Scribd
Upload
47 views6 pages

Adaptive Force Balancing An Unbalanced Rotor: Jasim Ahmed and Dennis S. Bernstein'

This 3 sentence summary provides the key details about the document: The document presents an adaptive feedback control law for balancing an unbalanced planar rotor. The control law estimates the location of the rotor's center of mass using adaptation techniques, and applies forces to cancel imbalance forces without requiring prior knowledge of the center of mass location. Numerical simulations show the control law is able to asymptotically stabilize the rotor under various conditions, including when the center of mass or rotor speed changes.

Uploaded by

Jessica Sanders
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
47 views6 pages

Adaptive Force Balancing An Unbalanced Rotor: Jasim Ahmed and Dennis S. Bernstein'

This 3 sentence summary provides the key details about the document: The document presents an adaptive feedback control law for balancing an unbalanced planar rotor. The control law estimates the location of the rotor's center of mass using adaptation techniques, and applies forces to cancel imbalance forces without requiring prior knowledge of the center of mass location. Numerical simulations show the control law is able to asymptotically stabilize the rotor under various conditions, including when the center of mass or rotor speed changes.

Uploaded by

Jessica Sanders
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 6

Proceedingsof the 38"'

Conference on Decision & Control


Phoenix, Arizona USA December 1999

TuMll

14:20

Adaptive Force Balancing of an Unbalanced Rotor


Jasim Ahmed and Dennis S. Bernstein'
Department of Aerospace Engineering, The University of Michigan
Ann Arbor, MI 48109-2140
(734) 7643719, (734) 763-0578 (FAX)

{jasim, dsbaero}Oumich.edu

tive feedforward methods cancel the imbalance forces


without affecting the stability of the system. By using adaptation techniques, these methods identify and
cancel the effects of imbalance [7]-[9]. However, these
methods do not utilize measurements of the spin rate of
the rotor but rather rely on frequency response models of the rotor. This frequency dependence thus requires that the algorithms re-adapt at each rotational
frequency.

Abstract

A planar rotor is asymptotically stabilized using an


adaptive feedback control law that does not require
knowledge of the location of the center of mass of the
rotor. A Lyapunov-based proof is used to show that
asymptotic stabilization is achieved. Numerical examples illustrate the ability of the control algorithm to
achieve asymptotic stabilization for various cases including changing center of mass and rotor speed.

In the present paper we develop an active force balancing feedback controller for planar rotors based upon
adaptive control principles and Lyapunov techniques.
This adaptive balancing controller, given by Theorem
1, is a 2nd-order dynamic compensator whose internal
states include estimates of the rotor center of mass.
This algorithm is closely related to the adaptive spacecraft controller considered in [lo].

1 Introduction

Virtually all rotating components suffer from imbalance


to some extent. In general, rotor imbalance causes shaft
vibration which leads to bearing deterioration and the
need for increased maintenance and possibly premature
failure. For example, a significant portion of aircraft
turbine maintenance can be directly attributed to excessive vibration levels, much of it due to rotor imbalance.

In Section 2 we derive the governing equations of motion for the rotor. In Section 3 we present the adaptive
force balancing algorithm. In Section 4 we numerically
illustrate the ability of the controller to asymptotically
stabilize the system when the rotor speed changes continuously and when the imbalance is changed abruptly.

The effects of rotor imbalance can be mitigated by a variety of means including support isolation as well as passive and semi-active mass balancers [l]-[3].However,
for high-precision applications, active force balancing
methods using feedback control may be required. In
active force balancing, forces are applied to the shaft
by magnetic bearings fixed in a nonrotating frame.

