Models of sampled data systems having an infinite-dimensional state space

The paper is concerned with design of models discrete-time dynamic systems in a state space from the specified input - output mapping (abstract Kalman design). A new method is described for realization using shift models whereby a modely is developed for nonstationary and nonlinear systems from a certain class of linear systems without constraints on the state space dimension. Examples are given of system realization from transfer matrices. Conditions for finite-dimension realization are formulated.