基于经验模态分解的非平稳信号趋势项消除

归纳概括了传统的趋势项消除方法,指出各类方法的优点和不足,提出了基于EMD(经验模态分解)的非线性、非平稳信号剔除方法.该方法通过数据驱动自适应构造基底函数IMF(本征模函数),再由若干阶IMF分量和剩余分量的重组获得趋势项,避免了对复杂趋势项的数学建模和分析计算.仿真结果表明,EMD法能够有效地提取和剔除非平稳信号中的复杂趋势项成分,获得平滑的趋向性信号.

Removal of Non-Stationary Signal Trend Items by Empirical Mode Decomposition

Following an overview of the strengths and weaknesses of conventional trend item-removal methods,this paper puts forward a method based on EMD(Empirical Mode Decomposition) to remove the trend items from nonlinear and non-stationary signals.The method defines adaptive IMF(Intrinsic Mode Function) function by data-driving and reconstructs trend items by some IMFs and residual components,and makes mathematical modeling and computing unnecessary for complex trend items.Simulation and test results show that the method effectively picks up and removes complex trend items from non-stationary signals and obtains smooth tendency signals.