Systematic and effective embedding of feedforward of target information into MPC

Abstract

Discussions on how to make effective use of advance information on target changes are discussed relatively rarely in the predictive control literature. While earlier work has indicated that the default solutions from conventional predictive control algorithms are often poor, very little work has proposed systematic alternatives. This paper proposes an embedding structure for utilising advance information on target changes within an optimum predictive control law. The proposed embedding is shown to be systematic and beneficial. Moreover, it allows for easy extension to deal with more challenging scenarios such as unreachable set points and guarantees of convergence/stability in the uncertain case.

Dual-Mode Model Predictive Control of an Omnidirectional Wheeled Inverted Pendulum

“…Instead of continuing with two separate modes as in (13), analysis can be simplified by the formation of an all-encompassing autonomous prediction model of the form Z k+1 = Ψ Z k . This also allows constraints of the form Gx ≤ f to be applied to predicted step Z k+n as GΨ n Z k ≤ f .…”

“…Simon [15] replaces r with a pseudo-referencer as an additional degree of freedom, penalising deviation of this from the true reference. Dughman [13] introduces an extra perturbation term c ∞ to the end of the control sequence c →k that acts as a constant perturbation to the input for k > n c . This constant perturbation has the same effect as a pseudoreference, with the equivalent pseudo-reference calculable as r = C(I − Φ) −1 Bc ∞ + r.…”

“…This controller topology was chosen for its systematic design approach, its numerical simplicity in comparison to nonlinear MPC approaches, its a priori stability guarantee, and its improved numerical conditioning in the prediction of open loop unstable plants [11]. Additionally, recent approaches to ensuring feasibility for the full set of unconstrained reference inputs are incorporated whilst maintaining optimality [12], [13]. The effect of controller parameters on the feasible state set is analysed, and a systematic design approach is described for the selection of control horizon n c and quadratic cost weighting matrices Q and R. Finally, both simulated and experimental results demonstrate the suitability and efficacy of this controller for the control of a CMD, performed using…”

Towards an Improved Hierarchical Control Strategy for a Solar Thermal Power Plant

“…Given n a -step ahead predictions of solar radiation and the field inlet temperature and with slight modifications to the one-step ahead prediction model in (4), one can in fact obtain n a -step ahead predictions of the reference temperature. It has been shown in [8] that an effective use of advance information on set point changes within an optimum predictive control law can be advantageous and beneficial and yet this has been little studied in the context of solar plant.…”

Triple-Mode Model Predictive Control Using Future Target Information

“…Their method only pertains to the unconstrained case. In [14], the authors proposed a dual-mode constrained MPC algorithm, which was composed of a linear state feedback and extra degrees of freedom. Under such a control law, the feedforward could not be designed individually and the future target horizon was determined through a trial and error process.…”