Structural System Identification: advanced approaches and applications. Tesi di Dottorato di Ricerca in Ingegneria Civile, Universita’ di Pavia.http://dipmec.unipv.it/dott/dottorandi_storico.php

The content of the present thesis is organized as follows: The second chapter deals with the identification of a multi-degree of freedom linear system by means of the eigensystem realization theory. Throughout the chapter will be shown how the realization of the system, combined with the observer Kalman filter theory, leads to the formulation of a systematic and stable procedure for the identification of the dynamics of the system. These informations will be used to treat the damage detection problem in a structural system and, together with an implemented control law, to control the dynamic system behavior. Chapter three deals with the problems of monitoring and damage assessment in a structure using the neural network approach. Multi-degree of freedom, linear and non-linear systems will be analyzed. Particular attention will be given to the feedforward, multi-layers network type. As learning set for the network the transfer functions of the structure, evaluated in different damage conditions, will be used. The chapter provides also several numerical applications. Chapter four shows a probabilistic approach to the system identification problem. To this end a particular class of potential systems will be introduced and it will be shown how the parameters of the system can be identified solving a set of algebraic equations having as coefficients the statistical moments of the system response and as unknowns the parameters itself. Single and multi-degree of freedom stochastic models will be studied. Finally chapter five contains the general conclusions and some remarks for further research.