Академический Документы
Профессиональный Документы
Культура Документы
systems
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 1 / 39
Outline
1 Introduction
4 Conclusions
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 2 / 39
Introduction
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 3 / 39
Introduction
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 4 / 39
Introduction
Linear MPC
Mid-Late 90’s: Linear MPC became a dominant advanced control
technology (Morari and Lee [1997], Young et al. [2001], Qin and
Badgwell [2003])
Properties of centralized linear MPCs well established (Sznaier and
Damborg [1990], Rawlings and Muske [1993], Mayne et al. [2000],
Bemporad et al. [2002])
Efficient large-scale solution strategies available (Antwerp and Braatz
[2000], Bartlett et al. [2002])
Current focus
Possibility of horizontal integration of subsystems’ MPCs to improve
overall system performance
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 5 / 39
Introduction
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 6 / 39
Nomenclature: Consider Two Interacting Units
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 7 / 39
Nomenclature: Consider Two Interacting Units
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 7 / 39
Nomenclature: Consider Two Interacting Units
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 7 / 39
Nomenclature: Consider Two Interacting Units
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 7 / 39
Nomenclature: Consider Two Interacting Units
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 7 / 39
Noninteracting systems
2
n, d, p
1 b
Φ2 (x)
x2 0
a
-1 Φ1 (x)
-2
-2 -1 0 1 2
x1
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 8 / 39
Weakly interacting systems
0.5
Φ2 (x)
0 b n, d
p
-0.5
x2
-1 Φ1 (x)
a
-1.5
-2
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
x1
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 9 / 39
Moderately interacting systems
1.5 Φ1 (x)
a
1
Φ2 (x)
0.5
x2 p
0 b d
-0.5
n
-1
-1.5
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
x1
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 10 / 39
Strongly interacting (conflicting) systems
2
1.5 Φ1 (x)
a
1
Φ2 (x)
0.5 p
x2 0 b d
-0.5
-1
-1.5
-2
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
x1
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 11 / 39
Strongly interacting (conflicting) systems
160
n
140
120
100
80
x2
60
40
20
Φ2 (x)
0 Φ1 (x)
-20
-10 0 10 20 30 40 50
x1
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 12 / 39
Modeling for distributed MPC
Decentralized, interaction models
Decentralized Model
xii (k + 1) = Aii xii (k) + Bii ui (k)
ui yi (k)
(Aii , Bii , Cii )
(local subsystem inputs)
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 13 / 39
Modeling for distributed MPC
Decentralized, interaction models
Decentralized Model
xii (k + 1) = Aii xii (k) + Bii ui (k)
P
ui yi (k) = j Cij xij (k)
(Aii , Bii , Cii )
++
(local subsystem inputs)
Interaction Model
xij (k + 1) = Aij xij (k) + Bij uj (k)
uj6=i
(Aij , Bij , Cij )
(external subsystem inputs)
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 13 / 39
Modeling for distributed MPC
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 14 / 39
Modeling for distributed MPC
x11
2 3
6 . 7
6 . 7
6 . 7
36
6 x
7
y1 C11 ··· C1M
2 3 2
6 1M
7
7
. 7 76 .
7
6
(k) =
6 .. 76 7 (k)
7
6 . 7 6 .