2 Equations of Motion

Consider a planar rotor with horizontal axis of rotation


supported by a radial active magnetic bearing (Figure
1). Let ( I , J ) be an inertial frame k e d at 0 and let the
rotor center P be given in this frame by its Cartesian
coordinates ( X , Y ) . Let ( i , j ) be a frame k e d to the
rotor at P where the angle between i and I is 8 (Figure
1). The rotor mass center C is fixed in (i,j ) and is given
in this frame by its Cartesian coordinates ((~1,a2). The
magnetic forces acting on the rotor in the I and J directions are denoted by F, and F,, respectively. The

Numerous active force balancing control algorithms


have been developed. Feedback methods [4]-[6] view
the imbalance as a disturbance. These methods achieve
disturbance rejection by modifying the rotor dynamics at the rotational frequency of the rotor, and thus
may encounter stability problems if the rotational frequency changes. On the other hand, open-loop or a d a p
'Research supported in part by the Air Force Office of Scientific
Research under Grant F49620-98-1-0037and the University of Michigan office of the Vice President for Research.

0-7803-5250-5/99/$10.000 1999 IEEE

773

J'

where the positive constants Kh and Ki, called the


displacement stiffness and current stiffness, depend on
bearing characteristics such as air gap thickness, bias
currents or field strength, and permeability of the material used. Detailed discussions on modeling active
magnetic bearings can be found in [ll]and [12].

Next we use rate and displacement feedback to stabilize


the rotor under the assumption that the rotor is balanced, that is, a1 = a2 = 0. To do this we assume that
M , Kh and Ki are known and let the control currents
be given by
\

c.

equations of motion are given by

where IC1 and k 2 are positive constants to be chosen,


and '211 and 212 are the control variables. Using (4), (5),
(6) and (7) we obtain

+ R2a1 cos(Rt)
- R 2 a 2 sin(Rt) + u1,
-kzY - I C ~+Ya2a1sin(Rt)
412

-02a2

where g is the acceleration due to gravity which is assumed to act in the negative J direction, it4 > 0 is the
mass of the rotor, and R = R(t) is the speed of the
rotor. The time history of R may be predetermined as
in a preprogrammed spin up or spin down maneuver,
or it can vary in response to the dynamics of the other
variables. However, for subsequent analysis, we assume
that R is constant and O(0) = 0. In this case (1) - (3)
become
X
Y

R2al

n2a1sin(Rt)

+Fy/M - 9 ,

(5)

Note that for simplicity the time argument is s u p


pressed except for the imbalance terms.

In general, F, and Fy are nonlinear functions of rotor


displacement and control currents. For small rotor displacements and small control currents, we approximate
F, and Fy by the linear model

F,

F9

= K h x Kii,,
= KhY+Kiiy,

212.

(11)

Remark 1. Although in practice bearing models are


more complex than (6) and (7), it is nevertheless assumed that an inner-loop control has been found for a
nominally balanced rotor such that the controlled rotor
dynamics with imbalance can be approximated to first
order by (10) and (11). Numerous magnetic bearing
control designs have been developed for this purpose;
see, for example, [12]-[14].

cos(Rt)

(10)

It is clear that if a1 = a 2 = 0 then the control law


given by (8) and (9) asymptotically stabilizes the rotor motion with u1 = '212 = 0. Thus u1 and 212 can
be used as outer-loop control variables for imbalance
compensation.

cos(Rt) - R2a2 sin(Rt) F,/M, (4)


o2a2c o s ( ~ t )

- k2X

3 Adaptive Rotor Stabilization

In this section we present a feedback control law that


stabilizes the planar rotor to the origin, i.e., achieves
X -+ 0, Y + 0 as t -+ CO. The control law does not
require any knowledge of the location of the center of
mass (a1,a2) of the planar rotor.

(6)
(7)

774

Since a1 and a 2 appear linearly in (10) and ( l l ) , we


can rewrite (10) and (11) as

derivative of V along the trajectories of the system is


given by
V(21,22,@)= -2TR21 - ZTRZ2,

(27)

which shows that V is negative semi-definite and


also does not explicitly depend on time.
Since
V(21(t),Z2(t),P(t)) 5 ~ ( 2 ~ ( 0 ) ~ 2 2 ( ~ ) , ~2( ~
0, ) ) ~
and since V is radially unbounded, it follows that 21,
2 2 and
are bounded. Thus & is bounded.
.

where

Next note that the total time derivative of V along


trajectories of the system is given by
V(21,22,p,t)

-2Z,TR[AZi
-2Z,TR[A22

Theorem 1: Let r E 8 2 x 2 , let R E 8 2 x 2 be


tive definite, and let p E 8 2 x 2 be the positive definite
solution to the Lyapunov equation

The control law (20), (21) is a 2nd-order integral compensator. Note that the control law requires no h o w l edge of the location of the center of mass of the rotor.
The control law (19)-(21) is adaptive in the sense of
[15], Chapter 1, where the state & is termed the adaptive parameter.

