EMD（经验模态分解，全称Empirical Mode Decomposition，一般指EMD算法）是Hilbert-Huang变换（HHT）的核心算法。
经验模态分解(EMD)算法是通过算法过程定义的，而并非由确定的理论公式定义的，所以对其进行准确的理论分析非常困难，我们目前只能借助大量的数字仿真试验不断对其性能进行深入的研究。 EMD算法的目的在于将性能不好的信号分解为一组性能较好的本征模函数（IMFIntrinsic Mode Function ），且IMF须满足以下两个性质：
(1)信号的极值点(极大值或极小值)数目和过零点数目相等或最多相差一个；
(2)由局部极大值构成的上包络线和由局部极小值构成的下包络线的平均值为零。
EMD算法的计算步骤如下：
（1）找出原数据序列X(t)的所有极大值点和极小值点，将其用三次样条函数分别拟合为原序列的上和下包络
线；上下包络线的均值为m1；将原数据序列减去m1可得到一个减去低频的新序列h，即h1=X(t)－m1；
一般h1不一定是平稳数据序列，为此需对它重复上述过程。如h1的包络均值为m11，则去除该包络平均所代表的低频成分后的数据序列为h11，即h11=h1－m11
重复上述过程，这样就得到第一个本征模函数分量c1，它表示信号数据序列最高频率的成分。(EMD (Empirical Mode Decomposition, generally referred to as EMD Algorithm) is the core algorithm of the Hilbert-Huang Transform (HHT).
The Empirical Mode Decomposition (EMD) algorithm is defined by the algorithm process and is not defined by the definite theoretical formula. Therefore, it is very difficult to conduct accurate theoretical analysis on it. We can only rely on a large number of digital simulation tests to continuously evaluate its performance. Conduct in-depth research. The purpose of the EMD algorithm is to decompose a signal with poor performance into a set of better-performing IMMF ntrinsic mode functions, and the IMF must satisfy the following two properties:
(1) The number of extreme points (maximum or minimum values) of the signal and the number of zero-crossings are equal or at most one difference;)