摘要：The problem of mode mixing in the empirical mode
decomposition (EMD) vastly restricts extraction of the useful
signal and the performance of denoising. An adaptive and
convenient demixing algorithm to promote the separation ability
of noise and expected signal is proposed. After decomposing the
original signal, a multiple-input and multiple-output mechanism
is designed to realize optimal reconstruction and decomposition
of the first order intrinsic mode functions (imfs). The correlation
of each imf is utilized to ascertain appropriate reconstruction
order without additive introduction of parameters and
components. Compared with EMD, the proposed algorithm
maintains the better demixing performance indicated in the
measurement results of synthetic signals and real-life frequency
modulated continuous wave radar signal.
