Akademik Veri Yönetim Sistemi
Tezin Türü: Yüksek Lisans
Tezin Yürütüldüğü Kurum: Orta Doğu Teknik Üniversitesi, Mühendislik Fakültesi, Elektrik ve Elektronik Mühendisliği Bölümü, Türkiye
Tezin Onay Tarihi: 2014
Öğrenci: GÖKHAN ÖZDOĞAN
Danışman: MEHMET KEMAL LEBLEBİCİOĞLU
Özet:The field of system identification takes a fundamental place in control engineering. In order to design an efficient robust controller, one requires an accurate mathematical model with its uncertainty for the dynamical system of interest. It is crucial that experiments are well designed so that parameters to be estimated through statistical methods result in least possible bias and minimum variance. In this study, four axes gyro stabilized Infrared Electro Optic gimbal system is modeled in the frequency domain through experimental investigation. The input and output signals are logged using xPC Target of MATLAB with a sampling frequency of 3kHz. The communication with real system is realized with RS-422 protocol. In system identification, the first step is experiment design. Various perturbation signals are analyzed and compared. A cost function is offered to optimize power spectrum for the input excitation signal while satisfying device specific constraints. Second step is frequency response function measurement. By averaging techniques, it is possible to reduce the variance of frequency response function measurement and decrease the error. Moreover, the averaging in the frequency domain provides the nonparametric noise model of the system. Periodic excitation signals are used and an integer number of signal periods have been measured. Before passing to the next step, a robust method to detect and quantify nonlinear distortion on frequency response functions measurements is studied. In the final step, real system is modeled by its parametric transfer function with plenty of different estimation techniques and their efficiencies, convergence properties, bias errors are compared and discussed.