Interpolation between unconstrained optimal input trajectories and feasible
trajectories forms the basis for a computationally efficient predictive co
ntrol algorithm but lacks robustness in that uncertainty can destroy the gu
arantee of feasibility. To overcome this problem it is possible to introduc
e into the interpolation process a further input trajectory which is referr
ed as 'mean level'.(2) This has been accomplished in an input-output settin
g and the purpose of the present paper is to show that it is possible to ge
t a considerably simpler algorithm by recasting the problem into state-spac
e form. Copyright (C) 2000 John Wiley & Sons, Ltd.