非线性动态系统相关性推断方法与应用

Dependence inferences in nonlinear dynamic system and applications

给出的基于分数维数的推断方法，用于推断两列观测数据是否来自于同一非线性动态系统，以解决非线性经济预测中的变量选择问题；随后对利用ＢＤＳ检验推断非线性可预测性的方法进行了讨论．在一个决策支持系统中，用该方法对国际有色金属市场价格趋势非线性特征进行统计推断与辅助建模，取得了较好效果．

In this paper, a statistical inference method based on the fractional dimension is proposed. This method is designed for testing the nonlinear dependence between time series, and can be used in variables selecting for economic forecasting. The BDS test is also discussed. Application results show that the methods is effective.