Sabit Parcali ve Anahtarlamali Dogrusal ve Periyodik Zamanli Sistemlerin Tanilamasi

This paper focuses on state-space identication for piecewise constant switching linear time-periodic (LTP) systems in the frequency domain. The proposed technique assumes full state measurement and known (or measurable) scheduling signals. In the case of linear time-invariant (LTI) systems, the statespace identication problem has been well studied and there are various techniques that work accurately in time and frequency domain. However, there are still many open issues in the statespace identication of LTP systems. In this paper, we specically focus on the family of LTP systems, which consist of a nite set of constant subsystems triggered by periodic scheduling signals. Although, this is a subset of the general family of LTP systems, here we explicitly model the availability of known periodic scheduling signal in the identication methodology. This greatly reduces the complexity of the estimated state-space models and potentially increases the system identication accuracy. We present a numerical example to demonstrate the efciency of our algorithm.