(18)

Then the control law

& =

r-l[u,T(t)Pzl

+ u?(~)Pz~],

u1 = -Q2[cos Rt - sin ~ t ] & ,


'112 = R2[sinRt
cos at]&,

(19)
4 Numerical Simulation

(20)
(21)
(22)

In this section we present numerical examples to illustrate the control law developed in Section 3. We
consider various scenarios including abrupt changes t o
the center of mass and varying rotor speed. The initial conditions and gains, which are the same for each
case considered, are X ( 0 ) = 2 mm, X ( 0 ) = 1 mm/sec,
Y ( 0 )= -1 mm, Y ( 0 ) = 2 mm/sec, &(O) = [-0.2 0.3IT
mm, IC1 = 6 sec-l, k2 = 125 sec-2 and R = Iz, where
I 2 is the 2 x 2 identity matrix. In each of the cases
considered we enable the control law at t = 0.5 sec.

achieves 21 + 0 and 2 2 + 0
t + ca. Furthermore,
& is bounded for all t 2 0 and Li -+ 0 as t -+ ca.

Proof: Define P = a - & and rewrite (12), (13), (19),


(20) and (21) as
21
22

+
+

= A21 U i ( t ) p ,
= A22 U2(t)P,
= -r-l(u:(t)m

(23)

+ u,T(t)pz2).

(24)

(25)

For this example the location of the center of mass relative t o the rotor center is given by a1 = 0.5 mm and
a 2 = -0.1 mm. The rotor rotates at a constant speed
of R = 1000 rpm. We observe from Figures 2 and 3
that X + 0, X + 0, Y + 0, and Y + 0. Figures 4
indicates the control effort. Thus the control law given
by Theorem 1 achieves its objective of asymptotically
stabilizing the rotor.

Next consider the candidate Lyapunov function V :


x R2 x X2 3 8 defined by

z2

V(Zi,Z2, p) = 2;'PZi

+ ZTP22 + PTr@.

+ Vz(t)P]. (28)

Since 21, 2 2 , P are bounded, and since U1 (t) and U2(t)


are bounded, it follows that V is bounded. Using Theorem 5.4 of [15] we conclude that 21 + 0 and 2 2 + 0.
Furthermore, since 21 + 0. and 2 2 + 0 it follows from
0
(25) that + 0 and thus & + 0.

The control objective is to specify U = [ul u2IT so that


21 -+ 0 and 2 2 + 0 as t + ca.

A ~ P P
+ A + R = 0.

+ Ui(t)p]

(26)

Note that although (23) - (25) is time varying, V does


not explicitly depend on time. Using (18) and (25) the

775

-1

0.1

secfd

--I

I
25

Figure 5: Displacement and velocity of rotor center in the


X direction

Next we let the center of mass of the rotor change


abruptly at t = 2 sec. The position of the center of mass
relative t o the rotor center is changed from a1 = 0.5
mm and a2 = -0.1 mm t o a1 = 4 mm and a2 = 3
mm. The rotor continues t o spin at a constant speed of
R = 1000 rpm. We observe from Figures 5 and 6 that
the controller is successful in achieving its objective of
X + 0, X + 0, Y + 0, and Y + 0. Figure 7 shows
that the adaptive parameter readjusts itself after the
center of mass imbalance is changed.
Finally, we accelerate the rotor linearly from R = 0 rpm
at t = 0 sec t o R = 2000 rpm a t t = 2 sec. The position
of the center of mass relative t o the rotor center is kept
constant at a1 = 0.5 mm and a2 = -0.1 mm. Note that
the rotor spinup violates the assumption of constant
rotor speed in Theorem 1. Nevertheless, Figures 8 and
9 suggest that the controller asymptotically stabilizes
X and Y despite the changing rotor speed. The control
effort is shown in Figure 10.

