Bulletin des sciences géographiques de l'INCT
Volume 19, Numéro 1, Pages 18-25
2015-10-31
Analysis Of Gps Coordinates Time Series By Kalman Filter
Authors : Gourine Bachir . Niati Abdelhalim . Benyahia Achour . Mokhfi Brahimi .
Abstract
In this paper we try to process time series of position coordinates using Kalman filter and Kalman smoo-ther and to predict the position coordinates in intervals that contain erroneous data. Coordinates are originally obtained by relative GPS in kinematic way. The Kalman filter which can be implemented in real time, utilizes for computing the current estimate the precedent measurements and current measurement together with their variances. In turn, the Kalman smoother, also known as the RTS (Rauch-Tung-Striebel) smoother gives better estimates than Kalman filter since it exploits all data, i.e., after the total measu-rements have been done, then, it can be used only in post-processing. Due to the unknown or badly known internal geophy-sical effects, a purely dynamic model of deformation that involves the actual causative forces is then very difficult to establish with sufficient accuracy. Consequently, in this study, two descriptive models of deformation are chosen to play the role of two different dynamic systems: first, the identity model whose state vector contains position, second, the kinematic model whose state vector contains position and velocity. After results analysis, we conclude that the identity model is more suitable one than the other, because it describes betterthe behavior of motion of the receiver antenna.
Keywords
Times series, Kalman filter, identity model, kinematic model, deformation analysis.
Les articles similaires
Taibi Djamel
.
Titaouine Abdenacer
.
Benchabane Fateh
.
Bennis Ouafae
.
pages 51-60.
Boussofa Ahmad
.
Kidouche Madjid
.
Ahriche Aimad
.
pages 266-273.
Behmanesh J.
.
Bateni M. M.
.
pages 1-132.
