þ w34 Ds_ ¼ vp3 Fc dt (14)
t
w42 ¼ m3 r; w43 ¼ w34 ; w44 ¼ m; w45 ~ 2;
¼ msa b r
w46 ¼ msa ; w47 ¼ ml1 sin q2 ð tþDt
ð l2 w41 Dq_ 1 þ w42 Dq_ 2 þ w43 Du_ 1 þ w44 Ds_ ¼ vp4 Fc dt (15)
t
B1r ¼ B1 ðx2 Þjx2 ¼r ; B1l2 ¼ B1 ðx2 Þjx2 ¼l2 ; B2 ¼ q2 u21 ðx2 Þdx2
0
where Dq_ 1 , Dq_ 2 , Du_ 1 , and Ds_ are the differences of q_ 1 , q_ 2 , u_ 1 ,
ð l2 and s_ before and after the impact.
B3 ¼ q2 u1 ðx2 Þx2 dx2 ; According to the momentum theorem, we have
0
ð l2
B4 ¼ q2 u1 ðx2 Þdx2 ; u1l2 ¼ u1 ðx2 Þjx2 ¼l2 ; u1r ¼ u1 ðx2 Þjx2 ¼r ð tþDt
0 vpj Fc dt ¼ vpj m0 v0 ð1  j  4Þ (16)
t
Ð l2
c11 is the damping of link 2, k11 ¼ E2 I2 0 f½d 4 u1 ðx2 Þ=dx2 4 g
u1 ðx2 Þdx2 , c22 ¼ c, and k22 ¼ k. Equations (12)–(16) are the impact dynamics equations. By
solving them, the unknown velocities after impact can be
3 Impact Equations determined.
Since the main goal of this study is to reduce impact vibration
responses, it is assumed (1) impact time Dt is infinitesimal, (2) 4 Perturbation Solution
impact is modeled as the point contact, (3) impact process is sim-
plified as an impulse force, and (4) the colliding object is treated Equations (5) and (6) describe the flexible dynamic behavior of
as a point mass. As a result, all velocities remain finite during this the flexible link and the vibration absorber. They play an impor-
infinitesimal impact time, and there are no changes in positions or tant role in vibration control. By analyzing nonlinear coupling
orientations of any bodies in the system as Dt ! 0; only the inte- between the flexible link and the vibration absorber in Eqs. (5)
grals involving q€1 , q€2 , u
€ 1 , s€, and Fc are not zero. and (6), the internal resonance relationship is expected to be
Let a moving mass m0 collides with link 2 at a certain point P. established and used to reduce vibration. Since our method aims
The relative velocity of the mass m0 with respect to the contact to control the vibration of the flexible link and its implementation
point P is v0 and the contact force is Fc 2 R3 . Based on Kane’s mainly relies on the vibration equations, only Eqs. (5) and (6) are
method, the generalized active forces resulted from Fc are considered. To solve them, the equations are nondimensionalized
using the definitions u ¼ pu 1 =l2 , s
  
ffi ¼ s=l2 , t ¼ xu t, q1 ¼ q1 ,
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

fcj ¼ vpj Fc ð1  j  4Þ (7) and q2 ¼ q2 , where xu ¼ k11 =m11 .
Substituting the dimensionless parameters into Eqs. (5) and (6)
where vPj is the jth partial speed of vP , and vP is the velocity of yields
point P.
The dynamics equations of the system undergoing the impact d 2 u d2 s du
can be written as 2
þm~ 12 2 þ g11 
dt dt dt
   
w11 q€1 þ w12 q€2 þ w13 u
€ 1 þ w14 s€ d2 q1 d2 q2  d2 q1 d2 q2 
þ 1 þ d1 2 þ d2 2 u þ d4 2 þ d3 2 s
¼ w15 q_ 21 þ w16 ðq_ 1 þ q_ 2 Þ2 þ w17 ðq_ 1 þ q_ 2 Þu_ 1  w18 u_ 2 dt dt dt dt
2 
  2  
w19 ðq_ 1 þ q_ 2 Þs_  w20 u_ 1 s_ þ s1 þ fc1 (8) d u du du ds d 2 u 
¼ d5 2 u  d6  d7  d 7 s
dt dt dt dt dt2
w21 q€1 þ w22 q€2 þ w23 u
€ 1 þ w24 s€    2   2
dq1 dq2 d2 q1 dq1 d2 q2
¼ w25 q_ 21  w26 ðq_ 1 þ q_ 2 Þu_ 1  w27 u_ 2  d8 þ  d 9  d 10  d 11
dt dt dt2 dt dt2
 w28 ðq_ 1 þ q_ 2 Þs_  w29 u_ 1 s_ þ s2 þ fc2 (9) (17)

