Optimization techniques for orbit estimation and determination to control the satellite motion

Abstract

This paper analyzes the methods of accurate orbit estimation and determination, which are the first and the most important phases in preparation of the satellite-control-maneuver. The paper specifically addresses the algorithmic implementation of the Least Mean Squares (LMS) and Kalman filters for the spacecraft orbit determination problem. The inherent advantages, disadvantages and trade-off, which are required to select the suitable method, have been analyzed. This approach has been refined using multi layer Adaptive Neural Network (ANN) with a nonlinear function. The three different approaches have been compared with regard to efficiency, performance, accuracy and running time in order to analyze the inherent characteristics of the used techniques. ANN has shown a good agreement with the definitive solution in case of the availability of old cycles data of the geostationary satellite. Depending on the application and available data one should carefully choose among these methods in orbit determination.