Estimation of orbital elements using methods of non-linear optimization

Nonlinear optimization techniques are applied to determine the orbital elements of a satellite from noisy ground observations. The problem is considered as a simultaneous estimation problem. The nature of the optimization problem for different kinds of estimators is discussed. Two different types of optimization methodologies, the gradient-dependent and the gradient-independent, are employed. Studies are made to determine the orbit from simulated multipass as well as single-pass data. Comparisons of the results show that the convergence characteristics of some of the optimization methods are far superior to those of the classical differential correction method. Some optimization methods are recommended for routine orbit determination.