Orbit Prediction Error Analysis Using Global Sensitivity Analysis (IEEE/CSAA GNCC)

Abstract

Global sensitivity analysis is applied to Satellite orbit prediction error to investigate where its uncertainty comes from, and variance-based sensitivity indices are adopted to rank the uncertainty contribution of different initial orbital elements. For the calculation of sensitivity indices, a new technique based on advanced moving least squares approximation is proposed, and the computational cost of the new technique is less than most other techniques available today. A numerical example is employed and three different calculating techniques are involved to demonstrate the accuracy and efficiency of the propose method, especially for models with severe nonlinearity. Finally, the proposed calculation technique is conducted on the global sensitivity analysis of orbit prediction error and all first-order and total-effect indices of all initial orbit elements are obtained, then experts can improve the precision of orbit prediction.