Model Order Selection in System Identification: New and Old Techniques

The thesis analyzes some techniques adopted for model order selection in system identification: both classical methods (cross-validation, information criteria, the F-test and the statistical tests on the residuals) and innovative ones are evaluated, such as PUMS criterion and kernel-based estimation. The theoretical description of these methods is accompanied by an experimental analysis. Two combinations of the methods are also introduced, proving that they allow a more robust order selection