The H-Matrix Representation and System Realization.
AEROSPACE RESEARCH LABS WRIGHT-PATTERSON AFB OHIO
Pagination or Media Count:
The H-matrix representation for constant parameter linear systems is presented as an inputoutput description which is well suited for system identification and realization theories. The concept of a canonical representation and the measurements required to determine this representation are discussed. The minimal complete realization problem is solved for the H-matrix representation based on exponential signals. The connections between this approach and the identification and realization theories based on the Hankel matrix of Markov parameters are exhibited. A more general basis for considering both the complete and partial realization problems of constant parameter, linear systems is proposed. Several interpretations of this generalization of the Markov parameter approach to the realization problems are considered these interpretations being with respect to both the type of information used and the nature of the partial realizations obtained. Aside from the Markov parameter interpretation there is the possibility of basing realizations on moment coefficients or the H-matrix representation among others. The relationship of control invariants to these realizations is exposed as a first step in the general characterization of the equivalence class of realizations. Finally, a stability property is introduced for the realizations in connection with the data sets of the inputoutput map. Author
- Statistics and Probability