基于小波变换和神经网络的卫星钟差预报分析

Satellite clock error prediction based on wavelet transform and neural network

目前IGS提供的实时钟差精度不够,事后精密钟差也有13d的延迟,有必要对钟差预报进行研究。文中利用小波神经网络模型进行钟差预报,首先利用小波对原始钟差序列进行分解、降噪,然后利用神经网络进行建模并预报,将得到的结果同灰色模型和二次多项式模型的结果进行对比分析,得出小波神经网络模型可以更好地进行钟差预报的结论。

At present,the application of GPS real-time precise single point positioning is more and more popular.However,the accuracy of real-time clock error is not enough,which has 13 days after the precision of the delay,so it is necessary to study the prediction of the clock error.Using wavelet neural network model for prediction of clock error,the paper first uses wavelet to decompose and reduce the original clock sequence.The results are compared with the gray model and the two polynomial model.