Journal of Computational and Nonlinear Dynamics JULY 2017, Vol. 12 / 041006-3

Downloaded From: http://computationalnonlinear.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jcnddm/936012/ on 02/10/2017 Terms of Use: http://www.as


d 2 s d2 u ds d2 v d2 / d2 w d2 w
2
þm~ 21 2 þ g22  þ x2su s 2
þ x2su v ¼ m~ 21 2  d14 21  d15 22
dt dt dt dt dt dt dt
  2   2
du d 2 q d2 q d2 q dv d/ d2 w
¼ d12 
þ d13 21 u  d14 21  d15 22 þ e f22  þ d12 
þ d13 21 /
dt dt dt dt dt dt dt
  
 2
  2  2  2 
dq1 dq2 dq1 dw1 dw2 dw1
þ d16 þ  d17 (18) þd16 þ   d17 (20)
dt dt dt dt dt dt

where The method of multiple scales is employed to obtain a set of


averaged equations that determine the amplitudes and phases of
~ 12 ¼ w34 =m
m ~ 11 ; ~ 2 s2 þ u2 Þ þ B2 þ mB u2 ;
~ 11 ¼ mðb
m the steady-state solutions on a slow scale. For the sake of simplic-
r a 1r 1l 2
ity, only the first-order approximate solutions will be sought in
g11 ¼ c11 =ðxu m
~ 11 Þ subsequent analysis. It is assumed that the solutions of Eqs. (19)
and (20) are represented by the form
~ 2 rÞ  l1 ðb
d1 ¼ m½sa ðB1r  b ~ 2 sa cos q2 þ B1r sin q2 Þ=m
~ 11
r r /ðt ; eÞ ¼ /0 ðT0 ; T1 Þ þ e/1 ðT0 ; T1 Þ;
(21)
vðt ; eÞ ¼ v0 ðT0 ; T1 Þ þ ev1 ðT0 ; T1 Þ
d2 ¼ msa ðB1r  ~ 2 rÞ=m
b ~ 11
r
where Ti ¼ ei t (i ¼ 0, 1). The derivatives of / and v with respect
to t then become expansions in terms of partial derivatives with
~ sa =m
d3 ¼ 2mb ~ 11
r respect to T0 and T1 given by

~ ð2sa  l1 sin q2 Þ=m


d 4 ¼ mb ~ 11 d @ @
r ¼ þe ¼ D0 þ eD1
dt @T0 @T1
2 2 (22)
d5 ¼ 2ml2 b ~ u1r Þ=m
~ sa ðB1r  b ~ 11 d @ @2
r r ¼ þ 2e ¼ D20 þ 2eD0 D1
dt2 @T02 @T0 @T1
d6 ¼ l2 w37 =m
~ 11
Substituting Eqs. (21) and (22) into Eqs. (19) and (20), and
equating each of the coefficients of e0 and e1 to zero, the following
d7 ¼ l2 w38 =m
~ 11 equations are obtained:
Order (e0 )
d8 ¼ w36 =ðm
~ 11 l2 Þ
D20 /0 þ /0 ¼ m
~ 12 D20 v0  d9 D20 w1  d11 D20 w2
(23)
d9 ¼ ½mðb~ s2 þ ru1r Þ þ B3 þ ml1 ðu1r cos q2  b
r a
~ sa sin q2 Þ
r
D20 v0 þ x2su v0 ¼ m
~ 21 D20 /0  d14 D20 w1  d15 D20 w2
þ l1 B4 cos q2 þ mB u1l2 ðl1 cos q2 þ l2 Þ=ðm
~ 11 l2 Þ
Order (e1 )

d10 ¼ w35 =ðm


~ 11 l2 Þ D20 /1 þ /1 ¼ 2D0 D1 /0  f11 D0 /0  m
~ 12 D20 v1  2m
~ 12 D0 D1 v0
 ðd1 D20 w1 þ d2 D20 w2 Þ/0  ðD20 w1 þ D20 w2 Þv0
d11 ¼ m½ðb~ s2 þ ru1r Þ þ B3 þ mB u1l l2 =ðm
~ 11 l2 Þ; m
~ 21 ¼ u1r ;
r a
pffiffiffiffiffiffiffiffiffiffiffiffi
2  d5 D20 /0 /0  d6 ðD0 /0 Þ2  d7 D0 /0 D0 v0
g22 ¼ c22 =ðxu mÞ; xs ¼ k22 =m; xsu ¼ xs =xu  d8 ðD0 w1 þ D0 w2 Þ2  d10 ðD0 w1 Þ2
(24)
d12 ¼ l2 w45 =m; d13 ¼ l1 b~ sin q2 ; d14 ¼ ðr þ l1 cos q2 Þ=l2 ;
r
D20 v1 þ x2su v1 ¼ 2D0 D1 v0  f22 D0 v0  m
~ 21 D20 /1
d15 ¼ r=l2 ; d16 ¼ sa =l2 ; d17 ¼ l1 sin q2 =l2
~ 21 D0 D1 /0 þ d12 ðD0 /0 Þ2 þ d13 D20 w1 /0
 2m
To make the damping and nonlinearities appear in the same þ d16 ðD0 w1 þ D0 w2 Þ2  d17 ðD0 w1 Þ2
perturbation equations, let g11 ¼ ef11 , g22 ¼ ef22 , u ¼ e/,
(25)
s ¼ ev, dq1 =dt ¼ edw1 =dt , and dq2 =dt ¼ edw2 =dt , where e is
introduced to represent the order of nonlinear terms and coupling, Since both w1 and w2 are determined by the operation tasks and
0 < e < 1. Then, Eqs. (17) and (18) can be expressed as are time-varying, it is difficult to give an analytical solution of
Eq. (23) about arbitrary operation tasks from arbitrary start time to
d2 / d2 v d2 w1 d 2 w2 arbitrary end time. To overcome this difficulty, we give an analyti-
þ / ¼ m~ 12  d9  d 11 cal solution of Eq. (23) about arbitrary operation tasks within a
dt2 dt2 dt2
 dt2 
 very small time interval. To this end, the time period of whole
d/ d 2 w1 d 2 w2 work process is divided into a large number of equal small time
þ e f11   d1 2 þ d2 2 /
dt dt dt intervals. Within each time interval, D20 w1 and D20 w2 are substituted
 
