Améliorations de la méthode EEMD

Abstract

The objective of this thesis of magister was to study various possibilities of improving the EEMD (Ensemble Empirical Mode Decomposition) method to reduce the number of tests. The EEMD algorithm was proposed to solve the problem of mixing modes present in the original EMD method. However, the effectiveness of EEMD based on a large number of tests which becomes a major drawback. Four methods of improvement (MEEMD, CEEMD, CEEMDAN, SEEMD) were considered and analyzed. Comparative studies of these algorithms and EEMD have been conducted to show the improvements made by each of these methods in relation to the EEMD in terms of the error signal reconstruction, the RMSE and the number of trials or time calculation. In addition, a comparative study of these four methods, based on the number of tests showed that SEEMD is the most effective. The analyzes were made by numerical simulations.