Theory and computation of disturbance invariant sets for discrete‐time linear systems

This paper considers the characterization and computation of invariant sets for discrete‐time, time‐invariant, linear systems with disturbance inputs whose values are confined to a specified compact set but are otherwise unknown. The emphasis is on determining maximal disturbance‐invariant sets X that belong to a specified subset Γ of the state space. Such d‐invariant sets have important applications in control problems where there are pointwise‐in‐time state constraints of the form χ ( t ) ∈ Γ . One purpose of the paper is to unite and extend in a rigorous way disparate results from the prior literature. In addition there are entirely new results. Specific contributions include: exploitation of the Pontryagin set difference to clarify conceptual matters and simplify mathematical developments, special properties of maximal invariant sets and conditions for their finite determination, algorithms for generating concrete representations of maximal invariant sets, practical computational questions, extension of the main results to general Lyapunov stable systems, applications of the computational techniques to the bounding of state and output response. Results on Lyapunov stable systems are applied to the implementation of a logic‐based, nonlinear multimode regulator. For plants with disturbance inputs and state‐control constraints it enlarges the constraint‐admissible domain of attraction. Numerical examples illustrate the various theoretical and computational results.