应用端部效应改善的LS+NN模型进行日长变化预报

现有ΔLOD(Delta Length-Of-Day,日长变化)预报模式在进行周期项与残差项拟合分离时,通常没有考虑LS(Least Squares,最小二乘)拟合序列的端部效应,预报精度难以取得较大提高。针对端部效应现象,首先采用时间序列分析模型在ΔLOD序列两端进行数据延拓,构成一个新序列,然后用新序列求得LS外推模型系数,再结合LS外推模型和NN(Neural Network,神经网络)对原始ΔLOD序列进行预测。算例表明,在ΔLOD序列两端增加延拓数据,能有效改善LS拟合序列的端部效应;端部效应改善的LS+NN模型的预报精度明显优于常规LS+NN模型,精度最大提高了17.86%。该方法不仅适用于LS+NN模型,也适用于LS外推模型与其他模型的组合。

Validating Effects of Ionospheric Delay Correction Models

With the increasing precision demand of spacecraft trajectory measurement data in space missions, ionospheric refraction error has become one of major sources affecting high precision measurement of Telemetry Tracking and Command(TT&C). Multifarious ionospheric models have been established based on a variety of observation data in the world. Because of different data and algorithms used in the models, there are different precision of correction results. This paper analyzes the characteristics of Klobuchar, International Reference Ionosphere(IRI) model and Chinese Reference Ionosphere(CRI) model which are used throughout the world. The three models were used for correction analysis for more than 50 passes of the measured tracking data of several on-orbit S-band satellites, and the result were compared with standard value from the satellite to the station in combination with coordinate transformation and geometric relationship from high precise ephemeris. The correction of tropspheric refraction error is done with microwave radiometer, and the results indicate that CRI model is better than the other two models for the Chinese region