2
d w 2
d w d2 / into Eq. (23) and all time-varying terms are considered as constants
 d4 21 þ d3 22 w  d5 2 / for the particular time interval. That is to say that both
dt dt dt
 2 ðd9 D20 w1  d11 D20 w2 Þ and ðd14 D20 w1  d15 D20 w2 Þ are consid-
d/ d/ dv d2 / ered as constant within the particular time interval. Then, they will
 d6  d7    d7 2 w
dt dt dt dt vary about next time interval and keep constant in that time inter-
 2  2  val. Therefore, it means that a constant exciting force is exerted to
dw1 dw2 dw1
 d8 þ   d10 (19) the system within the particular time interval. In this case, Eq. (23)
dt dt dt can be solved for each time interval using the method of multiple

041006-4 / Vol. 12, JULY 2017 Transactions of the ASME

Downloaded From: http://computationalnonlinear.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jcnddm/936012/ on 02/10/2017 Terms of Use: http://www.as


scales. At last, by repeating the above procedures, total vibration cc denotes the complex conjugate of the preceding term; x2n
responses can be obtained for whole work process. (n ¼ 1; 2) are the roots of
Based on the above analysis, the solution of Eq. (23) can be
written as ð1  m ~ 21 Þx4n  ð1 þ x2su Þx2n þ x2su ¼ 0
~ 12 m (27)

/0 ¼ A1 ðT1 Þexpðjx1 T0 Þ þ A2 ðT1 Þexpðjx2 T0 Þ  g1 þ cc


1  x2n ~ 21 x2n
m
v0 ¼ K1 A1 ðT1 Þexpðjx1 T0 Þ þ K2 A2 ðT1 Þexpðjx2 T0 Þ  g2 þ cc Kn ¼ ¼ 2 (28)
2
xn m~ 12 xsu  x2n
(26)

where A1 ðT1 Þ and A2 ðT1 Þ are functions of slow time T1 ; and xn are assumed to be distinct.
g1 ¼ ðd9 D20 w1 þ d11 D20 w2 Þ=2; g2 ¼ ðd14 D20 w1 þ d15 D20 w2 Þ=ð2x2su Þ; Substituting Eq. (26) into Eqs. (24) and (25) yields

D20 /1 þ /1 ¼ 2½jx1 A01 expðjx1 T0 Þ þ jx2 A02 expðjx2 T0 Þ  f11 ½jx1 A1 expðjx1 T0 Þ þ jx2 A2 expðjx2 T0 Þ
~ 12 ½jx1 K1 A01 expðjx1 T0 Þ þ jx2 K2 A02 expðjx2 T0 Þ  ðd1 D20 w1 þ d2 D20 w2 Þ
~ 12 D20 w1  2m
m
 ½A1 expðjx1 T0 Þ þ A2 expðjx2 T0 Þ  g1   ðD20 w1 þ D20 w2 Þ½K1 A1 expðjx1 T0 Þ þ K2 A2 expðjx2 T0 Þ  g2 
 d5 f2g1 x2 A1 expðjx1 T0 Þ þ 2g1 x2 A2 expðjx2 T0 Þ  ðx2 þ x2 ÞA1 A2 exp½jðx2  x1 ÞT0   x2 A2 expð2jx1 T0 Þ
1 2 1 2 1 1
 x22 A22 expð2jx2 T0 Þ  ðx21 þ x22 ÞA1 A2 exp½jðx2 þ x1 ÞT0   x21 A1 A1  x22 A2 A2 g  d6 fx21 A21 expð2jx1 T0 Þ
 x2 A2 expð2jx2 T0 Þ þ 2x1 x2 A1 A2 exp½jðx2  x1 ÞT0   2x1 x2 A1 A2 exp½jðx1 þ x2 ÞT0 Þ þ x2 A1 A1 þ x2 A2 A2 g
2 2 1 2
 d7 fðK1 þ K2 Þx1 x2 A1 A2 exp½jðx2  x1 ÞT0 Þ  K1 x21 A21 expð2jx1 T0 Þ  K2 x22 A22 expð2jx2 T0 Þ
 ðK1 þ K2 Þx1 x2 A1 A2 exp½jðx1 þ x2 ÞT0 Þ þ K1 x21 A1 A1 þ K2 x22 A2 A2 g  d7 f2g2 x21 A1 expðjx1 T0 Þ
þ 2g2 x2 A2 expðjx2 T0 Þ  K1 x2 A2 expð2jx1 T0 Þ  K2 x2 A2 expð2jx2 T0 Þ  ðK2 x2 þ K1 x2 ÞA1 A2 exp½jðx2  x1 ÞT0 Þ
2 1 1 2 2 1 2

