论文部分内容阅读
利用120°E、45°N上空的2008年年积日101~150d时间段内共600个电离层格网TEC数据,分析了该点上空电离层TEC参数的混沌特性,发现其关联维数为2.263 2,嵌入维数m=5,最大Lyapunov指数为0.083 3,该TEC时间序列具有混沌的特征,存在混沌现象。利用加权一阶局域法对TEC时间序列进行预测时,提出了利用夹角余弦和聚类分析方法对相似相点进行选择的方法,结果表明,在5维相空间中,该方法除在第4分向量略不及欧氏距离和夹角余弦方法外,其余4个分向量均优于后两种方法。利用该方法选择的相似相点进行一阶局域预测时,得到的标准差STD(0.618TECU)和RMS(0.623TECU)均小于欧氏距离和夹角余弦得到的STD和RMS,说明该方法可以准确地搜索到与基准点相关性更强的相似相点,预测精度更高。
Based on the TEC data of 600 ionospheric grids over a period of 101-150 d on the annual day 2008 at 120 ° E and 45 ° N, the chaotic characteristics of the ionospheric TEC parameters at that point were analyzed. The correlation dimension was found to be 2.263 2. The embedding dimension m = 5 and the maximum Lyapunov exponent 0.083 3. The TEC time series has chaos characteristics and chaos. When using weighted first-order local method to predict TEC time series, a method of selecting similar phase points by using the included angle cosine and clustering analysis method is proposed. The results show that in the 5-dimensional phase space, 4-point vector is slightly less than the Euclidean distance and included angle cosine method, the other four sub-vectors are better than the latter two methods. The first-order local prediction of similar phase points selected by this method shows that the standard deviations STD (0.618TECU) and RMS (0.623TECU) are both less than the Euclidean distance and the cosine of the angle obtained by STD and RMS, indicating that the method can Accurately searching for similar points with stronger correlation with the reference point, the prediction accuracy is higher.