2009 Second International Conference on Computer and Electrical Engineering

State Space Analysis and Control Design of Two-Mass Resonant System

Ghazanfar Shahgholian, Pegah Shafaghi, Mansoor Zinali, Sepehr Moalem

Department of Electrical Engineering - Islamic Azad University
Najaf Abad Branch
Esfahan, Iran
[email protected]

Abstract— In this paper a two-mass resonant system with a

speed controller by resonance ratio control (RRC) has been
simulated. The state space technique for two-mass system is
used. A linear transfer function model is also developed. The
gains of the controller can be designed by coefficient diagram
method (CDM). The simulation results are illustrated that the
proposed controller improves dynamic performance. Figure 1. Two-mass resonant system

Keywords- Two mass system; speed control; resonance ratio

control; optimal criterion. Such systems are often found in the steel mill drives. The
state equation of the two-mass resonant system is as
I. INTRODUCTION follows:
In general motion control application the flexible system
composed of a motor and load with flexible shaft. In d B 1 1
ωM = − M ωM − TS + TM (1)
mechanical resonance system, the measurement of system dt JM JM JM
variables is hard to obtain. When not all system variables are B B B B
d
measurable, the controller design system is very difficult [1- TS = (K S − M S ) ω M − (K S − L S ) ω L
2]. A simplest model of a multi-mass resonance system is dt JM JL
two-mass resonance system. Various controllers for two- 1 1 B B
mass such as classical control [3, 4], approximations and − BS ( + ) TS + S TM + S TL (2)
JM JL JM JL
system identification [5], resonance ratio control [6, 7] and
optimal control [8] are proposed. A systematic comparative d B 1 1
study of compensation schemes such as resonance ratio ωL = − L ωL + TS − TL (3)
dt JL JL JL
control and PID control for the coordinated motion control of
two-inertia mechanical systems presented in [9]. Neural
estimators of the load speed and torsional torque in the open- where:
and closed-loop control structure proposed in [10], which JM, JL: the motor inertia, the load inertia
additional feedbacks from the shaft torque and the difference BM, BL: the motor viscous damping coefficient, the load
between the motor and the load speeds is used. The viscous damping coefficient
tasks of speed controller for two-mass resonant system are KS: the shaft stiffness
suppression of shaft torsional vibration, faster tracking the ωM, ωL: the motor speed, the load speed
load speed of the speed reference without overshoot, TM, TS, TL: the motor torque, the shaft torsional torque, the
rejection the effect of the load disturbance torque and robust load disturbance torque
controller. In this paper, a resonance ratio control using by Increasing the inertia ratio KJ=JL/JM and shaft stiffness
optimal criterion for two-mass system is design. The will decrease the mechanical vibration of the system.
remainder of this paper is organized as follows. In section 2, With the state equations the circuit can be easily modeled
the state-space model of two-mass mechanical system in by using the functional blocks. The simplified block
open-loop is demonstrated. In section 3, the control structure diagram of the two-mass system is shown in Fig. 2. Genera-
with a feedback signal from the shaft torque is described.
lly, the speed ωM and position θM of the motor shaft differ
The simulation results in the open- and close-loop system are
show in section 4. Then, brief summary follow in section 5. from the respective variables ωL and θL, on the load side.
During transients, speeds of motor and load differ, and
II. TWO-MASS MECHANICAL SYSTEM torsional torque is given by:
The two-mass resonant system which has a motor and a
load connected with a flexible shaft is shown in Fig. 1. TS (s) = BS [ω M (s) − ω L (s)] + K S [θ M (s) − θ L (s)] (4)

