Presentation is loading. Please wait.

Presentation is loading. Please wait.

Working with Uncertainty in Model Predictive Control Bob Bitmead University of California, San Diego Nonlinear MPC Workshop 4 April, 2005, Sheffield UK.

Similar presentations


Presentation on theme: "Working with Uncertainty in Model Predictive Control Bob Bitmead University of California, San Diego Nonlinear MPC Workshop 4 April, 2005, Sheffield UK."— Presentation transcript:

1 Working with Uncertainty in Model Predictive Control Bob Bitmead University of California, San Diego Nonlinear MPC Workshop 4 April, 2005, Sheffield UK

2 Sheffield April 4, of 31 Outline Model Predictive Control Constrained receding horizon optimal control Based on full-state information or certainty equivalence How do we include estimated states? Accommodate estimate error — tighten the constraints Coordinated vehicles example Vehicles solve local MPC problems Interaction managed via constraints Estimation error affects the constraints — back-off Communication bandwidth affects state error Control, Performance, Communication tied-in

3 Model Predictive Control Works Full Authority Digital Engine Controller (FADEC) Commercial jet engine

4 Sheffield April 4, of 31 MPC with Constraints — Jet Engine Full-Authority Digital Engine Controller (FADEC) Multi-input/multi-output control 5x6 Constrained in Inputs - max fuel flow, rates of change States - differential pressures, speeds Outputs - turbine temperature Control problem solved via Quadratic Programming (every 10 msec) State estimator - Extended Kalman Filter State estimate used as if exact — Certainty Equivalence SENSORS ACTUATORS IGVs VSVs MFMV A8AFMV N2T2 N25 PS14 P25 PS3T4B STATIONS

5 Sheffield April 4, of 31 MPC Applied to Jet Engine Step in power demand Constraints Fuel flow Exit nozzle area Constraint-driven controller

6 Sheffield April 4, of 31 MPC Applied to Jet Engine Step in power demand Constraints Fuel flow Exit nozzle area Stage 3 pressure Two inputs One state

7 Sheffield April 4, of 31 Message MPC works in handling constraints on the model With accurate state estimates — this is fine for the real plant too

8 Sheffield April 4, of 31 What if the estimates are not accurate? Tighten the constraints imposed on the model — to ensure their satisfaction on the plant Remember. The MPC problem works on the estimate only

9 Sheffield April 4, of 31 Modifying constraints Want and we have Keep in MPC problem x+ x+ ^ x- x- ^ g t g-  t t+T x ^

10 Sheffield April 4, of 31 Handling uncertainty Two kinds of uncertainty Modeling errors State model is an inaccurate description of the real system State estimation errors Remember the MPC constrained control calculation works with the model and not the real system Constraints must be asserted on the real system

11 Sheffield April 4, of 31 Working with model error Total Stability Theorem (Hahn, Yoshizawa) Uniform convergence rate of nominal system + bounds on model error  bounds on state error MPC formulation of Total Stability Robust Control Lyapunov Function idea degree of stability model error bound

12 Sheffield April 4, of 31 Comparison model Main lemma: For any control The controlled behavior of dominates that of Uses a control Lyapunov function for the unconstrained system

13 Sheffield April 4, of 31 Including constraints  three systems real system comparison system model system

14 Sheffield April 4, of 31 MPC with Comparison Model Subject to If feasible at t=0 then Feasible for all t Real system is stable, constrained and

15 Sheffield April 4, of 31 Example From Fukushima & Bitmead, Automatica, 2005, pp

16 Sheffield April 4, of 31 Working with state estimates Kalman filtering framework Gaussian state estimate errors Probabilistic constraints are needed State estimate error Rework this as The constrained controller will need to be cognizant of This is a non-(certainty-equivalence) controller Information quality is of importance Same concept of tightening constraints

17 Sheffield April 4, of 31 Approximately Normal State Manage constraints by controlling the conditional mean state Use the control independence of  x n  i ~N( ˆ x n  i|n,  n  i|n ) Pr(x n  i  X)   ˆ x n  i|n  ( ,  )

18 Sheffield April 4, of 31 Pause for breath Our formulation so far Model errors Tighten constraints on the nominal system State estimate errors Tighten constraints to accommodate the estimate covariance Preserves the MPC structure and properties Original constraints inherited by real system Perhaps with probabilistic measures Feasibility and stability properties Via terminal constraint as usual Some examples …

19 Sheffield April 4, of 31 The Shinkansen Example One dimensional problem Three Shinkansen [Bullet Trains] on one track Uncertainty in knowledge of other trains’ positions Uniformly distributed with known width Follow the same reference with each train Constraint — no crash with preceding train Leader-follower strategy Each solves an MPC problem with state estimation

20 Sheffield April 4, of 31 Collision avoidance with estimation

21 Sheffield April 4, of 31 Train coordination All trains have the same schedule Osaka to Tokyo in three hours Depart at 09:00, arrive at 12:00 Each solves their own MPC problem Minimize departure from schedule No-collision constraint Estimates of other trains’ positions Trains separate early Separation reflects quality of position knowledge

22 Sheffield April 4, of 31 Back to the Trains Low Performance plus String Instability

23 Sheffield April 4, of 31 Relaxed Target Schedules Low Performance but no string instability Constraints not active

24 Sheffield April 4, of 31 Improved Communication High performance, no string instability

25 Sheffield April 4, of 31 Big Issues Constraints Quality of Information Communication Network and Control Architecture Tools for systematic design of complex interacting dynamical systems Model Predictive Control and State Estimation

26 Sheffield April 4, of 31 Single Node in Network Queue length q t is the state variable Constraint q t ≤Q else retransmission required Control signals are the source command data rates v i,t Propagation delays d i exist between sources and node Available bit rate  t is a random process Model as an autoregressive process P(q t ≥Q)<0.05

27 Sheffield April 4, of 31 Fair Congestion Control 50 retransmissions per 1000 samples

28 Sheffield April 4, of 31 Simulated Source Rates — Fair! mean = variance = mean = variance = mean = variance =

29 Sheffield April 4, of 31 A Tougher Example From Yan & Bitmead, Automatica, 2005 pp

30 Sheffield April 4, of 31 Network Control A variant of the train control problem Much greater degree of connectivity — higher dimension Improved performance is achievable by sending more frequent or more accurate state information upstream to control data flows This consumes network resources and must be managed MPC and State Estimation (Kalman Filtering) tools prove of value

31 Sheffield April 4, of 31 Conclusions MPC plus State Estimation Tools for coordinated control performance with managed communication complexity Information architecture Resource/bandwidth assignment … as a function of system task

32 Sheffield April 4, of 31 Acknowledgements Hiroaki Fukushima, Jun Yan, Tamer Basar, Soura Dasgupta, Jon Kuhl, Keunmo Kang NSF, Cymer Inc GE Global Research Labs, Pratt & Whitney, United Technologies Research Center My gracious UK and Irish hosts, IEEE

33 Sheffield April 4, of 31 Constraints in design The appeal of MPC is that it can handle constraints Constraints provide a natural design paradigm Lane keeping potential function

34 Sheffield April 4, of 31 A Design Bonus The MPC/KF design is much less sensitive to selection of design parameters than LQG Constraints work well in design — simplicity From Yan & Bitmead, Automatica, 2005 pp

35 Sheffield April 4, of 31

36 Sheffield April 4, of 31


Download ppt "Working with Uncertainty in Model Predictive Control Bob Bitmead University of California, San Diego Nonlinear MPC Workshop 4 April, 2005, Sheffield UK."

Similar presentations


Ads by Google