Projection-based Identification in the Frequency Domain

Abstract

This paper presents a frequency domain identification method for linear continuous-time systems. The method exploits the idea of the projection-based identification which has been developed by the authors in the time domain. One distinguished feature is that the convergence condition of the estimated parameters is given explicitly. It is also shown that the proposed method provides us a framework to view the existing methods such as IWLS, IV and Gauss-Newton ones in a unified way. Furthermore, the optimal convergence rate of the method is discussed, which explains the advantage of IV method over IWLS and Gauss-Newton ones. Finally, the validity of the discussion is demonstrated through a numerical example.