Brief paperMoving horizon estimation: Error dynamics and bounding error sets for robust control☆
Introduction
Moving horizon estimation (MHE) obtains the system state and disturbance estimates for a system by solving a constrained optimization problem (Findeisen, 1997, Rao et al., 2001). In the case that no system constraints are considered in the problem, the feedback controller and the state estimator can be designed separately following the separation principle (Rao, 2000). However, this is not the case if constraints are introduced in the control problem, since the estimation error can significantly degrade the controller performance and result in constraint violations (Mayne et al., 2006, Rawlings and Mayne, 2009). In order to address this problem, the estimation and control of the system have to be considered simultaneously; this is also known as the output-feedback control problem (Mayne et al., 2006).
In order to design a robust controller, which takes the estimation error into account, the bounds on the estimation error have to be known, and equivalently the estimation error dynamics, as well as the dynamics of the estimated system. Methods for obtaining these have been developed by Mayne et al. (2006) for the Luenberger observer, by Alessandri, Baglietto, and Battistelli (2003) and Sui, Feng, and Hovd (2008) for the unconstrained MHE with only state estimates and by Voelker, Kouramas, and Pistikopoulos (2010b) for the unconstrained MHE that includes both the estimation of the state and the disturbances.
In this work we present a method for the general case of constrained moving horizon estimation to obtain: (i) the estimation error dynamic equations, (ii) the estimation error bounds, and (iii) the state estimation dynamic equations. This paper presents an overview of the methodology and the interested reader can find more details and the proofs in Voelker (2011).
Section snippets
Problem definition
We consider linear discrete-time dynamic systems with disturbances and measurement noise where , , are the states, outputs and inputs of the system, while and are the process disturbance and measurement noise respectively. The system is subject to the following constraints:
Assumption 1 (i) The pair is controllable, (ii) is observable, and (iii) the sets , ,
Explicit/multi-parametric MHE
The explicit solution of MHE by using multi-parametric programming methods was first presented in Darby and Nikolaou (2007) for the case of MHE with filtered arrival cost. In this section we revisit their results for the smoothed arrival cost. By considering (i) the estimated values as the optimization variables, (ii) the measurements, the inputs, and the arrival cost as the problem parameters , and (iii) by substituting the state-space formulation of the
Dynamic equations and the bounding set of the estimation error and the state estimate of the constrained MHE
We proceed now to present a method for the determination of (i) the dynamic equations of the estimation error, (ii) the bounding set of the estimating error and (iii) the dynamics of the state estimate for the constrained MHE.
An illustrative example
The calculation of the error dynamics and the bounding error set is demonstrated with the following illustrative example: The error bounding set is obtained from the estimation error dynamics, as it was shown in Section 4.3. In our simulation the system initial condition is while the estimator initial condition is
Anna Voelker received her Ph.D. in 2011 in Control Engineering from Imperial College London. She obtained an M.Sc. in Automation and Robotics from the University of Dortmund in 2007 and in 2004 a B.Eng. in Information Technology from the Georg-Simon-Ohm University of Applied Sciences Nuremberg.
References (19)
- et al.
The explicit linear quadratic regulator for constrained systems
Automatica
(2002) - et al.
A parametric programming approach to moving-horizon state estimation
Automatica
(2007) - et al.
Robust output feedback model predictive control of constrained linear systems
Automatica
(2006) - et al.
Computation of arrival cost for moving horizon estimation via unscented Kalman filtering
Journal of Process Control
(2009) - et al.
Constrained linear state estimation—a moving horizon approach
Automatica
(2001) - et al.
Receding-horizon estimation for discrete-time linear systems
IEEE Transactions on Automatic Control
(2003) - Findeisen, P.K. (1997). Moving horizon state estimation of discrete time systems. Master’s Thesis. University of...
- Kouramas, K., Rakovic, S.V., Kerrigan, E.C., Allwright, J.C., & Mayne, D.Q. (2005). On the minimal robust positively...
- et al.
Multi-parametric programming, Vol. 1
(2007)
Cited by (0)
Anna Voelker received her Ph.D. in 2011 in Control Engineering from Imperial College London. She obtained an M.Sc. in Automation and Robotics from the University of Dortmund in 2007 and in 2004 a B.Eng. in Information Technology from the Georg-Simon-Ohm University of Applied Sciences Nuremberg.
Konstantinos Kouramas is a Research Associate in the Centre for Process Systems Engineering at the Department of Chemical Engineering, Imperial College London. He obtained a diploma in Electrical Engineering and Computer Technology from Patras University in 1997, an M.Sc. in Control Systems and a Ph.D. in Control Systems from the Department of Electrical and Electronic Engineering, Imperial College London. His research interests include theory and computational algorithms for control and optimization, and their applications in process, energy and automotive/aeronautics systems.
Efstratios N. Pistikopoulos, Director of the Centre for Process Systems Engineering (CPSE) from 2002 to September 2009, is a Professor of Chemical Engineering, at Imperial College London. He obtained a diploma from the Aristotle University of Thessaloniki in 1984, a Ph.D. from Carnegie Mellon University, USA in 1998, and was with Shell Chemicals (Amsterdam, the Netherlands) before joining Imperial in 1991. His research interests focus on the development of theory, algorithms and computational tools for multi-parametric programming and explicit model predictive control, and their applications in biomedical, energy and process systems.
- ☆
The financial contribution of EPSRC (EP/E047017/1 and EP/G059071/1) and European Research Council (MOBILE, ERC Advanced Grant No:226462) is gratefully acknowledged. The material in this paper was not presented at any conference. This paper was recommended for publication in revised form by Associate Editor Lalo Magni under the direction of Editor Frank Allgöwer.
- 1
Tel.: +44 2075946620, +44 207594662; fax: +44 2075941129.