Professional Documents
Culture Documents
DAMPING INJECTION CONTROL OF WINDING SYSTEMS BASED ON PO - 2007 - IFAC Proceedin
DAMPING INJECTION CONTROL OF WINDING SYSTEMS BASED ON PO - 2007 - IFAC Proceedin
HAMILTONIAN SYSTEMS
a
Groupe de recherche en électronique industrielle, Department of ECE
Université du Québec à Trois-Rivières, Trois-Rivières (Québec), Canada, G9A 5H7
b
Defense Research and Development Canada, Québec (Québec), Canada, G3J 1X5
Abstract: In web and metal sheet processing systems and winders, flexibility of the web is
a source of vibrations and resonance amongst motor drives. Port-Controlled Hamiltonian
with Dissipation (PCHD) modelling is considered to develop stabilization strategies with
a physical interpretation and motivation of the control action. A web transport system is
modelled as a PCHD system and the control action is defined to obtain asymptotically
stable operating points for the controlled system by a passivity argument. The controller
negative output feedback gain matrix is interpreted as the realization of virtual dampers
added to the system. Copyright © 2007 IFAC
b 12 b32
Design of the controllers for the first step follows the Power M1 M2 M3
Flow
procedure described by Thiffault et al. (2005),
without considering the dampers that will be added
b1 b2 b3
in the second step. The second step of the design is
presented, in particular to inject damping to obtain
disturbance rejection properties: the virtual dampers
(b) Cross coupled structure
detune the controllers for disturbance rejection.
Fig. 2. Equivalent system with damping injection.
Master-slave control structures are commonly used Legend: Thick lines represent the web, double
in web transport systems. In master speed control, arrows represent energy exchange and small
the velocity of a selected driven roller is set at a pre- circles represent measurement points without
defined constant value (‘‘master speed’’). Generally, energy exchange.
a feedback control system is employed to produce
accurate control of the velocity of the selected driven
roller. In this case, the velocity of the output shaft of is satisfied for
the motor driving the selected roller is measured and
compared continuously with its reference. Other bi ≥ 0 (i=1, 2, 3) and
(23)
drives may be slaved to the master (Shin, 2000). b2 – (b12 + b32)/4 ≥ 0.
Two realizations are considered for matrix C (C1 and System (11) in closed loop is passive, and is
C2) for the system of figure 1, with M2 identified as stabilized with the output feedback controllers.
the master that imposes speed of the web:
⎛ b1 + b12 − b12 0 ⎞ 4.2 Interpretation of the damping injection
⎜ ⎟
C1 = ⎜ 0 b2 0 ⎟ (18) structures.
⎜ 0 − b32 b3 + b32 ⎟⎠
⎝ Using matrix C=C1 (18),(20) is physically interpreted
as inserting virtual dampers at the locations shown in
⎛ b1 + b12 − b12 0 ⎞ figure 2a. Each control input ui contains velocity
⎜ ⎟
C2 = ⎜ − b12 b2 + b12 + b32 − b32 ⎟ (19) feedback term –bivi/ri, which represents a damping
⎜ 0 − b32 b3 + b32 ⎟⎠ torque exerted upon roll i; it can be represented by a
⎝
virtual damper with coefficient bi for each motor Mi.
The respective control laws are defined by Feedback terms in velocity differences in control
inputs u1 and u3 represent damping torques acting
⎧U 1 =−b1 y1 −b12 ( y1 − y 2 ) upon rolls 1 and 3, due in particular to variations of
⎪
C1: ⎨U 2 =−b2 y 2 (20) velocity of motor 2. The energy required by the
⎪U =−b y −b ( y − y ) virtual dampers is injected or absorbed by the
⎩ 3 3 3 32 3 2
controller but no energy corresponding to these terms
is exchanged with motor M2. Motor M2 is identified
⎧U 1 =−b1 y1 −b12 ( y1 − y 2 ) as the master (is not affected by the dampers) and the
⎪ (21) two other motors are identified as slaves (they are
C2: ⎨U 2 =−b2 y 2 −b12 ( y 2 − y1 )−b32 ( y 2 − y1 )
⎪U =−b y −b ( y − y ) affected by the virtual dampers).
