Abstract：To eliminate noises involved in dynamic weighting signal, empirical mode decomposition （EMD） is applied. For the end points of data sequence being usually not the extremes, the upper and lower envelopes swing sharply there, which cause end effect and result in low precision of the sifting results. A new method called the second extremes mirror method is proposed, considering the given antiend-effect method with low precision and strict requirement for data volume. The improved EMD method is adopted in dynamic weighting trial. The results indicate that the precision of this new method is very high, the new anti-end-effect method is the most effective one among the given methods, and the error of weighting can be controlled below 0.8%.
2008, Vol. 29 
