Method of ensemble empirical mode decomposition with partial adaptive noise
In order to solve the problem of large noise number and computation amount demand in ensemble empirical mode decomposition(EEMD)and its improved forms,the extrema and energy variation law of white noise intrinsic mode function(IMF)are statistically analyzed,and the empirical formula of the white noise IMF extrema number varying with length and order is summarized.The high-order IMF of white noise can't effectively adjust the extrema distribution of the signal,and an EEMD with partial adaptive noise(EEMDPAN)is proposed.Compared to complete EEMD with adaptive noise(CEEMDAN),EEMDPAN has two improvements:it does not use all independent adaptive noise,but instead uses complementary adaptive noise that adds up to 0 in pairs;does not add all levels of adaptive noise,but stop at a certain level in the middle,and then use the classic EMD method.Decomposing two artificial signals,experiments show that EEMDPAN effectively inherits the ability of EEMD to suppress modal aliasing.Compared to CEEMDAN,the computational complexity is reduced to one-third,and the low-order component signal of the decomposition result has less additional noise,while the high-order component signal has higher credibility.
国家科技期刊平台
NSTL
万方数据