⎩ 3 3 3 32 3 2
The same interpretation can be used when using
With feedback controllers (20) and (21) and outputs C=C2 given by (19) and (21). However, mutual
y (16), dissipation inequality interactions (cross-coupling) between every pair of
adjacent motors, result into the structure of the
m
(dH /dt )≤∑ yiU i ≤0 (22) equivalent system of figure 2b: all motors are
i =1 affected by the virtual dampers. We associate this
structure to a master-slave structure with coupling.
-3
x 10
6 5
Table 1 Main system parameters
Speed error(m/s)
4
speed(m/s)
E S L1 = L2 r1 = r 3 2 0
(N/m2) (m2)
Velocity reference
(m) (m) 0 Without Damp.Contr
Speed error(m/s)
Speed error(m/s)
Coupled structure(b2=0)
0.15 10 2.15 0.04 0.2 0.2 Coupled structure (Damp.Contr)
0 0
parameters -0.4
0 2 4 6
-0.4
0 2 4 6
Time (s) Time (s)
T1ref = T2ref v2ref b1 = b3 b2 b12 = b32 Fig. 3. Web velocity reference and error signals.
(N) (m/s)
100 5 100 6·105 2·105 150
110
Tension T1(N)
Tension T1(N)
100 105
0 95
0.2 0.4 0.6 0.8 1 1.2 5 5.2 5.4
The simulation results were obtained in the Matlab- Time (s) Time (s)
SimulinkTM environment with a model comprising 101
150
not only the web model, but also the model of the
Tension T2(N)
Tension T2(N)
100.5
motor drives and of their controllers, which 100
100
implement field oriented control. PI velocity 50
99.5
controllers and IP tension controllers are used to 0
99
implement the structure described in figure 1; the 0.2 0.4 0.6 0.8 1 1.2 5 5.05 5.1
Time (s) Time (s)
effect of adding the damping controller is then
evaluated. In particular, the two implementations on (a) Tension step and (b) Load step torque
the damping controller are evaluated with respect to velocity ramp. disturbance on M1.
a tension set point variation followed by ramping in Fig. 4. Unwind tension T1 and Winder tension T2
web velocity and then a load disturbance equivalent responses without damping controller.
to 10 N applied on M1. The main parameters of the
winding system, based on (Pagilla et al., 2003), and 100
106
of the damping controller are given in tables 1 and 2. 104 Master-slave structure
Tension T1(N)
Tension T1(N)
80
Coupled structure
60 102
Tension T2(N)
80
significantly affect tension and cause oscillations. 60
100
Smoothing the speed reference signal (e.g. S-profile) 40 Master-slave structure 99.8
would alleviate the problem for velocity set point 20 Coupled structure Master-slave structure
Coupled structure
99.6
variations, but not for load torque disturbance. 0
0.5 1 1.5 5 5.2 5.4 5.6
Time (s) Time (s)
Figure 5 shows the results with the damping (a) Tension step and (b) Load step torque
controller using the parameters from table 2. velocity ramp. disturbance on M1.
Sensitivity to the disturbances was decreased
significantly with only a marginal effect on the Fig. 5. Unwind tension T1 and Winder tension T2
tension set point response. The response of the two responses with damping controller.
implementations of the damping controller only
differ significantly in the way the load torque of the effect of the imperfect velocity-tension
disturbance propagates to affect T2 in the coupled decoupling as compared to figure 4. One notes that
structure, whilst there is no effect in the master-slave with b2=0, condition (27) is not satisfied; however,
structure. However, in both cases, ramping of the this condition is sufficient and not necessary for
line velocity is slowed down as shown in figure 3. stability, in particular considering that the PI velocity
controllers inject some damping into the system.
Removing the damping term b2 on motor M2 has
little effect in the coupled structure (figure 6) but this
term is very important during the velocity ramping 5. CONCLUSION
period in the master-slave structure. With b2=0, web
velocity follows the ramp reference signal closely A PHCD model was established to represent the
(Figure 3) so that there is a much smaller reduction model of a web transport system. Realizations of the
Jeon, S.H., J.M. Kim, K.C. Jung, S.K. Sul and J.Y.
110
Choi (1999). Decoupling control of bridle rolls
for steel mill drive system. IEEE Transactions
Tension T1(N)
Tension T1(N)
105
100
Tension T2(N)
100 100
via µ-analysis. In: Proc. of. 10th IEEE Int.
