Geometric techniques for implicit two-dimensional systems

Abstract

Geometric tools are developed for two-dimensional (2-D) models in an implicitFornasini–Marchesini form. In particular, the structural properties of controlled and conditionedinvariance are defined and studied. These properties are investigated in terms ofquarter-plane causal solutions of the implicit model given compatible boundary conditions.The definitions of controlled and conditioned invariance introduced, along with the correspondingoutput-nulling and input-containing subspaces, are shown to be richer than theone-dimensional counterparts. The analysis carried out in this paper establishes necessaryand sufficient conditions for the solvability of 2-D disturbance decoupling problems andunknown-input observation problems. The conditions obtained are expressed in terms ofoutput-nulling and input-containing subspaces, which can be computed recursively in a finitenumber of steps.