The problem investigated is that of identification and orbit determination of a tethered satellite system, when sparse observation is available. Two approaches to the problem are discussed. First, a standard least-squares batch filter is employed. The model contained within the filter accounts for the gross orbital motion of a tethered satellite as well as in-plane libration, and includes gravity effects due to the Earth's oblateness. The second approach utilizes an artificial neural network to predict the state of the system at epoch time given a batch of observation data. The network is trained to emulate the dynamics of a tethered system using observation data sets for which the system's state values are known.
展开▼
