大气阻力参数修正对低轨空间目标轨道预报精度的改进

对于低轨空间目标, 大气阻力是影响轨道预报精度的主要摄动力. 本文提出了一种 基于空间环境数据和神经网络模型的空间目标大气阻力参数修正方法, 基于目 标的历史两行元根数, 通过模拟得到外推一天轨道预报中预报结果与观测数据 符合最好的阻力调制系数, 分析表明其与太阳F_10.7指数和地磁Ap指数具有很好的相关性. 根据已有数据, 构建神经网络模型, 实现对阻力调制系数 的补偿计算, 从而改进低轨目标外推一天的轨道预报. 结果表明, 神经网络模 型相比两行元根数能够更及时地对空间环境变化进行响应. 将该方案应用于天 宫一号和国际空间站的外推一天轨道预报, 验证了方案的正确性和普适性, 对 地磁扰动引起的较大预报误差改进效果更好, 误差能够降低50%~60%; 平均而言, 预报精度可以提高约30%, 改进成功率达到80%左右. 

Atmospheric Drag Coefficient Modification for Orbit Prediction Precision Improvement of LEO Space Objects

Atmospheric drag is the primary disturbing force to LEO space objects since the atmospheric density is still considerable. This paper proposes a new method based on space environment indices and neural network model to modify drag coefficient. According to TLE (Two Line Element) sets, by simulating the orbit prediction and comparing prediction semi-major axis to real-time value, the optimal values of drag coefficients (B₀sup*) are selected. It is found that optimal values are one or two days ahead the values in TLE, and they are all corresponded with F_10.7 and Ap indices. Based on historic data, the neural network is built for drag coefficient correction to improve the orbit prediction precision. Result shows that the neural network model could timely response to space environment disturbance. This method is applied in Tiangong-1 (TG-1) and International Space Station (ISS) orbit prediction to verify its validity and universality, and it shows that the orbit prediction accuracy is improved by 50%~60% during geomagnetic disturbance while the errors are biggest. Generally, this method could improve the orbit prediction precision by 30%, the and success rate of improvement is about 80%.