Choice of models for the identification of linear multivariable discrete-time systems

A critical study is presented of the relative effectiveness of four types of models which have been used in the area of linear multivariable discrete-time systems identification. Each model's features and their effect on the complexity of the identification algorithm are studied. The structural parameters required to characterise each model and the number of parameters to be estimated are examined and compared. The characteristics of the parameter estimates of each model when using the least-squares method are also investigated. Results of a simulated example are given which show the advantages and disadvantages of each model when used for identification.