Reduced-Order Parameter Estimation for Continuous Systems From Sampled Data

Abstract

An indirect method for estimating the parameters of the reduced continuous-time model from the sampled input/output data is presented. In this method, a discrete-time ARMA model is first identified. Then, the order of the continuous-time model is minimized by the dispersion analysis and/or accumulated dispersion analysis with the criterion of minimum discrepancy in sense of energy contribution between the original system and the reduced model. Finally, the reduced continuous-time model is matched to the identified discrete ARMA model in frequency domain. The proposed approach is applied to the identification of a power system stabilizer. The results show that the estimated continuous-time models are rather close to those supplied by the vender.