 ðK2 x21 þ K1 x22 ÞA1 A2 exp½jðx2 þ x1 ÞT0 Þ  K1 x21 A1 A1  K2 x22 A2 A2 g  d8 ðD0 w1 þ D0 w2 Þ2  d10 ðD0 w1 Þ2 þ cc
(29)

D20 v1 þ x2su v1 ¼ 2½jx1 K1 A01 expðjx1 T0 Þ þ jx2 K2 A02 expðjx2 T0 Þ  f22 ½jx1 K1 A1 expðjx1 T0 Þ þ jx2 K2 A2 expðjx2 T0 Þ
m ~ 21 ½jx1 A01 expðjx1 T0 Þ þ jx2 A02 expðjx2 T0 Þ þ d12 fx21 A21 expð2jx1 T0 Þ  x22 A22 expð2jx2 T0 Þ
~ 21 D20 /1  2m
þ 2x1 x2 A1 A2 exp½jðx2  x1 ÞT0   2x1 x2 A1 A2 exp½jðx1 þ x2 ÞT0 Þ þ x2 A1 A1 þ x2 A2 A2 g
1 2

þ d13 D20 w1 ½A1 expðjx1 T0 Þ þ A2 expðjx2 T0 Þ  g1  þ d16 ðD0 w1 þ D0 w2 Þ2  d17 ðD0 w1 Þ2 þ cc (30)

where ð0 Þ represents the derivatives with respect to the slow time T1 . R21 ¼ 2jH9 A01  jH10 A1 þ H11 A1 þ H12 A1 A2 expðjrT1 Þ
(35)
5 Internal Resonance Analysis R22 ¼ 2jH13 A01  jH14 A2 þ H11 A2  H15 A21 expðjrT1 Þ
In this work, the absorber is used to reduce impact vibration of
the flexible link based on 1:2 internal resonance, i.e., 2x1  x2 . And
Let x2 ¼ 2x1 þ er, where r is a detuning parameter. We have H1 ¼ x1 ð1 þ m
~ 12 K1 Þ; H2 ¼ x1 f11

2x1 T0 ¼ x2 T0  rT1 ; ðx2  x1 ÞT0 ¼ x1 T0 þ rT1 (31) H3 ¼ d1 D20 w1 þ d2 D20 w2 þ ðD20 w1 þ D20 w2 ÞK1 þ 2d5 g1 x21

To determine the solvability conditions of Eqs. (29) and (30), þ 2d7 g2 x21
we seek a particular solution in the form
H4 ¼ d5 ðx21 þ x22 Þ  2d6 x1 x2  d7 ðK1 þ K2 Þx1 x2
/1 ¼ p11 expðjx1 T0 Þ þ p12 expðjx2 T0 Þ;
þ d7 ðK2 x21 þ K1 x22 Þ
v1 ¼ p21 expðjx1 T0 Þ þ p22 expðjx2 T0 Þ (32)
H5 ¼ x2 ð1 þ m
~ 12 K2 Þ
Substituting Eq. (32) into Eqs. (29) and (30), using Eq. (31), and
then equating the coefficients of expðjx1 T0 Þ and expðjx2 T0 Þ on H6 ¼ x2 f11 ;
both sides, one obtains
H7 ¼ d1 D20 w1 þ d2 D20 w2 þ ðD20 w1 þ D20 w2 ÞK2
ð1  x2n Þp1n m ~ 12 x2n p2n ¼ R1n ; þ 2d5 g1 x22 þ 2d7 g2 x22
~ 21 x2n p1n
m þ ðxsu  x2n Þp2n
2
¼ R2n (33)
H8 ¼ ðd5 þ d6 þ 2d7 K1 Þx21 ; H9 ¼ x1 ðK1 þ m21 Þ;
where H10 ¼ x1 f22 K1