978-0-7695-3925-6/09 $26.00 © 2009 IEEE 97

DOI 10.1109/ICCEE.2009.105
constant decreases the ωR without affecting the ωA. Conver-
sely, increasing the mechanical stiffness of the shaft (KS)
increases both ωR and ωA, and also the mechanical bandwid-
th in closed-loop system. Therefore, in the large systems,
which have large inertias, generally produce low natural
frequencies. The system parameters used in this paper for
two cases are listed on Table I.
TABLE I. TWO-MASS SYSTEM PARAMETERS
Symbol Value Unit
KS 138 Nm/rad
BS 0.1000 Nm/rad/s
JM 0.0480 kg.m2
Figure 2. Block Diagram of the two-mass mechanical system
BM 0.0013 Nm/rad/s
JL 0.0086 kg.m2
BL 0.0690 Nm/rad/s
The transfer function from TM to ωM, which plays an KJ 0.1792 -
important role in the closed loop design, is given by: ωA 126.7 rad/s
ωR 137.6 rad/s

ω M (s) 1 J L s 2 + (B L + B S ) s + K S
H M (s ) = = (5) Mechanical systems are notoriously under damped, hence
TM (s) J L J M Δ (s ) ascertaining the damping factors amounts to an educated
guess. The dominant eigenvalues for open-loop system are
where the characteristic equation of the open loop system is P1=-1.24 and P2,3= -10.26±j137.14. The damping factor of
given as: the original plant without the controller is η=0.0498. The
step response of shaft torque due change in motor torque
B M + BS B L + BS 2 and load torque without controller is shown in Fig. 3. In this
Δ (s) = s 3 + ( + )s figures, the dashed line and the thick solid line show
JM JL
corresponds relative to motor torque and relative to load
K S (J M + J L ) + B S (B M + B L ) + B M B L torque. Figs. 4-5 show the open-loop frequency response
+( )s
JMJL from the motor speed, shaft torque and load speed to the
K S (B M + B L ) motor torque and to load torque, respectively.
+ (6)
JMJL III. RESONANCE RATIO CONTROL
In this section various close-loop transfer functions
Since damping losses usually considered being relatively describing the dynamic of the system following change in
low, they are neglected with out significantly affecting the reference speed and disturbance torque are derived. Fig. 6
accuracy of the forgoing analysis. Neglecting friction terms, show the complete block diagram of the two-mass system
the transfer function HM(s) is given by: with resonance ratio controller. The controller structure
consists of an integral controller and proportional
1 s 2 + ω 2A compensation with a additional feedback signal from the
H M (s) = (7) shaft torque with transfer function GR(s).
J M s (s 2 + ω 2R )

We see that ωR represents the mechanical resonance

frequency of the open-loop system. The anti-resonance
frequency ωA is lower thanωR.

KS
ωA = (8)
JL
1 1
ωR = K S ( + ) (9)
JM JL

The control bandwidth in closed-loop motion control Figure 3. Step response of shaft torque in open-loop system
system is limited by the ωA. An increase in the motor inertia

98
Figure 4. Frequency response characteristics of transfer function relative Figure 5. Frequency response characteristics of transfer function relative
to motor torque to motor torque

Figure 6. Block diagram of control system

The motor speed and load speed are depends on the ω M (s) s G S (s) G M (s) G L (s) [1 + G R (s)]
disturbance torque TL and reference speed ωC: H RT (s) = = (14)
TL (s) Δ (s)
ω L (s) G L (s)
ω M (s) = H RS (s) ω C (s) − H RT (s) TL (s) (10) H RL (s) = =
TL (s) Δ (s)
ω L (s) = H RC (s) ω C (s) − H RL (s) TL (s) (11)
[s + G M (s) (sK P + K I + s G S (s) + s G S (s) G R (s))] (15)
If GS(s), GM(s) and GL(s) are the transfer functions of
shaft, motor and load, respectively, the close-loop transfer where Δ(s) is given by:
functions of the reference speed to motor speed and the
reference speed to load speed (tracking dynamics) can be Δ (s) = [1 + G S (s) G L (s)][s + (K I + K P s) G M (s)]
expressed in (12) and (13), respectively: +s G S (s) G M (s) [1 + G R (s)] (16)

