Minimal continuous model identification via Markov parameter estimation

In this paper, an attractive and novel algorithm for improving irreducible model identification of continuous time(CT) MIMO systems has been presented. The algorithm is based on least - squares (LS) estimates of Markov parameters (MP) using input output data and residual whitening. By choosing a linear-in-parameters model structure, the estimation becomes linear and asymptotically robust to zero-mean additive disturbances. CT Markov parameters may result in diverging approximations even for stable systems. To remove the existing limitations in the case of systems with low or zero damping, Markov Poisson parameters have been used to lend much flexibility to the estimation model. The MIMO problem has been divided into a set of MISO subproblems which are identified independently. Finally, the proposed approach has been applied to a boiler.