摘要
利用海口、广州、满洲里、兰州、北京和乌鲁木齐2003年的电离层电子浓度总含量(TEC)数据讨论了电离层TEC的混沌特性及其预报.采用自相关法确定延迟时间τ,综合Cao算法和饱和关联维法(G-P算法)的结果确定嵌入维数m.结果表明:当τ=2,m=6时,可对电离层TEC进行相空间重构,且重构后轨迹的李雅普诺夫指数均为正,应证电离层TEC具有混沌特性.在此基础上采用相空间重构和支持向量机相结合的方法对以上6个站点2003年的电离层TEC数据进行预报,将得到的结果同常用的加权一阶局域法与自适应预报进行比较,发现其预报效果最佳,预报的平均相对精度相对于加权局域法均提高10%以上,最大提升了31.1%;预报的相对精度大于95%的占50%以上,而自适应预报只有25%左右.
Chaotic properties of ionospheric total electron content （TEC）and prediction based on the TEC data over Haikou,Guangzhou,Manzhouli,Lanzhou,Beijing and Urumqi in 2003 was discussed in this paper.The auto correlation method was used to determine the time delay,and the result of Cao method and Saturated correlation dimension method （G-P）algorithm was combined to determine the embedding dimension m.The result shows whenτ=2,m=6,the phase of ionospheric TEC can be re-constructed and the Lyapunov exponent of the reconstructed traj ectory is positive,which proves that the ionospheric TEC has chaotic characteristics.On this basis,a combination of phase space recon-struction and support vector machine method was used to forecast the ionospheric TEC of above six station,and compared with the common used add-weighted one-rank local-region method and adaptive prediction,the prediction effect is best.The comparison shows that the mean relative precision is increased by more than 1 0% compared with the add-weighted method,and the maximum increase is 31.1%.And the proportion of prediction value whose relative precision more than 95% is above 50%, while adaptive prediction is only 2 5%.
作者
胡耀垓
彭铭
赵正予
周晨
Hu Yaogai Peng Ming Zhao Zhengyu Zhou Chen(School of Electronic Information, Wuhan University, Wuhan 430072, Chin)
出处
《华中科技大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2016年第9期26-31,共6页
Journal of Huazhong University of Science and Technology(Natural Science Edition)
基金
国家自然科学基金资助项目(41375007)
湖北省自然科学基金资助项目(2011CDA099)
