A Behavioral Approach to H2 Optimal Control

In this paper we consider a formulation of theH2 optimal control problem in which the controller is not viewed as an operator that transforms measurements to controls. Instead, the controller is assumed to be a device that constrains a set of a-priori specified interconnection variables so as to achieve a controlled system whose manifest variables have a free component and satisfy an upperbound on the L2 norm of its impulse response. We formalize and solve this problem in a behavioral setting, that is, without reference to system representations and input-output structures. The equivalance between the H2 control problem, the generalized H2 control problem and the LQ optimal control problem are established.