| dc.description.abstract | This master s thesis studies the recently proposed initialization scheme for the H2-optimal static output feedback problem [1]. The initialization scheme consists of solving a convex quadratic programming (QP) problem to obtain the controller (K0) that in the open-loop sense is closest to the linear quadratic regulator (LQR). For the special case when the available measurements y(k) = Cx(k) , do not contain all information about the system states, we have that the resulting control law is u(k) = KC y(k) = K0 y(k) , where K is the LQR. In this thesis we show that for a class of systems this controller is suitable for initializing the H2-optimal static output feedback problem. In particular, we have tested the syntesized H2-optimal controllers on the thermal/optical plant uDAQ28/LT. A short study of the convexity properties of the H2-optimal static output feedback problem has shown that the problem is non-convex for some systems. Thus, we cannot expect to find the globally H2-optimal static output feedback for cases where the initialization procedure fails to provide a good controller guess. [1] Manum, H., Skogestad, S., & Jäschke, J. (2009). Convex initialization of theH2-optimal static output feedback problem. In Proceedings of the American Control Conference, (pp. 1724 1729). | nb_NO |
