Abstract: In the practice of satellite navigation data processing, it is found that there is uncertainty and regularity in the broadcast ephemeris orbit error. For the reason that this kind of error information cannot be represented by a definite mathematical model, an error prediction model based on the particle swarm optimization back propagation (BP) neural network is established. In this model, the particle swarm optimization (PSO) is used to globally optimize the initial weights and thresholds of the BP neural network. The satellite position and velocity, calculated by broadcast ephemeris, with time information and perturbation correction parameters, are combined together to train and test the neural network. The results show that model’s fitting ability and prediction effect to the broadcast ephemeris orbit error are better. This model can be used to provide error compensation for satellite position calculation, so the accuracy of satellite orbit determination can be improved effectively and the system-level error can be reduced.

Key words: broadcast ephemeris orbit error, back propagation neural network, particle swarm optimization （PSO）, perturbation correction parameters