Abstract: The Empirical Mode Decomposition （EMD） is an advanced method for signal analysis. Especially, it has a significant effect on EEG and other nonlinear and non-stationary signals. But there is an involved end issue in the courage of getting two envelops of the data using spline interpolation. On the basis of existing algorithms for solving the problem, a new method is proposed. Generally, it fits the sequence near the end with a fitting function, and then extends a extreme point outside the endpoint with the fitting function. Finally, it utilizes the mirror extension method to realize the process of EMD algorithm. For the extended data, the spline does not swing at both ends of the data. The experimental results prove that the new method can decompose EEG effectively.

Key words: Empirical Mode Decomposition（EMD）, data extending, mirror extension, ending effect

摘要： 经验模态分解（EMD）是一种先进的数据处理方法，对脑电信号（EEG）等非线性非平稳信号的处理非常有效。但是其在利用三次样条曲线构造上下包络时，端点附近的包络存在严重的摆动。针对该问题，在镜面延拓算法的基础上，提出了二次延拓算法。根据邻近端点的数据计算出该信号在端点处的拟合函数；利用该拟合函数在左右端点各延拓出一个极值点；采用镜面延拓算法对延拓后的信号进行EMD分解。算法考虑了信号端点处的变化趋势，使得端点处的延拓更加合理，从而使三次样条曲线在端点处不会出现大的摆动。仿真结果表明，该算法能有效地对脑电信号进行分解。

关键词: 经验模态分解, 数据延拓, 镜面延拓, 端点效应