4 . 5 4 . 56
6 . 7
6 7
yM CM1 ··· CMM 6 xM1
6
7
7
6 . 7
6 . 7
4 . 5
xMM
Centralized model
A minimal realization of the composite plant model
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 14 / 39
Distributed MPC
Assumptions and formulations
Assumptions
All MPC cost functions are positive definite, quadratic
Each subsystem represented by a linear, state-space model
All interaction models are stable
Local input inequality constraints (e.g., input bounds)
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 15 / 39
Communication-based MPC1
converge
First move in each converged
input trajectory injected into Process Process
the plant 1 2
y1 y2
1
Similar schemes proposed by Jia and Krogh [2001], Camponogara et al. [2002]
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 16 / 39
Communication-based MPC1
converge
First move in each converged
input trajectory injected into Process Process
the plant 1 2
y1 y2
1
Similar schemes proposed by Jia and Krogh [2001], Camponogara et al. [2002]
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 16 / 39
Communication-based MPC
Distillation column of Ogunnaike and Ray [1994]
0.5
Outputs T21 , T7 ; Inputs L, V 0.25
setpoint
cent-MPC
comm-MPC
Two SISO MPCs 0
-0.25
Intentionally choose bad pairing: T21
-0.5
MPC-1 : T21 − V -0.75
MPC-2 : T7 − L -1
-1.25
Can communication-based MPC fix 0 50 100
Time (sec)
150 200
0.5
V 0
-0.5
-1
-1.5
0 50 100 150 200
Time (sec)
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 17 / 39
Communication-based MPC
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 18 / 39
Cooperation-based MPC
Tasks involved
Model interconnections between subsystems
Exchange state and input trajectories among interconnected
subsystems
Replace local objectives by a suitable global objective e.g.,
X M
X
Φ= w i Φi wi ≥ 0, wi = 1
i i=1
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 19 / 39
Cooperation-based MPC
Intermediate termination
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 20 / 39
Feasible cooperation-based MPC (FC-MPC)
Tasks involved
Model interconnections between subsystems
Exchange input trajectories among interconnected subsystems
Replace local objectives by a suitable global objective e.g.,
X M
X
Φ= w i Φi wi ≥ 0, wi = 1
i i=1
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 21 / 39
Geometry of FC-MPC
Φ2
Φ1
u2
u1
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 22 / 39
Geometry of FC-MPC
Φ2
Φ1
p
u2
u1
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 22 / 39
Geometry of FC-MPC
Φ
Pareto optimal surface
Φ2
Φ1
p
u2
u1
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 22 / 39
Geometry of FC-MPC
Φ
Pareto optimal surface
Φ2
Φ1
p
u2
0
FC − MPC1(u1)
u1
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 22 / 39
Geometry of FC-MPC
FC − MPC2(u2)
Φ
Pareto optimal surface
Φ2
Φ1
p
u2
0
FC − MPC1(u1)
u1
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 22 / 39
Geometry of FC-MPC
FC − MPC2(u2)
Φ
Pareto optimal surface
∗(0)
Φ2
Φ1
p
u2
0
FC − MPC1(u1)
u1
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 22 / 39
Geometry of FC-MPC
FC − MPC2(u2)
Φ
Pareto optimal surface
∗(0)
Φ2
Φ1
p
u2
1
0
FC − MPC1(u1)
u1
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 22 / 39
Feasible cooperation-based MPC (FC-MPC)
Properties
1 All iterates are plantwide feasible
2 The sequence of cost functions is a non-increasing function of the
iteration number
Also bounded below, hence convergent
3 The sequence of iterates converges to an optimal limit point
(centralized MPC solution)
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 23 / 39
Closed-loop properties of FC-MPC
Properties
Nominal closed-loop stability under intermediate termination
Disturbance scenarios that destabilize FC-MPC also destabilize
centralized MPC
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 24 / 39
Performance of FC-MPC
Distillation column of Ogunnaike and Ray [1994]
0.5
Outputs T21 , T7 ; Inputs L, V 0.25
setpoint
cent-MPC
comm-MPC
Two SISO MPCs 0
-0.25
Intentionally choose bad pairing: T21
-0.5
MPC-1 : T21 − V -0.75
MPC-2 : T7 − L -1
-1.25
Communication-based MPC cannot 0 50 100
Time (sec)
150 200
0.5
V 0
-0.5
-1
-1.5
0 50 100 150 200
Time (sec)
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 25 / 39
Performance of FC-MPC
Distillation column of Ogunnaike and Ray [1994]
0.5
Outputs T21 , T7 ; Inputs L, V 0.25
setpoint
cent-MPC
comm-MPC
Two SISO MPCs 0 FC-MPC (1 iterate)
-0.25
Intentionally choose bad pairing: T21
-0.5
MPC-1 : T21 − V -0.75
MPC-2 : T7 − L -1
-1.25
Communication-based MPC cannot 0 50 100
Time (sec)
150 200
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 25 / 39
Integrated styrene polymerization plants
Plant 1
Plant 1:
Fs0 , cs, Tf0 Manipulate Finit0 to control
Frecy, Cmr , Tr
Fm0 , cm0 , Tf0
T1
