应用径向基函数神经网络处理EMD方法中的边界问题

Disposal of Boundary Problem in EMD Method Applying RBF Neural Network

利用RBF神经网络分析方法对一个数据序列的两端加以延拓,分别得到两个附加的极大值点和极小值点,然后利用三次样条函数将附加的极值点与原时间序列的极值点连接起来,拟合出原始数据序列的上下包络线,实现了准确的EM D分解,较好地解决了其边界问题.

This paper extends two sides of a given data series using RBF neural network analysis method, then receives two additive maximum points and minimum points separately, next, connects the additive extremum points and extremum points of the original time series utilizing cubic spline function. In this way, the super and infra wrapping lines are drawn up and have achieved the exact EMD decomposition, which settle its boundary problem preferably.