Kanonik Korelasyon Analizi Iie Sistem Tanıma
Toker, Kemal Gürkan
xmlui.mirage2.itemSummaryView.MetaDataShow full item record
In this thesis, canonical correlation analysis is introduced for linear system identification. By taking into account the similarities between difference equation representation of a linear discrete time system and canonical correlation analysis, it is shown that canonical correlation analysis can be used for system identification.Canonical correlation analysis (CCA) is a powerful method which is used to measure the relationship between two multidimensional variables. This approach tries to find the linear combinations of these two variables with maximum correlation. Data sets are obtained from input and output data of the systems and system parameters are estimated using these data.Linear, discrete time SISO and MIMO systems have been studied using this analysis method. Also, a numerical algorithm is introduced for real time system identification implementation.