MIMO D-T system identification using subspace based methods

Abstract

This paper presents a MIMO system identification algorithm based on calculating Markov parameters directly from available input-output data. The identification algorithm discussed in the paper can be considered as a version of the subspace model identification schemes. In selecting a state space model corresponding to input-output data, the algorithm makes use of recently introduced concept of pseudo-observability indices, which greatly increases the flexibility of the procedure. An illustrative computational example is included.