Algebraic properties and design of sampling rates in hybrid linear systems under multirate sampling (Q1610243)

The paper is concerned with the properties of reachability, observability and stabilizability in linear hybrid systems under multirate sampling of the state, input and output of the open-loop hybrid system with constant or time-varying parameters. The fulfilment of such properties is established in terms of the full rank property of matrices or nonsingularity of the associated Grammians. Several design issues have been addressed, like the selection of the sampling periods according to the parametrical optimization of a loss function consisting of a linear combination of worst-case \(L_{2}\) and \(l_{2}\)-induced norms related to the overall set of admissible inputs over the set of constraints of sampling periods. An alternative design of the sampling periods has been proposed in terms of minimizing the condition numbers of the observability or reachability matrices or their associated Grammians in such a way that the transmission of relative errors from the data towards the results when solving the associated algebraic problem is minimized. Two numerical examples are given. The first is related to the design of sampling rates for all possible choices of the inputs in the worst-case design context by using loss functions involving linear combinations of the \(L_{2}\) and \(l_{2} \)-induced norms. The second example addresses the choice of sampling rates for an optimal transmission of possible rounding or measurement errors in the reachability problem.