Improvement of local mean approximation in empirical mode decomposition for gear fault detection

Empirical mode decomposition (EMD) is an adaptive time-frequency domain signal processing method that is completely driven by data itself. The cubic spline interpolation method has been used to approximate the local mean in the sifting process of EMD. This study explores approaches to improve local mean approximation to obtain better EMD performance. A modifi ed monotone piecewise Hermite interpolation (MMPHI) method is applied to envelope mean approximation, because it demonstrates advantages over the cubic spline method. A type of direct approximation of the local mean, i.e., the windowed local mean (WLM) approach, is also investigated and its merit in identifying impulses is demonstrated.