R11 ¼ 2jH1 A01  jH2 A1  H3 A1 þ H4 A1 A2 expðjrT1 Þ H11 ¼ d13 D20 w1 ; H12 ¼ 2d12 x1 x2 ; H13 ¼ x2 ðK2 þ m21 Þ;
(34)
R12 ¼ 2jH5 A02  jH6 A2  H7 A2 þ H8 A21 expðjrT1 Þ H14 ¼ x2 f22 K2 ; H15 ¼ d12 x21

Journal of Computational and Nonlinear Dynamics JULY 2017, Vol. 12 / 041006-5

Downloaded From: http://computationalnonlinear.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jcnddm/936012/ on 02/10/2017 Terms of Use: http://www.as


Thus, the problem of determining the solvability conditions of a numerical method is used here. First, m=m2 , sa =l2 , and r=l2 are
Eqs. (29) and (30) is reduced to that of determining the solvability viewed as search variables. Second, the search ranges of these var-
conditions of Eq. (33). iables are defined, respectively. In this example, m=m2 2 ð0; 1,
Since the determinant of the coefficient matrix of Eq. (33) is sa =l2 2 ð0; 0:2, and r=l2 2 ð0; 1. Then, the sign of v is searched
zero according to Eq. (27), the solvability conditions are in terms of the given search variables and search ranges. The final
  search result is v > 0. As a result, a1 and a2 in Eq. (46) are always
 R1n m ~ 12 x2n  bounded. Since the damping is neglected, the system is conserva-
 ¼0 (36)
R x2su  x2n  tive and the energy level remains constant. Therefore, if the
2n
response of the modal amplitude a1 is governed by a periodic
function, then the modal amplitude a2 will be periodic and out of
or
phase with a1 . It means that internal resonance can be successfully
~ 12
m created in the absence of damping, and the energy in the system is
R1n ¼ Kn R2n (37) continuously exchanged between the fundamental mode of the
~ 21
m flexible link and the vibration mode of the absorber.
As a matter of fact, it is always a damping system, i.e., f11 > 0
on account of Eq. (28).
and f22 > 0, the equilibrium points for this case are defined by the
Substituting Eqs. (34) and (35) into Eq. (37) yields
solution of the set of steady state secular term equations, namely,
2jC1 A01  jC2 A1  C3 A1 þ C4 A1 A2 expðjrT1 Þ ¼ 0 (38) C2 C4
 a1 þ a1 a2 sin c ¼ 0 (47)
C1 4C1
2jC5 A02  jC6 A2  C7 A2 þ C8 A21 expðjrT1 Þ ¼ 0 (39)
C6 C8 2
where  a2  a sin c ¼ 0 (48)
C5 4C5 1
~ 12
m m~ 12
C1 ¼ H1 þ K1 H9 ; C2 ¼ H2 þ K1 H10 ; C7 C8 2 C3 C4 2
~ 21
m m~ 21 a2  a cos c  2 a2 þ a cos c þ ra2 ¼ 0 (49)
C5 4C5 1 C1 2C1 2
~ 12
m m~ 12
C3 ¼ H3  K1 H11 ; C4 ¼ H4 þ K1 H12 ;
~ 21
m m~ 21 By inspection, it is determined that the system possesses an infi-
~ 12
m m~ 12 nite number of equilibrium points defined by a1 ¼ 0, a2 ¼ 0, and
C5 ¼ H5 þ K2 H13 ; C6 ¼ H6 þ K2 H14 ; c 2 R.Therefore, by evaluating the Jacobian, the stability of the
~ 21
m m~ 21
~ 12
m m~ 12 system can be ascertained.
C7 ¼ H7  K2 H11 ; C8 ¼ H8  K2 H15 The Jacobian matrix for this case is
~ 21
m m~ 21
2 3
l1 0 0
Substituting An ¼ ð1=2Þan expðjhn Þ; ðn ¼ 1; 2Þ with real an and 4 0 l2 0 5
hn in Eqs. (38) and (39), and then separating the real and imagi- (50)
nary parts, one obtains 0 0 0

C2 C4 where l1 ¼ C2 =C1 and l2 ¼ C6 =C5 .


