Tezin Türü: Yüksek Lisans
Tezin Yürütüldüğü Kurum: Orta Doğu Teknik Üniversitesi, Mühendislik Fakültesi, Elektrik ve Elektronik Mühendisliği Bölümü, Türkiye
Tezin Onay Tarihi: 2017
Öğrenci: MEHMET İPEK
Danışman: UMUT ORGUNER
Özet:In this thesis, satellite orbit determination problem is investigated using various types of Kalman filters based on several ground and space based measurement types. First, the performance of the Gauss' method for initial orbit determination is evaluated and the measurement time separation values required for it to give sufficiently good estimates are determined. Second, filter initialization methods based on a few measurements are proposed for angles only (AE), angles and range (AER), angles and Doppler (AED) and full state observation cases and their performances are investigated. Third, comparison of the relative performances of extended Kalman filter (EKF), unscented Kalman filter (UKF) and continuous-discrete EKF (CD-EKF) is carried out for the sample scenarios assuming that AER and full state measurements are available. It is shown that continuous-discrete EKF running with a reasonable prediction step size, gives the best results in terms of root mean square (RMS) estimation error and computation time. The performance of the CD-EKF is further examined by comparing its RMS errors with posterior Cramer Rao Lower Bound (PCRLB) for all measurement types and it is shown that CD-EKF is almost efficient. In addition to these, relative orbit determination performances obtained using AE, AER and AED measurements are compared. It is shown that the use of the Doppler measurement in addition to angles only measurements significantly improves the estimation performance in terms of position and velocity RMSE's. Furthermore, usefulness of a second observing station (located sufficiently apart from first) is studied. It is found that using a second station provides a significant improvement in the estimation performance especially when the range and Doppler data are not available.