Finit0 , ci0 , Tf0 End use grade fraction: (1 − β)
Plant 2
Produces grade A (lower
Fc0 , Tc0 grade) of polymer
Fm1 , cm1 , T1 Plant 2:
Fs2 , cs2 , Tf2
Fm2 , cm2 , Tf2
Two units–polymerization
Finit2 , ci2 , Tf2
L D reactor and separator
Fm3 , cm3 , T2
MPC manipulates Finit2 ,
Fc2 , Tc2 V
Frecy and V to control T2 ,
B
Cp, Cmbot , Cinitbot Cmr and Cp
Produces grade B (higher
grade) of polymer
Two MPCs, one for each plant
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 26 / 39
Integrated styrene polymerization plants
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 27 / 39
Integrated styrene polymerization plants
Performance of different MPC frameworks
0 setpoint 0 setpoint
Centralized MPC -1 Centralized MPC
-2 Decentralized MPC Decentralized MPC
-2
-4 -3
T1 T2 -4
-6
-5
-8 -6
-10 -7
-8
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Time (hrs) Time (hrs)
0.05 0.5
Centralized MPC Centralized MPC
0 Decentralized MPC 0.4 Decentralized MPC
-0.05
0.3
-0.1
Finit0 Finit2 0.2
-0.15
0.1
-0.2
-0.25 0
-0.3 -0.1
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Time (hrs) Time (hrs)
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 28 / 39
Integrated styrene polymerization plants
Performance of different MPC frameworks
0 setpoint 0 setpoint
Centralized MPC -1 Centralized MPC
-2 Decentralized MPC Decentralized MPC
FC-MPC (1 iterate) -2 FC-MPC (1 iterate)
-4 -3
T1 T2 -4
-6
-5
-8 -6
-10 -7
-8
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Time (hrs) Time (hrs)
0.05 0.5
Centralized MPC Centralized MPC
0 Decentralized MPC 0.4 Decentralized MPC
FC-MPC (1 iterate) FC-MPC (1 iterate)
-0.05
0.3
-0.1
Finit0 Finit2 0.2
-0.15
0.1
-0.2
-0.25 0
-0.3 -0.1
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Time (hrs) Time (hrs)
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 28 / 39
Integrated styrene polymerization plants
Performance comparison
Λcost Performance loss
(w.r.t centralized MPC)
Centralized-MPC 18.84 -
Decentralized-MPC 1608 8400%
FC-MPC (1 iterate) 18.94 0.54%
FC-MPC (5 iterates) 18.84 0%
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 29 / 39
MPC with partial cooperation
Operational objective. Use u1
y1
u1 Plant
to track y1 , u2 to track y2
Weak interaction Decentralized control gives poor
u2
Strong interaction
control performance
Centralized control uses both u1
y2
and u2
pFC − MPC2
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 30 / 39
MPC with partial cooperation
Operational objective. Use u1
y1
u1 Plant
to track y1 , u2 to track y2
Weak interaction Decentralized control gives poor
u2
Strong interaction
control performance
Centralized control uses both u1
y2
and u2
pFC − MPC2
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 30 / 39
Geometry of pFC-MPC
0.5
0.25 Φ2 (x)
n
0 b d
-0.25
x2 Φ(x)
-0.5 p p0
-0.75
-1 Φ1 (x)
a
-1.25
-1.5 -1 -0.5 0 0.5 1 1.5
x1
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 31 / 39
MPC with partial cooperation
1.5 0.35
1 0.3
0.25
0.5 0.2
u1 0 u2 0.15
-0.5 0.1
0.05
-1 cent-MPC 0 cent-MPC
pFC-MPC pFC-MPC
-1.5 -0.05
0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8
Time Time
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 32 / 39
Conclusions
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 33 / 39
Conclusions
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 33 / 39
Conclusions
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 33 / 39
Conclusions
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 33 / 39
Future research
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 34 / 39
Future research
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 34 / 39
Future research
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 34 / 39
Future research
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 34 / 39
Acknowledgments
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 35 / 39
Further Reading I
J. Antwerp and R. Braatz. Model predictive control of large scale processes. J. Proc.
Control, 10:1–8, 2000.
T. Başar and G. J. Olsder. Dynamic Noncooperative Game Theory. SIAM, Philadelphia,
1999.
R. Bartlett, L. Biegler, J. Backstrom, and V. Gopal. Quadratic programming algorithms
for large-scale model predictive control. J. Proc. Cont., 12(7):775–795, 2002.
A. Bemporad, M. Morari, V. Dua, and E. Pistikopoulos. The explicit linear quadratic
regulator for constrained systems. Automatica, 38(1):3–20, 2002.
D. P. Bertsekas and J. N. Tsitsiklis. Parallel and Distributed Computation.
Prentice-Hall, Inc., Englewood Cliffs, New Jersey, 1989.
E. Camponogara, D. Jia, B. H. Krogh, and S. Talukdar. Distributed model predictive
control. IEEE Ctl. Sys. Mag., pages 44–52, February 2002.
J. E. Cohen. Cooperation and self interest: Pareto-inefficiency of Nash equilibria in finite
random games. Proc. Natl. Acad. Sci. USA, 95:9724–9731, 1998.
