Identification of continuous systems via markov parameters and time moments

摘要

This paper proposes an identification algorithm for multi-input and multi-output (MIMO) continuous systems via the Markov parameters and the time moments. The original system to be identified is transformed into a continuous-time system such that the transformed system is stable, the generalized time moments M{sub}i(α) of the transformed system exist even for unstable original systems, and the Markov parameters G{sub}i of the original system are equal to M{sub}i(α) /i!(i=0, 1, …). The minimum realization of the MIMO continuous-time system is identified by the ε-minimal realization method using the Markov parameters, which can be achieved by approximately calculating the generalized time moments of the transformed system using the input-output data sampled from the impulse response of the original system. Several illustrative examples will show that the algorithm provides good identification results even if the original system is unstable and there exists observing noise.