| 利用人工神经网络预测电离层foF2参数 |
| |
| 引用本文： | 孔庆颜,柳文,凡俊梅,焦培南,冯静,王俊江.利用人工神经网络预测电离层foF2参数[J].地球物理学报,2009,52(6):1438-1443. |
| |
| 作者姓名： | 孔庆颜 柳文 凡俊梅 焦培南 冯静 王俊江 |
| |
| 作者单位： | 中国电波传播研究所青岛分所，电波环境特性及模化技术国家重点实验室，青岛 266107 |
| |
| 摘 要： | 利用人工神经网络技术实现了电离层foF2参数提前1小时预测.从foF2时间序列本身的变化特征出发，根据时间序列相关分析结果确定网络输入参数.选用当前时刻foF2值，预测时刻前一天的foF2值，预测时刻前7天foF2平均值，当前时刻前7天foF2平均值，foF2的一阶差分及表示当前时刻t的变量共六个参数作为神经网络输入，下一时刻值作为神经网络输出.对于太阳活动高年平均预测相对误差小于6%，均方根误差小于0.6 MHz，太阳活动低年平均预测相对误差小于10%，均方根误差小于0.5 MHz
|
| 关 键 词： | 神经网络 foF2 短期预测 电离层 |
| 收稿时间： | 2008-12-11 |
| 修稿时间： | 2009-3-29 |
On the prediction of foF2 using artificial neural networks |
| |
| KONG Qing-Yan,LIU Wen,FAN Jun-Mei,JIAO Pei-Nan,FENG Jing,WANG Jun-Jiang.On the prediction of foF2 using artificial neural networks[J].Chinese Journal of Geophysics,2009,52(6):1438-1443. |
| |
| Authors: | KONG Qing-Yan LIU Wen FAN Jun-Mei JIAO Pei-Nan FENG Jing WANG Jun-Jiang |
| |
| Institution: | China Research Institute of Radiowave Propagation, National Key Laboratory of Electromagnetic Environment, Qingdao 266107, China |
| |
| Abstract: | A method to predict oF2 one hour ahead using neural networks is developed. Based on the time variation characteristics of oF2, six parameters are determined as inputs by time series analysis, which are the present observation of oF2, one-day ahead oF2 at the same time as the predicted one, the seven-day average oF2 before the predicted hour and the current hour, the first difference of oF2 and time of day. The output is the oF2 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. |
| |
| Keywords: | Neural networks foF2 Short-term prediction Ionosphere |
|
| 点击此处可从《地球物理学报》浏览原始摘要信息 |
| 点击此处可从《地球物理学报》下载免费的PDF全文 |
|
