A Closed-loop Identification Method based on AIC and Its Application in the Distillation Column Reactor of Ethanol- Water System

Abstract

Considering process with time delay and limitations of traditional closed-loop identification methods, a new closed-loop identification method based on Akaike’s information criterion (AIC) was proposed. Firstly, the time delay term is approximated by first-order Padé and the equivalent closed-loop transfer model structure of closed-loop system is calculated. Secondly, the actual closed-loop transfer function model was identified by open-loop identification methods (e.g. ARX). Here, a method to select model structure in determination of discrete identification based on AIC was put forward. The high-order discrete model structure which is determined by this method has high identification accuracy. Finally, the equivalent closed-loop transfer function which is similar with first-order Padé and is gained from after transformation and reduction of the high- order discrete model owns an identification model with same structure. Hence, solving methods of process parameters could be deduced by coefficient of same power and two typical time delay process. It achieves satisfying identification effect in the closed-loop system identification of temperature and liquid level in distillation column reactor of ethanol -water system.