Closed-Loop Identification Methods for LQ Control Design

Abstract In connection with the current interest in the interaction between identification and control design, a closed-loop identification method is proposed that takes into account a subsequent LQ control design step. It appears that this method unifies existing closed-loop identification schemes. Hence it facilitates comparison of these schemes. The properties of the resulting model depend on the chosen parametrization. It is shown that the right choice of parametrisation results in a model, of which the bias distribution is independent of the noise, which is an advantage with respect to some classical approaches. The method is illustrated with a simulation example.