The invention provides an LSTM-based space target orbit error prediction method and system, and the method comprises the steps: obtaining satellite precise ephemeris data, TLE data and SGP prediction orbit data, and constructing a data set; providing a long-short-term memory network model, and performing network structure debugging and network hyper-parameter setting on the long-short-term memory network model to obtain an initial prediction model; training the initial prediction model by using the data set to obtain a final error prediction model; and predicting the space target orbit error by using the error prediction model. According to the LSTM-based space target orbit error prediction method and system provided by the invention, the dependence on physical parameters is reduced, complex physical conditions do not need to be considered, and historical orbit data is completely taken as a learning object to predict a future orbit; meanwhile, the training amount is reduced, and the model does not need to be trained independently for each fragment.
本发明提供了一种基于LSTM的空间目标轨道误差预测方法及系统,获取卫星精密星历数据、TLE数据以及SGP预测轨道数据,构建数据集;提供一长短时记忆网络模型,对所述长短时记忆网络模型进行网络结构调试和网络超参数设置,得到初始预测模型;利用所述数据集对所述初始预测模型进行训练,得到最终的误差预测模型;利用所述误差预测模型对空间目标轨道误差进行预测。本发明上提供的基于LSTM的空间目标轨道误差预测方法及系统,减少了对物理参数的依赖,不需要考虑复杂的物理情况,完全以历史轨道数据为学习对象,对未来轨道进行预测;同时减少了训练量,不需要针对每个碎片单独训练模型。
LSTM-based space target orbit error prediction method and system
基于LSTM的空间目标轨道误差预测方法及系统
2024-07-09
Patent
Electronic Resource
Chinese
TCN-LSTM-based underground space passenger flow prediction method and system
European Patent Office | 2024
|LSTM LSTM-based future threat prediction method and apparatus
European Patent Office | 2021
|Space target space-based monitoring efficiency evaluation optimization orbit determination method
European Patent Office | 2024
|