【摘要】 改进了混沌系统中的两个重要特征量:嵌入维数和时间延迟的计算,根据计算得出的上述两个参数重构相空间;然后在相空间中作轨迹的线性拟合,选择轨迹中的最近邻点作一次性的预测。提出的算法在相空间中很好地把轨迹的线性拟合与最近邻方法结合起来,解决了现有的时间序列分析和预测算法中主观性太强的缺点,通过对话务量时间序列和太阳黑子时间序列的验证,与其它算法相比,该算法的分析结果稳定而准确、预测精度高、运行时间比较短。更多还原

【Abstract】 This paper improved the computing of two important characteristic measure embedding dimension and time delay in chaos system,and reconstructed phase space based on these two parameter.And then it simulated track li-nearly in phase space,selecting nearest neighbor for one time predicate.The new algorithm combines the linear track simulation and the nearest neighbor method well,sloving disadvantage that the subjectivity is too strong in existing time serials and predicate method.By validating the phone number time serials and sunspot serials,comparing other algorithm,the analysis result of our method is steady and exact,predicate precision of it is high and the running time of it is short.更多还原