摘要
提出了一种基于小波变换和最小二乘支持向量机的卫星钟差预报方法。首先通过小波变换把钟差时间序列分解成具有不同频率特征的分量,然后根据各分量的特点构建不同的最小二乘支持向量机模型进行预报,最后将各分量的预报结果进行叠加得到最终的钟差预报值。实验结果表明,该方法的预报效果优于单一的最小二乘支持向量机模型以及常规的二次多项式模型和灰色系统模型。
This paper proposes a novel method for prediction of satellite clock bias(SCB)based on wavelet transform and least squares support vector machines(LSSVM).The method first decomposes SCB time series into different frequency components through the wavelet transform algorithm.Then,according to the different characteristics of the decomposed components,the proper LSSVM are constructed to predict the components.Finally,the final prediction value of SCB is yielded by the linearity superposition for the respective predicted result.The experiment shows that the proposed method can improve the prediction accuracy of SCB when the quadratic polynomial(QP)model,GM(1,1)model and LSSVM model.
出处
《武汉大学学报(信息科学版)》
EI
CSCD
北大核心
2014年第7期815-819,共5页
Geomatics and Information Science of Wuhan University
基金
国家自然科学基金资助项目(11103025)~~
关键词
卫星钟差
预报
小波变换
最小二乘支持向量机
satellite clock bias
prediction
wavelet transform
least squares support vector machines