APEEMD及其在转子碰摩故障诊断中的应用
Adaptive Partly Ensemble Empirical Mode Decomposition and Its Application for Rotor Rubbing Fault Diagnosis

总体平均经验模态分解(ensemble empirical mode decomposition, 简称EEMD)是抑制经验模态分解(empirical mode decomposition, 简称EMD)模态混叠的有效方法，针对EEMD分解效果依赖于添加噪声的大小、筛分次数和总体平均次数等参数的选择及噪声残留大、分解不完备等问题，提出了自适应部分集成经验模态分解。该方法通过成对地向目标信号加入自适应噪声，并对每个内禀模态函数(intrinsic mode function,简称IMF)自动选择筛选次数，通过排列熵检测筛分出高频IMF，再对剩余信号进行EMD分解。将提出的方法应用于仿真和转子碰摩故障试验数据分析，结果表明提出的方法能够有效地应用于转子碰摩故障诊断，而且在分量的精确性、完备性和模态混叠的抑制等方面优于EEMD方法。
Ensemble empirical mode decomposition (EEMD) was an effective method for restraining mode mixing of empirical mode decomposition (EMD). However, the effect of EEMD relies on the selection of the size of the added noise, times of iteration and ensemble, and large residual noise will result in an incomplete decomposition. In light of such problems, the adaptive partly ensemble empirical mode decomposition (APEEMD) method was proposed. APEEMD added the adaptive noise in pairs to the targeted signal and automatically set the iteration times for each intrinsic mode function (IMF). Then, the permutation entropy was employed to detect the high frequency IMFs, and the residual signal was decomposed using EMD. The simulation and rotor rubbing experimental results indicate that the proposed method was effective for fault diagnosis and superior to EEMD in accurate decomposition and inhibition of mode mixing.