经验模态分解停止准则及在故障诊断中的应用

STUDY ON THE STOPPING CRITERION OF EMPIRICAL MODE DECOMPOSITION AND ITS APPLICATION IN FAULT DIAGNOSIS

停止准则是决定经验模态分解能否完整准确地提取出信号的内禀模态函数的重要因素之一,文中在总结现有的标准偏差(standard deviation,SD)准则、G.Rilling准则和能量差跟踪法的基础上,提出新的结合G.Rilling准则和能量差跟踪法的停止准则,并对这些准则进行仿真和实测数据的比较验证,试验结果表明,文中提出的停止准则优于SD准则和G.Rilling准则,为故障诊断提供更可靠的依据。

The stopping criterion is one of the important factors which determine the integrality and accuracy of the intrinsic mode functions（IMF） extracted by empirical mode decomposition（EMD）.Firstly,the standard deviation（SD） criterion,the G.Rilling criterion and the method of tracking for energy difference（MTED） are discussed and summarized.Then,a new stopping criterion is proposed,which integrates the G.Rilling criterion and MTED.At last,the simulation and case study are given out to compare various stopping criterion and demonstrate the effect of the proposed stopping criterion.The result suggests that the proposed criterion is very efficient.