Identification of an Ethane-ethylene distillation column system — A procedure for MIMO system identification using the IDENT toolbox

This paper deals with the identification process of an ethane-ethylene distillation column system and introduces a procedure for MIMO system identification using Matlab´s IDENT toolbox. An existing, five inputs and three outputs, ninety samples dataset has been analyzed. Four separate datasets are analysed, each having a different level of noise. As a part of the proposed procedure, both nonparametric and parametric identification methods have been implemented and results have been discussed. A comparative analysis of the results for all of the tested model structures has been carried out and the best performing model has been chosen. A Simulink model of the identified system has been implemented based on the best performing parametric model. It has been stimulated with the existing input datasets and the resulting output signals have been compared to the existing output datasets. The obtained results confirm the quality of the obtained model and affirm the correctness of the proposed and implemented procedure.