Ph.D. received on: 7/7/1997

E-mail: canale@comau.com

Tutor: Prof. Giovanni Fiorio, Dipartimento di Automatica e Informatica, Politecnico di Torino

 ___________________________________________________________________________________________________________

Interaction between Set Membership Identification and H-infinity Robust Control   ___________________________________________________________________________________________________________

Advisor:

Prof.  Giovanni Fiorio, Dipartimento di Automatica e Informatica, Politecnico di Torino

Summary:

In the thesis the problem of the identification oriented to control systems design has been approached in a Set Membership context. In this context the interation between the identification and the robust control has been investigated. The research has been focused on the Set Membership identfication of uncertainty models in a form well suited for the synthesis of   robust controllers. In the thesis it is presented the procedure for the identification of mixed parametric - non parametric uncertainty models made up by a nominal model of reduced complexity and an additive upper bound on the unmodeled dynamics. It is also shown how it is possible, by using these uncertainty models, guarantee the same performance level that can be obtained by a control system designed on the basis of the same uncertainty models with the nominal part of higher complexity. This way we can observe a “saturation” effect on the achievable performances by increasing the complexity of the nominal part of the model due to the fact that the identified upper bound on the unmodeled dynamics does not improve by increasing the order of the nominal part of the model. The practical results that follows is that the resulting designed compensator is also of reduced complexity as its order depends on the order of the nominal part of the model employed in the synthesis. The identification of uncertainty models has been performed with data collected both in time and frequency domain. The use of these uncertainty models can also be extended to the design of fixed structure robust compensators (e.g. PID) by means of interval mathematics techniques.
 
 
 _______________________________________