Robust Multi-Parametric Model-Based Control

Tuesday, November 18, 2008: 4:15 PM
Room 202-A (Pennsylvania Convention Center)

Efstratios N. Pistikopoulos, Centre for Process Systems Engineering, Imperial College London, London, United Kingdom
Konstantinos Kouramas, Centre for Process Systems Engineering, Department of Chemical Engineering, Imperial College London, London, United Kingdom
Nuno P. Faísca, Imperial College London, London, United Kingdom

In this work we present the fundamental theoretical developments for a unified and general framework for the design of Robust Multi-Parametric Controllers (Pistikopoulos et al. 2007ab, ParOS 2007). Given a dynamic system with model uncertainties both in the right and left hand-sides (of constraints of the underlying uncertain optimisation problem), the proposed framework is realised in three steps:

1.robust reformulation of the constraints: the state/output and input constraints are reformulated to account for the worst-case model uncertainty (Ben-Tal and Nemirovski, 2001, Lim et al., 2004).

2.dynamic programming: the uncertain optimisation problem is reformulated with a dynamic programming multi-stage setting (Bellman, 2003, Faísca et al., 2008).

3.multi-parametric programming: each of the stages of the DP problem are solved as multi-parametric programming problems and the control input is obtained as an explicit function of the incumbent state (Pistikopoulos et al, 2007ab).

The above framework can guarantee the design of multi-parametric MPC (mp-MPC) controllers that can ensure constraints satisfaction for all admissible uncertainties as opposed to the previous non-robust mp-MPC design techniques (Sakizlis et al., 2004). The results are demonstrated with examples that illustrate the proposed framework.

References:

Bellman, R. (2003). Dynamic Programming. Dover Publications

Ben-Tal, A., Nemirovski, A., (2000). Math. Program 88, 411-424

Faísca, N.P., Kouramas, K.I., Saraiva, P.M., Rustem, B., Pistikopoulos, E.N., (2008). Optimization Letters 2, 267-280

Lim, X., Janak, S.L., Floudas, C.A., (2004). Comp. & Chem. Eng. 28, 1069-1085

Parametric Optimisation Solutions (ParOS) Ltd, Improved Process Control, European Patent EP1399784, 2007

Pistikopoulos, E.N., Georgiadis, M. & Dua, V., (2007). Multi-parametric Parametric Programming. Weinheim: Wiley-VCH

Pistikopoulos, E.N., Georgiadis, M. & Dua, V., (2007). Multi-parametric Parametric Model-based Control. Weinheim: Wiley-VCH

Sakizlis, V., Kakalis, N.M.P., Dua, V., Perkins, J.D., Pistikokpoulos, E.N., (2004). Automatica 40, 189-201

Extended Abstract: File Not Uploaded
See more of this Session: Advances in Model Predictive Control
See more of this Group/Topical: Computing and Systems Technology Division