采用具有局域控制特性的B样条函数直接拟合均值线，采用局域标准差终止准则判断原型模态函数（Proto-mode Function，PMF）的对称性，得到一种适用于流数据分析的快速经验模态分解（Empirical Mode Decomposition，EMD）算法。相比经典EMD算法，该快速算法不仅时间复杂度和空间复杂度均有所下降，而且分解精度基本维持不变。该快速算法适用于战场侦察传感网等流数据分析场合，实时性好，能够得到与序列等长的连贯本征模态函数（Intrinsic Mode Function，IMF）。

A new fast EMD (Empirical Mode Decomposition) algorithm was developed relying on moving mean averages of the extrema as combinations of B splines and local stopping criteria estimating the symmetry of proto-mode function, which is suitable for streaming data analysis. Compared to classical EMD, both of time complexity and space complexity of the fast algorithm have been decreased markedly, and the decomposition precision retains. It is especially appropriate for streaming data analysis such as in occasion of war sensor networks for its timeliness and continuous intrinsic mode functions acquired.