ω M (s) K I G M (s) [1 + G S (s) G L (s)] If the resonance ratio control is PD, GR(s)=KC+KDs, the
H RS (s) = = (12)
ω C (s ) Δ (s) state-space equations of two-mass system with controller are
ω L (s) K I G M (s) G S (s) G L (s) given by:
H RC (s) = = (13)
ω C (s ) Δ (s)
d K K K B (B + K P ) K P + B M
ω M = [− D S + D S M − ]ω M
The close-loop transfer functions of the disturbance torque dt JM JM JM JM
to motor speed and disturbance torque to load speed K D BS K C 1 K +1
(regulation dynamics) can be expressed in (14) and (15), +[ ( + ) − C ] TS
JM JM JL JM
respectively:

99
KD B B 1 B ω 2A τ 2 J γ3 γ3 γ 1
+ [K S − S L ] ω L + E V + S TL (17) K C = [−J M ω 2A + ( − 1) M 21 23 3 − K S ] (24)
JM JL JM JL γ1 ωA τ KS
d −B L 1 1 J M γ 13 γ 32 γ 3 ω 2A τ 2
ωL = ωL + TS − TL (18) KD = ( − 1) (25)
dt JL JL JL K S ω 2A τ 3 γ 12 γ 2
d B (B + K P )
TS = [K S − S M ] ωM
dt JM In this criterion, for KC>0 and KD>0, the stability index
BS B L K 1 B must be (26) and (27), respectively.
+ [−K S + ] ω L − B S ( C + ) TS + S TL (19)
JL JM JL JL
τ 2 ω 2A
d γ2 < (26)
E V = K I (ω C − ω M ) (20) γ 12
dt
(K J + 1) τ 4 ω 4A
γ3 > (27)
where KC and KD are gains proportional and derivative of γ 12 γ 22 (τ 2 ω 2A − γ 1 )
controller.
The characteristic equation polynomial in close-loop with IV. SIMULATION RESULTS
RRC is given by: The simulation results of the speed control in the two-
mass using the proposed controller will be shown in this
Δ RRC (s) = J M s 4 + (K S K D + K P ) s 3 section in order to demonstrate the efficiency of the
+ (J M ω2A + K I + K S K C + K S ) s 2 + K P ω2A s + K I ω 2A (21) controller. In order to confirm the performance of the
controller, we accomplish a simulation study using Matlab
Simulink as shown in Fig. 7.
The Coefficient Diagram Method (CDM) is an indirect
The controller parameters from CDM criterion without
pole placement method to design an appropriate characteris-
and with derivative of the proposed control scheme of the
tic polynomial. The CDM can give a controller design
system are listed in Tables II and III, respectively. In to
which is both stable and robust, and it has the desired
cases the equivalent time constant is 0.0304 sec.
system response speed. CDM needs some design parameters
with respect to the characteristic polynomial coefficients TABLE II. CONTROLLER PARAMETERS WITHOUT DERIVATIVE
which are the equivalent time constant (τ) and the stability index
index (γk) [11]. The controller gains obtained from CDM KP KI KD KC
stability
criterion as followers: γ1=2.5
γ2=2.37 9.37 308.09 0 4.4292
3 3 γ3=1
JM γ γ γ3
1 2 γ1=3
KI = 2 4
(22)
ω τ
A
γ2=1.65 12.49 410.79 0 5.1564
3 3 γ3=1.6
JM γ γ γ3
1 2 γ1=2.5
KP = (23)
2
ω τ 3 γ2=2.37 14.05 462.14 0 9.9345
A
γ3=1.5

load speed
integral load
LOAD controller mechanical angle
1 1

shaft JL.s+BL shafte damping s

torsional torque load cofficient
disturbance torque
BS

KS
PID
integral shaft stiffness
controller PD controller MOTOR
KI 1 1
s JM.s+BM s
motor torque motor
integral mechanical angle
proportional motor torque controller
controller
KP
motor speed