99.9
50 Conference on Control Applications, 5-7.
Master-slave structure(b2=0) 99.8
Coupled structure(b2=0)
Master-slave structure(b2=0)
Coupled structure(b2=0)
Macchelli, A. (2004). Port Hamiltonian systems, A
0
0.2 0.4 0.6 0.8 1 1.2 5 5.2 5.4 unified approach for modeling and control finite
Time (s) Time (s)
and infinite dimensional physical systems. Ph.D.
(a) Tension step and (b) Load step torque Thesis, University of Bologna.
velocity ramp. disturbance on M1. Maschke, B.M. and A.J. van der Schaft (1992). Port-
controlled Hamiltonian systems: modelling
Fig. 6. Unwind tension T1 and Winder tension T2
origins and system theoretic properties. In: Proc.
responses with damping controller and b2=0.
of the Second IFAC Symposium on Nonlinear
Control Systems Design, NOLCOS92, 282-288.
controllers were interpreted physically as adding Ortega, R., A.J. van der Schaft, B.M. Maschke and
virtual dampers to the system, giving some practical G. Escobar (1999). Energy-shaping of port-
insight in choosing the structure of the control gain controlled Hamiltonian systems by
matrix. The damping controller was used to detune interconnection. In: Proc. of 38th IEEE Conf. on
the velocity and tension controller for disturbance Decision and Control, 1646-1651.
rejection. It appears that the best response would be Ortega, R., M.W. Spong, F. Gómez-Estern and G.
obtained by using a smooth velocity reference Blankenstein (2002). Stabilization of a class of
trajectory to reduce the effect of imperfect velocity- underactuated mechanical systems via
tension decoupling and to use the master-slave interconnection and damping assignment. IEEE
damper structure to reduce propagation of Trans. on Automatic Control, 47, 1218-1233.
disturbances within the system. Pagilla, P.R., I. Singh and R.V. Dwivedula (2003). A
Study on Control of Accumulators in Web
The results can be applied to other types of systems Processing Lines. In: Proc. American Control
such as automated vehicles travelling in convoys or Conf. 3684-3689.
flight formation. Future work includes the use of a Pagilla, P.R, N.B. Siraskar and R.V. Dwivedula
Linear Matrix Inequalities (LMI) approach to tune (2007). Decentralized Control of Web
the gains of the control laws in order to meet the Processing Lines. IEEE Transactions on Control
control objectives (performance, robustness, System Technology, 15, 106-117.
constraints on the actuators …) and the experimental Sepulchre, R., M. Janković and P. Kokotović (1999).
evaluation of the control approach with web Constructive nonlinear control. Springer-Verlag,
processing systems and the vehicle convoy problem. London.
Shin, K.H. (2000). Tension control. Tappi Press,
Atlanta GA.
ACKNOWLEDGEMENTS Thiffault, C., P. Sicard and A. Bouscayrol (2005).
Desensitization to voltage sags of a rewinder by
This work was supported by a grant from Natural using an active dancer roll for tension control.
Sciences and Engineering Research Council of In: Proc. of IEEE, International Electric
Canada. Machines and Drives Conferences, 466-473.
van der Schaft, A.J. (2000) L2-gain and passivity
techniques in nonlinear control, Springer-
REFERENCES Verlag, London.
Wang, C., Y. Wang, R. Yang and H. Lu (2004).
Åström, K.J., P. Albertos and M. Blanke (2000). Research on tension control system based on
Control of complex systems, Springer. neural network. IEEE Transactions on Industrial
Geddes, J.E. and M. Postlethwaite (1998). Electronics, 51, 381-386.
Improvements in product quality in tandem cold Wang, W., D. Cheng, C. Li and Y. Ge (2003).
rolling using robust multivariable control. IEEE Dissipative Hamiltonian realization and energy-
Trans. on Control System Techn., 6, 257-267. based L2 disturbance attenuation control of
Hoshino, I., Y. Okamura and H. Kimura (1996). multimachine power systems. IEEE Tran. on
Observer-Based Multivariable Tension Control Automatic Control, 48, 1428-1433.
of Aluminium hot rolling mills. In: Proc. of 35th Xi, Z., G. Feng, D. Cheng and Q. Lu (2003).
IEEE Conf. on Decision and Control,1217-1222. Nonlinear decentralized saturated controller
Inman, D.J. (1989). Vibration with control, design for power systems. IEEE Transactions on
measurement, and stability. Prentice-Hall Inc. Control System Technology, 11, 539-546.