Figure 3: Displacement and velocity of rotor center in the


Y direction
I

200,

150

@a loo
0

\ 5 0

-50

5 Conclusions

-im
-150

A planar rotor is asymptotically stabilized using an

I
05

sec1.

2.5

adaptive feedback control law that requires no knowledge of the position of the center of mass. Current
research is focusing on the extension to multiple plane
force balancing.

Figure 4: Control effort 211 and 212

776

P SP

777

30,
I:

..

20

10

v)

[8] Larsonneur, R., and Herzog, R., Feedforwaxd


Compensation of Unbalance: New Results and Application Experiences, Proceedings, IUTA M Symposium on
the Active Control of Vibration, Bath, U.K., September
1994.

[9] Shafai, B., Beale, S., LaRocca, P., and Cusson,


E., Magnetic Bearing Control Systems and Adaptive
Forced Balancing, IEEE Contr Sys Mag, Vol. 14, No.
2, pp. 4-13, 1994.

E
0

-10

[lo] J. Ahmed, V. T. Coppola and D. S. Bernstein,


Asymptotic Tracking of Spacecraft Attitude Motion
with Inertia Matrix Identification, AIAA J. Guid.
Contr. Dyn., Vol. 21, pp. 684-691, 1998

-20

Figure 10: Control effort 111 and

[ll] Imblach, J., Blair, B. J. and Allaire, P. E., Measured and Predicted Force and Stiffness Characteristics
of Industrial Magnetic Bearings, Tkans ASME, Vol.
113, pp. 784788, October 1991.

112

[12] Siegwart, R., Design and Application of Active


Magnetic Bearings for Vibration Control, Vibration

References
[l] Thearle, E. L., A New Type of Dynamic Balancing Machine, Tkans. ASME, Vol. 54, pp. 131-141,
1932.

and Rotor Dynamics, von Karnaan Institute for Fluid


Dynamics Lecture Series 1992-06, Sep. 21-25, 1992.
[13] Humphris, R. R., Kelm, R. D., Lewis, D. W.,
and Allaire, P. E., Effect of Control Algorithms on
Magnetic Journal Bearing Properties, Pans ASME,
Vol. 113, pp. 784788, October 1991.

[2] Inoue, J., Hayashi, S., and Araki, Y., On the


Self-synchronizationof Mechanical Vibrators, Bulletin
of the Japan Society of Mechanical Engineers, Vol. 10,
NO. 41, pp. 755-762,1967.

[14] Kim, C. S., and Lee, C. -W., Isotropic Optimal Control of Active Magnetic Bearing System, Pro-

Bovik, P., and Hogfors, C., Autobalancing of


Rotors, Sound and Vib, Vol. 111, No. 3, pp. 429-440,
1986.

[3]

ceedings, Fourth International Symposium on Magnetic


Bearings, pp. 35-40, August 1994.
[15] K. J. Astrom, and B. Wittenmark, Adaptive Control, Addison-Wesley, Reading, MA, 2nd edition, 1995.

[4] Higuchi, T., Mizumo, T., and Tsukamoto, M.,


Digital Control System for Magnetic Bearings with
Automatic Balancing, Proceedings, 2nd International
Symposium on Magnetic Bearings, pp- 27-32, Tokyo,
Japan, July 1990.
[5] Mizuno, T., and Higuchi, T., Design of Magnetic
Bearing Controllers Based on Disturbance Estimation,

Proceedings, 2nd International Symposium on Magnetic


Bearings, pp. 281-288, Tokyo, Japan, July 1990.
[6] Mohamed, A. M., and Busch-Vishniac, I., Imbalance Compensation and Automatic Balancing in
Magnetic Bearing Systems Using the q-Parametrization
Theory, IEEE %ns. Contr Sys Tech, Vol. 3, No. 2,
pp. 202-211,1995

[7] Knospe, C. R., Hope, R. W., Fedigan, S. J., and


Williams, R. D., New Results in the Control of Rotor Synchronous Vibration, n Technical Report, Center
for Magnetic Bearings, University of Virginia, Charlottesville, VA, 1994.

778

You might also like