A hands-on approach to teaching system identification using first-order plus dead time modelling of step response data

Abstract

This paper describes three step response-based system identification methods of increasing complexity, together with a range of exercises that will enhance student understanding of this area in an engaging and practical way. For illustration purposes and practicality, it is assumed that the model to be identified is of the first-order plus dead time type. The first method uses a popular graphical technique, which is easy to understand and apply, but inaccurate when the response data are not ideal. The second uses the Nelder–Mead simplex method, which is a more powerful technique and has the added benefit of introducing undergraduate students to the concepts of numerical optimisation. The third uses an integral equation algorithm. The latter two methods, which can be readily extended to other model structures and input types, are also demonstrated using experimental data obtained from a tank level control system.