a01 ¼  a1 þ a1 a2 sin c (40) The corresponding eigenvalues are ðl1 ; l2 ; 0Þ. Since C2 > 0
C1 4C1
and C6 > 0, the sign of l1 and l2 is determined by C1 and C5 ,
C6 C8 2 respectively. Similarly, because it is difficult to determine the sign
a02 ¼  a2  a sin c (41) of C1 and C5 in symbolic form, a numerical method is used. Since
C5 4C5 1
C1 and C5 have relation to m, sa , and r when the structural
C3 C4 parameters of the flexible arm are given, m=m2 , sa =l2 , and r=l2
h01 ¼  a2 cos c (42) are viewed as three search variables and m=m2 2 ð0; 1,
C1 4C1 sa =l2 2 ð0; 0:2, and r=l2 2 ð0; 1 are defined, respectively, in this
example. The final search result is l1 < 0 and l2 < 0. Therefore,
C7 C8 2
a2 h02 ¼ a2  a cos c (43) the modal amplitudes a1 and a2 are stable, as indicated by the neg-
C5 4C5 1 ative eigenvalues. It means that, in the presence of damping, the
impact vibration energy of the flexible link can be successfully
c ¼ h2  2h1 þ rT1 (44) migrated to and dissipated by the absorber.
Eliminating h1 and h2 from Eqs. (42)–(44) gives
6 Simulation and Analysis
C7 C8 2 C3 C4 2
a2 c0 ¼ a2  a1 cos c  2 a2 þ a cos c þ ra2 (45) 6.1 Numerical Simulation. To verify the above theoretical
C5 4C5 C1 2C1 2 analysis, a two-link robotic arm with the second link flexible is
used in the following numerical simulations, as shown in Fig. 1.
In the absence of damping (i.e., f11 ¼ f22 ¼ 0), one has Link 1 is a uniform rigid beam and made of steel, which length
a01 ¼ ½C4 =ð4C1 Þa1 a2 sin c and a02 ¼ ½C8 =ð4C5 Þa21 sin c. Multi- is 1.0 m, the cross section is square, and side length is 0.05 m.
plying a01 by a1 and a02 by va2 , where v ¼ C4 C5 =ðC1 C8 Þ, and then Link 2 is a uniform Euler–Bernoulli flexible beam and made of
adding and integrating them, yields aluminum with elastic modulus of 71 GPa and mass density of
2710 kg/m3, which length is 1.0 m, the rectangle cross-sectional
a21 þ va22 ¼ E ¼ const: (46) height and width are 0.003 m and 0.05 m, respectively, i.e.,
h  b ¼ 0.003 m  0.05 m, and a tip mass mB ¼ 0:2 kg.
where E is a integration constant proportional to the initial energy Suppose the joint motion of two arms is
of the system. ( pffiffiffi 
In Eq. (46), v is determined by the structural parameters of both q1 ¼ 0:15 sinð 2t=5Þ
the flexible link and the vibration absorber, i.e., m, sa , and r. ð0  t  80 s
Because it is difficult to determine the sign of v in symbolic form, q2 ¼ 0:2 cosðpt=7Þ

041006-6 / Vol. 12, JULY 2017 Transactions of the ASME

Downloaded From: http://computationalnonlinear.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jcnddm/936012/ on 02/10/2017 Terms of Use: http://www.as


At t ¼ 10 s, a moving object M0 with the mass of 0.1 kg Fig. 6. At the state of internal resonance, the endpoint deformation
impacts the endpoint of link 2 paralleling y2 axis in the plane is decreased quickly by 50% within 6 s and 90% within 14 s, com-
o2 x2 y2 , as shown in Fig. 2. Its relative velocity with respect to the pared with the uncontrolled case (as shown in Fig. 3).
endpoint is 0.4 m/s and impact time period is 0.01 s. The corre- As we know, when an absorber is added to the flexible robotic
sponding impact vibration of the endpoint is shown in Fig. 3. It arm, several parameters are also introduced, including mass of the
can be seen that, although there are small vibration responses vibrator (m), equilibrium position of the vibrator (sa ), damping of
before impact, large-amplitude impact vibration responses arise the absorber (c22 ), and position of the absorber (r) (i.e., the
suddenly due to abrupt changes of system velocities in the impact distance from the absorber to the joint of the hinged flexible link),
phase. Besides, since the flexible robotic arm keeps moving after as shown in Fig. 1. They will affect vibration responses of the
the collision, newly emerged vibration responses are accumulated flexible robotic arm. Therefore, it is necessary to seek appropriate
on the existing impact vibration responses. In this example, the values to obtain better results. In this example, the aim is to
largest vibration deformation of the endpoint is 0.053 m. Because attenuate the impact vibration quickly and effectively. The optimi-
of low damping, vibration responses are reduced very slowly. zation object and constraints are
To alleviate impact vibration, a vibration absorber is attached
to link 2 at r ¼ 0:5 m, which mass m ¼ 0:05 kg and equilibrium min ! ts
position sa ¼ 0:05 m, as shown in Fig. 1. The same impact param-
eters are used. At the state of internal resonance, Eqs. (40)–(44) s:t: mmin  m  mmax
are integrated numerically in the absence of damping. The rela-
tionship between the modal amplitude a1 (solid line) and a2 sa min  sa  sa max
(dashed line) is shown in Fig. 4. It can be seen that the peaks and
troughs of the responses are exactly 180 deg out of phase, which c22 min  c22  c22 max
means that there is continuous energy exchange between vibration
modes, and the internal resonance has been successfully created. rmin  r  rmax
When the damping of the absorber is considered, i.e.,
c11 ¼ 0:002 and c22 ¼ 0:008, Eqs. (40)–(44) are integrated where ts is the time when the impact vibration response is
numerically. The modal amplitude a1 (solid line) and a2 (dashed decreased to a certain threshold (e.g., 0.01 m in this example),
line) are illustrated in Fig. 5. It is shown that a transfer of energy small ts means quick alleviation; mmin and mmax are the lower and
between modes is existent, because the peaks and troughs of the upper permitted masses of the vibrator; sa min and sa max are the
responses are also exactly 180 deg out of phase. Furthermore, the lower and upper permitted equilibrium positions of the vibrator;
modal amplitudes decrease with time and approach their equilib- c22 min and c22 max are the lower and upper permitted damping of the
rium values (a1 ¼ a2 ¼ 0), indicating that the vibration energy
has been effectively dissipated through internal resonance. The
endpoint deformation of the flexible robotic arm is shown in

