SATELLITE ORBIT ESTIMATION USING ON-LINE NEURAL NETWORKS
This paper presents satellite orbit estimation using artificial neural networks. A multilayer Perceptron is used to estimate the position of a low-earth orbit satellite. The main goal is to filter out noisy or incomplete data received from sensors. The algorithm is applied to the CHAMP satellite. The same orbit is estimated using the extended Kalman filter. Simulation results show superior performance of the neural network as compared to the extended Kalman filter.