基于经验模式分解和最小二乘支持向量机的卫星钟差预报

发布时间：2018-03-16 03:03

本文选题：天体测量学　切入点：时间　出处：《天文学报》2014年03期 　论文类型：期刊论文

【摘要】：针对卫星钟差(satellite clock bias,SCB)呈现非线性、非平稳变化的特性,提出结合经验模式分解(empirical mode decomposition,EMD)和最小二乘支持向量机(least squares support vector machines,LSSVM)的钟差预报方法.首先对钟差相邻历元间作一次差,并利用经验模式分解将差分序列分解成若干不同频率的平稳分量,分解后的分量突出了差分序列不同的局部特征;然后根据各个分量的变化规律,选择合适的核函数和相关参数构造不同的最小二乘支持向量机模型分别预报;最后将各分量预报值叠加得到一次差预报值,再将其还原得到钟差预报值.实验结果表明,所提方法与常用的二次多项式(quadratic polynomial,QP)模型、灰色系统(grey model,GM)模型和单一的最小二乘支持向量机模型相比,具有较高的预报精度和较强的泛化能力.
[Abstract]:According to the satellite clock (satellite clock, bias, SCB) is nonlinear, non-stationary characteristics change, combining with empirical mode decomposition (empirical mode, decomposition, EMD) and least squares support vector machine (least squares support vector machines, LSSVM) prediction method of clock error. First the clock to make a difference between two adjacent epochs. And the decomposition of difference sequence is decomposed into several stationary components of different frequencies by using empirical mode decomposition, the component highlights the difference sequence of different local features; then according to the variation of each component, select the appropriate kernel function and related parameters to construct different least squares support vector machine model prediction; finally components the forecast value superposition a bad forecast value, then the reduction of the clock error prediction value. The experimental results show that the proposed method and the commonly used two polynomial (quadratic polynomial The QP model, the grey system (grey model, GM) model and the single least squares support vector machine model have higher prediction accuracy and stronger generalization ability.

【作者单位】：中国科学院国家授时中心;中国科学院时间频率基准重点实验室;中国科学院大学;中国科学院精密导航定位与定时技术重点实验室;
【基金】：国家自然科学基金项目(10573019)资助
【分类号】：P127

【参考文献】