Extreme-point weighted mode decomposition

- A new signal decomposition method termed EWMD with two algorithms is proposed.
- A new IMF criterion closely related to its definition is established.
- The accuracy, decomposition capability and orthogonality of EWMD are studied.
- EWMD is compared with EMD by analyzing synthetic and vibration signals of rolling bearing.

empirical mode decomposition (EMD) is an effective method for nonlinear and nonstationary signal analysis. In this paper a new signal decomposition method termed extreme-point weighted mode decomposition (EWMD) is proposed for improving the accuracy of EMD. In EWMD method, an newly intrinsic mode function (IMF) with physically meaning is defined to overcome the drawback of EMD that constructs mean curve by interpolating local extreme-points. Based on that, a new mean curve is constructed by using the weighting values of adjacent extreme-points for sifting process. Also a new IMF criterion closely related to its definition is established. We have deeply studied and compared the proposed method with EMD method by analyzing synthetic and mechanical vibration signals and the results show the superiority of proposed method in IMF accuracy, decomposition capability and orthogonality.