Control Oriented System Identification

Abstract : The research goals for this grant were to obtain algorithms for control oriented system identification is to construct dynamical models of systems based primarily on measured data that are compatible with robust control design techniques. The research carried out under this grant has continued the research on control oriented identification originated by the PI and his collaborators, has extended control oriented identification methods to new classes of dynamical systems and has initiated a study unifying identification and control laws design. The research that has extended existing problem formulation concerns the construction of algorithms for linear shift invariant systems using a combination of apriori and experimental information. Algorithms for the identification of continuous time systems and efficient linear algorithms have been constructed. The research that has extended the existing problem formulation concerns the development of algorithms for the construction of parameterized linear families form a combination of apriori and measured information. Algorithms for this type of nonlinear system identification have been given that produce models suitable for gain scheduled controllers. Finally research into the integration control oriented identification and robust control for slowly time varying systems was initiated under this grant.