On the prediction of f○F2 using artificial neural networks

被引：2

作者：

Kong Qing-Yan ^{[1
]}

Liu Wen ^{[1
]}

Fan Jun-Mei ^{[1
]}

Jiao Pei-Nan ^{[1
]}

Feng Jing ^{[1
]}

Wang Jun-Jiang ^{[1
]}

机构：

[1] China Res Inst Radiowave Propagat, Natl Key Lab Electromagnet Environm, Qingdao 266107, Peoples R China

来源：

CHINESE JOURNAL OF GEOPHYSICS-CHINESE EDITION | 2009年 / 52卷 / 06期

关键词：

Neural networks; f(circle)F(2); Short-term prediction; Ionosphere;

D O I：

10.3969/j.issn.0001-5733.2009.06.003

中图分类号：

P3 [地球物理学]; P59 [地球化学];

学科分类号：

0708 ; 070902 ;

摘要：

A method to predict f(circle)F(2) one hour ahead using neural networks is developed. Based on the time variation characteristics of f(circle)F(2), six parameters are determined as inputs by time series analysis, which are the present observation of f(circle)F(2), one-day ahead f(circle)F(2) at the same time as the predicted one, the seven-day average f(circle)F(2) before the predicted hour and the current hour, the first difference of f(circle)F(2) and time of day. The output is the f(circle)F(2) one hour ahead. During high solar activity period, the average relative error is less than 6 percent and RMS less than 0.6 MHz, while during solar minimum, the average relative error is less than 10 percent and RMS less than 0.5 MHz.

引用

页码：1438 / 1443

页数：6

共 9 条

[1] Forecasting of ionospheric critical frequency using neural networks [J].