基于小波神经网络的电离层TEC短期预报

提出利用神经网络中的小波神经网络技术(WNN)对全球的电离层总电子含量进行预报,以国际全球定位系统服务(IGS)发布的电离层TEC数据资料为样本,并用发布的已知TEC进行检核,分析了全球预报精度及不同纬度的预报精度。试验结果表明,小波神经网络的短期预报结果能很好地符合已知数据,预报的精度较高,预报值和已知值之差小于3TECU的电离层格网点数占总格网点数的百分比都在90%以上,随着时间的推移,比值下降并不明显。

Predicting Shortdated Ionospheric TEC Based on Wavelet Neural Network

The wavelet neural network on predicting ionosphere is proposed for the whole world. According to the ionsopheric TEC data as swatchs released by IGS(international GPS service),The TEC prediction value is checked by original value,the whole prediction precision and the different latitude precision are analysed. The results of experimentation show that the predicted TEC calculated with wavelet neural network accord with the ionospheric TEC ,and the precision of the predicted TEC is high,the percentage of the frids number of the difference between prediction value and original value less than three TECU beyond 90% ,as time going,this percentage descends unconspicuously.