Abstract: In view of the lack of research on BeiDou satellite clock error prediction, this paper constructs a new combined prediction model based on grey model and neural network model. The combined clock error prediction model effectively combines the advantages of two single models through the optimal weight method to realize the short-term prediction of Beidou clock error. Finally, Taking the atomic clock data carried by the three types of Beidou satellites as an example, the weight of each single model is calculated, and the short-term prediction of clock error is realized by constructing the optimal weight prediction model. The prediction results are better than the two single models, which proves the effectiveness and applicability of the combined prediction model in the short-term prediction of clock error.