An Approach to Parameter Identification of Continuous-Time System Based on λ Transform
Yang Wu, Tohru KATAYAMA
This paper deals with the identification of parameters of continuous-time system based on λ-transform (λ=1/ (τs+1)) due to Johansson. Using the pre-filtering technique based on the λ transform followed by a sampling of the pre-filtered signals, we derive a linear regression model to which the_least-squares (LS) method is applied. Noting that the residuals associated with the LS estimate are dependent on the parameter of the pre-filter, we present a simple search method of choosing the optimal parameter so that the sum of squared residuals is minimized. To show the applicability, we present simulation results for the identification of vibrating systems excited by an impulse input where the output acceleration is measured.