System Identification from PI Control Operation Data

Abstract

This paper presents the system identification from PI control operation data. A general pre-filtering procedure is not required in this method. A tow-step procedure for estimating Laplace transfer function of a process is proposed. The closed loop transfer function model (called the Laguerre model) is represented by Laplace transforms of Laguerre polynomials approximated to a true response, exploiting orthonormal properties of Laguerre functions. Then, the process transfer function model is computed from the Laguerre model and PI controller. Simulation and experiment results show that the proposed method is effective for non-linear systems in modelling.