Fig. 2 Flexible robotic arm without vibration absorber Fig. 4 Undamped modal amplitudes

Fig. 3 Endpoint vibration undergoing an impact Fig. 5 Damped modal amplitudes

Journal of Computational and Nonlinear Dynamics JULY 2017, Vol. 12 / 041006-7

Downloaded From: http://computationalnonlinear.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jcnddm/936012/ on 02/10/2017 Terms of Use: http://www.as


Fig. 9 A two-link flexible robotic arm with a vibration absorber
in ADAMS

Fig. 6 Endpoint deformation of the controlled arm

Fig. 10 Deformation of the endpoint and the displacement of


the vibrator (without damping)

method can reduce impact vibration of the flexible robotic arm


more quickly and effectively with optimality parameters. Of
course, other optimization objectives and constraints can also be
chosen.

6.2 Virtual Prototype Simulation. In order to further vali-


date the proposed method, several virtual prototype simulations
Fig. 7 Optimized endpoint deformation of the controlled arm
are conducted using ADAMS software. A two-link flexible robotic
arm with the same parameters as the aforementioned optimal
absorber; rmin and rmax are the lower and upper permitted results is used in these simulations. When the arm is not equipped
positions of the absorber. with a vibration absorber, the corresponding impact vibration of
An optimization algorithm based on particle swarm optimiza- the endpoint is shown in Fig. 8, in which the endpoint deformation
tion (PSO) is used to obtain the optimal values. Several important increases abruptly and decreases slowly.
parameter values are chosen as follows: 0:05 kg  m  0:2 kg , If a vibration absorber is attached to the flexible robotic arm (as
0:05 m  sa  0:1 m, 0:001  c2  0:02, and 0:1 m  r  0:9 m. shown in Fig. 9), at the state of 1:2 internal resonance without the
The same impact parameters as the above case are used, damping of the vibration absorber, the relationship between the
and the corresponding optimization results are obtained, i.e., endpoint deformation of the flexible link and the displacement of
m ¼ 0:074 kg, sa ¼ 0:060 m, c2 ¼ 0:012, and r ¼ 0:797m. In this the vibrator is shown in Fig. 10. As can be seen in Fig. 10, the
case, the endpoint deformation is shown in Fig. 7, in which the endpoint amplitude gradually increases and gradually decreases,
impact vibration is decreased to 0.01 m within 9.5 s. As for the and then repeats again and again; so does the vibrator amplitude.
case without optimization (as shown in Fig. 6), the impact vibra- In addition, when the endpoint amplitude decreases, the vibrator
tion is decreased to 0.01 m within 11 s. It is demonstrated that this amplitude increases simultaneously, indicating the vibration

Fig. 8 Endpoint deformation undergoing a collision (without a vibration absorber)

041006-8 / Vol. 12, JULY 2017 Transactions of the ASME

Downloaded From: http://computationalnonlinear.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jcnddm/936012/ on 02/10/2017 Terms of Use: http://www.as


bridge for the transfer of impact vibration energy between the
flexible link and the absorber via nonlinear coupling. Through
perturbation analysis, the 1:2 internal resonance is successfully
created and its stability is proven for the flexible robotic arm with
rigid motion. In the presence of damping of the absorber, the
impact vibration energy of the flexible link can be effectively
migrated to and dissipated by the absorber. Numerical simulations
and virtual prototype simulations verify its effectiveness and fea-
sibility in alleviating impact vibration of the flexible robotic arm
undergoing a collision.

Acknowledgment
This project is supported by National Natural Science Founda-
tion of China (Grant No. 51675017) and Pre-Research Foundation
of GEH PLA (Grant No. 9140A34030315 KG18081).

