首页> 中文期刊>空间科学学报 >利用人工神经网络提前1h预报电离层TEC

利用人工神经网络提前1h预报电离层TEC

     

摘要

A handy method of forecasting the ionospheric TEC one hour ahead by Artificial Neural Network (ANN) is presented in this paper. Considering of the practical application, the observations of TEC are used as inputs without any other data. The input parameters are the present observation of TEC, the first difference and relative difference of TEC, and the local time. The output is the TEC one hour ahead. Ionospheric TEC data evaluated from GPS measurements at Xiamen receiving station is used to checkout the forecasting method. The relative error is 9.3744%, and the cross correlation coefficient between the observed and forecast TEC values is 0.96678. The accuracy rate of relative error less than 15% is 79.59%, during the geomagnetic storms, but 98.81% for the quiet or moderate geomagnetic conditions. These conclusions suggest that the value of forecasting is very the geomagnetic level. It is shown that the Artificial Neural Network is promising in forecasting of ionospheric TEC one hour ahead.%提出了一种利用人工神经网络提前1 h预报电离层TEC的简便方法.考虑到实际工程应用要求,没有使用其他空间天气参数,而是选择电离层TEC观测数据本身作为输入参数.输入参数为当前时刻TEC、一阶差分、相对差分和时间,输出参数为预报时刻TEC.利用文中介绍的GPS/TEC处理方法解算厦门站2004年电离层TEC观测数据,对预报方法进行评估,全年平均相对误差为9.3744%,预报结果与观测值相关性达到了0.96678.结果表明,利用人工神经网络方法提前1 h预报电离层TEC有很好的应用前景.

著录项

相似文献

  • 中文文献
  • 外文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号