基于补偿波形调整的导航卫星轨道预报方法

发布时间：2018-03-31 09:06

本文选题：神经网络　切入点：轨道预报　出处：《武汉大学学报(信息科学版)》2017年08期

【摘要】：针对利用动力学模型得到的预报轨道随时间推移精度衰减较快的问题,尝试采用神经网络作为建模工具改进北斗导航卫星轨道预报精度。对影响神经网络模型补偿效果的因素进行了详细分析,基于神经网络补偿波形调整策略制定了适应导航卫星短期、中期和长期预报的神经网络优化模型。利用实测数据进行了试验分析,结果表明,该方法可以显著改进利用动力学模型得到的预报轨道精度。短期预报中,当采用的训练样本距离当前时刻大于10d时,应移动补偿波形;中长期预报中均应移动补偿波形。相比补偿波形不调整的神经网络模型,采用基于补偿波形调整的神经网络优化模型后,预报弧长为8、15、30d时,改进率分别提高了2.3%、6.7%、10%。
[Abstract]:In order to solve the problem that the prediction orbit obtained by using the dynamic model decays rapidly with time, this paper attempts to use neural network as a modeling tool to improve the prediction accuracy of Beidou navigation satellite orbit.The factors affecting the compensation effect of the neural network model are analyzed in detail. Based on the adjustment strategy of the neural network compensation waveform, the neural network optimization model for the short, medium and long term prediction of navigation satellite is developed.The experimental results show that this method can significantly improve the accuracy of the prediction trajectory obtained by using the dynamic model.In short term forecast, the compensation waveform should be moved when the distance between the training sample and the current time is more than 10 days, and the compensation waveform should be moved in the medium and long term forecast.Compared with the neural network model which does not adjust the compensation waveform, after adopting the neural network optimization model based on the compensation waveform adjustment, when the prediction arc length is 81515 / 30 days, the improvement rate is increased by 2.3 / 6.7and 10 / 10 respectively.
【作者单位】：北京卫星导航中心;中国科学院测量与地球动力学国家重点实验室;61287部队;
【基金】：国家自然科学基金(41204022,61603397) 大地测量与地球动力学国家重点实验室开放基金课题(SKLGED2017-3-3-E)~~
【分类号】：TN967.1;V412.41

【相似文献】