Fig. 11 Endpoint deformation of the flexible link (without


References
damping and with damping) [1] Khulief, Y. A., 2013, “Modeling of Impact in Multibody Systems: An
Overview,” ASME J. Comput. Nonlinear Dyn., 8(2), p. 021012.
[2] Dupree, K., Hu, G., and Dixon, W. E., 2008, “Adaptive Lyapunov-Based Con-
trol of a Robot and Mass–Spring System Undergoing an Impact Collision,”
IEEE Trans. Syst. Man Cybern., Part B, 38(4), pp. 1050–1061.
[3] Uemura, M., and Kawamura, S., 2010, “Passivity-Based Controllers for
Periodic Motions of Multi-Joint Robots With Impact Phenomena,” IEEE/RSJ
International Conference on Intelligent Robots and Systems, Oct. 18–22, pp.
664–669.
[4] Garcia, A., Feliu, V., and Somolinos, J. A., 2001, “Gauge Based Collision
Detection Mechanism for a New Three-Degree-of-Freedom Flexible Robot,”
IEEE International Conference on Robotics and Automation, Vol. 4,
pp. 3853–3858.
[5] Xu, W. L., and Yue, S., 2004, “Pre-Posed Configuration of Flexible Redundant
Robot Manipulators for Impact Vibration Alleviating,” IEEE Trans. Ind. Elec-
tron., 51(1), pp. 195–200.
[6] Liu, S., Wu, L., and Lu, Z., 2007 “Impact Dynamics and Control of a Flexible
Fig. 12 Comparison of the endpoint deformation Dual-Arm Space Robot Capturing an Object,” Appl. Math. Comput., 185(2),
pp. 1149–1159.
[7] Orszulik, R. R., and Shan, J., 2011, “Vibration Control Using Input Shaping
energy is being transferred from the flexible link to the absorber and Adaptive Positive Position Feedback,” J. Guid. Control Dyn., 34(4),
pp. 1031–1044.
and vice versa. This “beat” phenomenon can verify the existence [8] Abiko, S., and Yoshida, K., 2010, “Adaptive Reaction Control for Space
of internal resonance. Robotic Applications With Dynamic Model Uncertainty,” Adv. Rob., 24(8–9),
In the presence of the damping of the vibration absorber, the pp. 1099–1126.
endpoint deformation of the flexible link at the state of internal [9] Dadfarnia, M., Jalili, N., Xian, B., and Dawson, D. M., 2004, “A Lyapunov-
Based Piezoelectric Controller for Flexible Cartesian Robot Manipulators,”
resonance is shown in Fig. 11. The endpoint deformation without ASME J. Dyn. Syst., Meas., Control, 126(2), pp. 347–358.
the vibration absorber is also shown in Fig. 11. For the conven- [10] Golnaraghi, M. F., 1991, “Regulation of Flexible Structures Via Nonlinear
ience of comparison, these two cases are all listed in Fig. 12. It Coupling,” Dyn. Control, 1(4), pp. 405–428.
can be seen that the endpoint deformation of the flexible link has [11] Tuer, K. L., Golnaraghi, M. F., and Wang, D., 1994, “Development of a Gener-
alised Active Vibration Suppression Strategy for a Cantilever Beam Using
been effectively decreased by way of internal resonance. Internal Resonance,” Nonlinear Dyn., 5(2), pp. 131–151.
Through above simulations and analyses, it is verified that the [12] Oueini, S. S., and Golnaraghi, M. F., 1996, “Experimental Implementation of
proposed method is effective in alleviating impact vibration of the the Internal Resonance Control Strategy,” J. Sound Vib., 191(3), pp. 377–396.
flexible robotic arm undergoing a collision. [13] Pai, P. F., Rommel, B., and Schulz, M. J., 2000, “Non-Linear Vibration Absorb-
ers Using Higher Order Internal Resonances,” J. Sound Vib., 234(5),
pp. 799–817.
7 Conclusion [14] Yaman, M., and Sen, S., 2007, “Determining the Effect of Detuning Parameters
on the Absorption Region for a Coupled Nonlinear System of Varying
In this paper, a vibration absorption method is put forward to Orientation,” J. Sound Vib., 300(1–2), pp. 330–344.
alleviate impact influence of the flexible robotic arm. To dissipate [15] Hui, C. K., Lee, Y. Y., and Ng, C. F., 2011, “Use of Internally Resonant Energy
Transfer From the Symmetrical to Anti-Symmetrical Modes of a Curved Beam
the impact vibration energy, a slider mass–spring–dashpot mecha- Isolator for Enhancing the Isolation Performance and Reducing the Source
nism is used as a vibration absorber and attached to the flexible Mass Translation Vibration: Theory and Experiment,” Mech. Syst. Signal
robotic arm. Internal resonance is sufficiently utilized to provide a Process., 25(4), pp. 1248–1259.

Journal of Computational and Nonlinear Dynamics JULY 2017, Vol. 12 / 041006-9

Downloaded From: http://computationalnonlinear.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jcnddm/936012/ on 02/10/2017 Terms of Use: http://www.as

You might also like