利用神经网络预报短期电离层TEC变化  

Short-term ionospheric TEC change prediction by neural network

作　　者：李淑慧[1] 彭军还[1] 徐伟超[1] 杨红磊[1] 

机构地区：[1]中国地质大学(北京)土地科学技术学院,北京100083

出　　处：《测绘科学》2013年第1期8-9,12,共3页Science of Surveying and Mapping

基　　金：国家自然科学基金项目(41104025;41074009);国土资源部公益性行业科研专项(200911015)

摘　　要：利用神经网络,基于IGS提供的(40°N,115°E)网格点TEC数据,本文构造了该点处提前一天的TEC预报模型。神经网络模型的预测目标是待预测日一天内的12个TEC数值。输入参数包括预测日前一日的太阳黑子数、地磁Kp指数、预测日前27天的太阳黑子数三角函数拟合参数,以及预测日前2天16个Kp指数的多项式拟合参数。实验结果表明训练后的神经网络模型可以反映出不同季节的TEC周日变化以及地磁暴情况下的TEC特征。The prediction model one day before was constructed through Neural Network （NN）, based on the ionospheric TEC data provided by 1GS at the gridding point （40°N, 115°E） in the paper. The Neural Network model is to calculate the 12 TEC series on the predicted day. The input parameters include the sun spot number, the geomagnetic Kp index on the day before the predicted day, the trigonometric function fitting parameters of the sun spot number series on all the 27 days before the predicted day, and the polyno- mial fitting paraneters of 16 Kp indexes on both the 2 days before the predicted day. The experimental result indicated that the trained NN model can be used to reflect the TEC change characteristics, diurnal at different levels of solar activities as well as for the iono- spheric storm in case of the geomagnetic storm.

关 键 词：电离层 电子总量(TEC) 地磁Kp指数 太阳黑子数 神经网络 

分 类 号：P228[天文地球—大地测量学与测量工程]