P. M. Hidalgo and C. B. Brosilow. Nonlinear model predictive control of styrene
polymerization at unstable operating points. Comput. Chem. Eng., 14(4/5):481–494,
1990.
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 36 / 39
Further Reading II
Y.-C. Ho. On Centralized Optimal Control. IEEE Trans. Auto. Cont., 50(4):537–538,
2005.
D. Jia and B. H. Krogh. Distributed model predictive control. In Proceedings of the
American Control Conference, Arlington, Virginia, June 2001.
D. Jia and B. H. Krogh. Min-max feedback model predictive control for distributed
control with communication. In Proceedings of the American Control Conference,
Anchorage,Alaska, May 2002.
T. Keviczky, F. Borelli, and G. J. Balas. A study on decentralized receding horizon
control for decoupled systems. In Proceedings of the American Control Conference,
Boston, Massachusetts, July 2004.
R. Kulhavý, J. Lu, and T. Samad. Emerging technologies for enterprise optimization in
the process industries. In J. B. Rawlings, B. A. Ogunnaike, and J. W. Eaton, editors,
Chemical Process Control–VI: Sixth International Conference on Chemical Process
Control, pages 352–363, Tucson, Arizona, January 2001. AIChE Symposium Series,
Volume 98, Number 326.
S. Li and T. Başar. Distributed algorithms for the computation of noncooperative
equilibria. Automatica, 23(4):523–533, 1987.
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 37 / 39
Further Reading III
J. Lu. Challenging control problems and emerging technologies in enterprise
optimization. Control Eng. Prac., 11(8):847–858, August 2003.
J. Lunze. Feedback Control of Large Scale Systems. Prentice-Hall, London, U.K, 1992.
W. L. Luyben. Dynamics and control of recycle systems. 1. simple open-loop and
closed-loop systems. Ind. Eng. Chem. Res., 32:466–475, 1993.
D. Q. Mayne, J. B. Rawlings, C. V. Rao, and P. O. M. Scokaert. Constrained model
predictive control: Stability and optimality. Automatica, 36(6):789–814, 2000.
T. Meadowcroft, G. Stephanopoulos, and C. Brosilow. The Modular Multivariable
Controller: 1: Steady-state properties. AIChE J., 38(8):1254–1278, 1992.
R. Monroy-Loperena, R. Solar, and J. Alvarez-Ramirez. Balanced control scheme for
reactor/separator processes with material recycle. Ind. Eng. Chem. Res., 43:
1853–1862, 2004.
M. Morari and J. H. Lee. Model predictive control: past, present and future. In
Proceedings of joint 6th international symposium on process systems engineering
(PSE ’97) and 30th European symposium on computer aided process systems
engineering (ESCAPE 7), Trondheim, Norway, 1997.
B. A. Ogunnaike and W. H. Ray. Process Dynamics, Modeling, and Control. Oxford
University Press, New York, 1994.
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 38 / 39
Further Reading IV
S. J. Qin and T. A. Badgwell. A survey of industrial model predictive control
technology. Control Eng. Prac., 11(7):733–764, 2003.
J. B. Rawlings and K. R. Muske. Stability of constrained receding horizon control. IEEE
Trans. Auto. Cont., 38(10):1512–1516, October 1993.
L. P. Russo and B. W. Bequette. Operability of chemical reactors: multiplicity behavior
of a jacketed styrene polymerization reactor. Chem. Eng. Sci., 53(1):27–45, 1998.
Y. Samyudia and K. Kadiman. Control design for recycled, multi unit processes. J.
Proc. Control, 13:1–14, 2002.
N. R. Sandell-Jr., P. Varaiya, M. Athans, and M. Safonov. Survey of decentralized
control methods for larger scale systems. IEEE Trans. Auto. Cont., 23(2):108–128,
1978.
D. Siljak. Decentralized Control of Complex Systems. Academic Press, London, 1991.
M. Sznaier and M. J. Damborg. Heuristically enhanced feedback control of constrained
discrete-time linear systems. Automatica, 26(3):521–532, 1990.
R. E. Young, R. D. Bartusiak, and R. W. Fontaine. Evolution of an industrial nonlinear
model predictive controller. In J. B. Rawlings, B. A. Ogunnaike, and J. W. Eaton,
editors, Chemical Process Control–VI: Sixth International Conference on Chemical
Process Control, pages 342–351, Tucson, Arizona, January 2001. AIChE Symposium
Series, Volume 98, Number 326.
Rawlings, Venkat and Wright (Wisconsin) Distributed, Large-scale MPC 2005 NMPC Workshop 39 / 39