System Identification with Multi-Laguerre filters

Abstract

An identification scheme of a plant with widely spread poles is proposed.The Laguerre approach enables us to set the sampling period of the experimental data for the identification independently of the plant poles but the point of the poles of the Laguerre filters must be set appropriately in relation to the plant poles.Therefore if the plant has widely spread poles,it is not easy to identify the plant with the Laguerre approach.The proposed scheme uses plural sets of the Laguerre filters with different poles to detect widely spread poles.Least squares are used for the parameter estimation and the performance index is the sum of the performance indexes for the ordinal least squares for each Laguerre parameter.