A novel approach for end effect processing of empirical mode decomposition

The end effect appears as a troublesome problem in application of empirical mode decomposition (EMD) to decompose the complex signal into the sum of limited number of IMFS. A novel end effect processing method is proposed to solve this problem. By studying on physical property of the heat conductivity equation, end effect processing method based on PDE is proposed. The extrapolation of the signal through PDE at both endpoints to form the primary expansion signal, then the primary signal is further expanded through expansion, and EMD is performed on the resulting signal to obtain reduced end effects. With the proposed method, the end effect can be controlled in both ends of the signal and the middle component can be correctly decomposed. Experiments on simulative and real signals verify that the proposed method can enormously decrease the end effect of EMD.