Figure 7. Block diagram of control system in Matlab/Simulink

100
TABLE III. CONTROLLER PARAMETERS WITH DERIVATIVE
index
KP KI KD KC
stability
γ1=2.5
γ2=2 6.62 217.49 0.0091 1.2215
γ3=1
γ1=2.5
γ2=1.8 8.05 264.25 0.0188 2.90
γ3=1.5
γ1=3
γ2=1.6 11.72 384.84 0.0028 4.46
γ3=1.6

The dominant eigenvalues for close-loop system are P1,2=-

101.44±j144.85 and p2,3=-66.13±j23.81 for without derivati-
ve and p1,2=-51.78±j130.57, p3=-128.78 and p4=-50.93 for
with derivative. Fig. 8 show the step response of motor
speed in closed-loop system with RRC controller with and
without derivative. Fig. 9 show the frequency response
characteristics of transfer function of the reference speed to
motor speed. Figure 10. Frequency response characteristics of transfer function of the
reference speed to motor speed

REFERENCES
[1] G.Shahgholian, J.Faiz, “An analytical approach to synthesis and
modeling of torque control strategy for two-mass resonant systems”,
Inte. Revi. of Auto. Cont. (IREACO), Vol.2, No.4, pp.459-468, July
2009.
[2] K.Erenturk, “Nonlinear two-mass system control with sliding-mode
and optimized proportional-integral derivative contoroller combined
with a grey estimator”, IET Contr. The. Apll., Vol.2, No.7, pp.635-
642, 2008.
[3] A.Shoulaie, M.Bayati-Poudeh, G.Shahgholian, “Damping torsional
torques in turbine generator shaft by novel PSS based on genetic
algorithm and fuzzy logic”, Jou. of Elec. and Pow. Engi. (JEPE),
Figure 8. Step response of motor speed in closed-loop system with RRC Vol.1, No.1, pp.3-10, Winter 2009.
controller [4] Y.Takeichi, S.Komada, M.Ishida, T.Hori, “Speed control of
symmetrical type three-mass resonant system by PID controller”,
IEEE/AMC, Vol.2, pp.594-599, March 1996.
[5] G.Shahgholian, P.Shafaghi, “PID controller for torque control in two-
mass resonant system”, Pro. Int. Univ. Pow. Eng. Con. (UPEC), Sep.
2009.
[6] S.Hashimoto, K.Hara, H.Funato, K.Kamiyama, “AR-based
identification and control approach in vibration suppression”, IEEE
Tran. On Ind. Appl., Vol.37, No.3, pp.806-811, May/June 2001.
[7] B.Nam, H.Kim, H.Lee, D.Kim, “Optimal speed controller design of
the two-inertia stabilization system”, PWASET Vol.31, pp.155-160,
July 2008.
[8] S.Katsura, K.Ohnishi, “Force servoing by flexible manipulator based
on resonance ratio control”, IEEE Trans. On Indu. Elec., Vol.54, No.1,
pp.539-547, February 2007.
[9] Y.Hori, H.Sawada, Y.Chun, “Slow resonance ratio control for
Figure 9. Shaft torque in closed-loop system with RRC controller vibration suppression and disturbance rejection in torsional system,”
IEEE Trans. on Ind. Elec., Vol.46, No.1, pp.162–168, February 1999.
V. CONCULSION [10] D.Szabo, S.Kerekes, “A fuzzy sliding mode approach for the two
mass system”, IEEE/ISIE, pp.348-352, 1999.
A space-state mathematical analysis and design methods [11] T.M.O’Sullivan, C.M.Bingham, N.Schofield, “High performance
for a two-inertia system is develop. The simulation results in control of dual-inertia servo drive systems using low cost integrated
controlling the speed of the-mass system by the proposed SAW torque transducers”, IEEE Tran. on Indu. Elec., Vol.55, No.4,
controller are investigated to verify the effectiveness of the pp.1226-1237, August 2006.
[12] S.Manabe, “Importance of coefficient diagram in polynomial
proposed method. method”, IEEE/DCM, pp.3489-3494, December 2003.

101

View publication stats