Mode Mixing Suppression Algorithm for Empirical Mode Decomposition Based on Self-Filtering Method

Abstract

The Hilbert-Huang transform (HHT) is a classic method in time-frequency analysis field which was proposed in 1998. Since it is not limited by signal type, it is generally applied in medicine, target detection and so on. Empirical mode decomposition (EMD) is a pre-processing part of HHT. However, EMD still has many imperfect aspects, such as envelope fitting, the endpoint effect, mode mixing and other issues, of which the most important issue is the mode mixing. This paper proposes a mode mixing suppression algorithm based on self-filtering method using frequency conversion. The proposed algorithm focuses on the instantaneous frequency estimation and the false components removing procedures, which help the proposed algorithm to update or purify the designated intrinsic mode function (IMF). According the simulation results, the proposed algorithm can effectively suppress the mode mixing. Comparing with ensemble empirical mode decomposition (EEMD) and mask method, the suppression performance is increased by 26%.