Finitely parameterised implementation of receding horizon control for constrained linear systems

Recent work by the authors has provided a finitely parameterised characterisation of receding horizon control (RHC) for linear models with quadratic performance index and linear constraints. Using a closed-form solution to quadratic programming, we have expressed the RHC law via a "look-up table" consisting of a partition of the state space into regions in which the corresponding control law has an explicit form. This paper investigates numerical properties of the table look-up implementation of the RHC solution and compares its performance with traditional methods for on-line optimisation (such as active constraint methods for quadratic programming). Issues that we consider include computation times, code complexity and data storage.

[1]  T. Johansen,et al.  On explicit suboptimal LQR with state and input constraints , 2000, Proceedings of the 39th IEEE Conference on Decision and Control (Cat. No.00CH37187).

[2]  Masoud Soroush,et al.  Analytical Model Predictive Control , 2000 .

[3]  G. Goodwin,et al.  Global analytical model predictive control with input constraints , 2000, Proceedings of the 39th IEEE Conference on Decision and Control (Cat. No.00CH37187).

[4]  Alberto Bemporad,et al.  The explicit linear quadratic regulator for constrained systems , 2003, Autom..

[5]  David Q. Mayne,et al.  Constrained model predictive control: Stability and optimality , 2000, Autom..

[6]  Alberto Bemporad,et al.  The explicit solution of model predictive control via multiparametric quadratic programming , 2000, Proceedings of the 2000 American Control Conference. ACC (IEEE Cat. No.00CH36334).