A Complex Co-localization Local Mean Decomposition
Yuanduo Niu, Qianqiang Lin, Haoran Du · 2024
To address the issue of primary components intertwined with rotating assembly components during the separation process, we propose the Complex Co-localization Local Mean Decomposition method (CCLMD). This method replaces traditional linear interpolation for obtaining the average envelope function with piecewise cubic Hermite interpolation, enhancing the accuracy of local extrema identification. By applying a threshold, intrinsic mode function (IMF) components are extracted and subjected to Local Mean Decomposition (LMD). The separation process concludes once the IMF components are determined to represent pure frequency-modulated signals. Simulation results show that compared to Complex Empirical Mode Decomposition (CEMD) and Complex Local Mean Decomposition (CLMD), CCLMD effectively captures local signal characteristics, achieving up to 96.03% separation effectiveness, and maintaining 86.42% at 0 dB, with minimal fluctuation in performance due to signal-to-noise ratio variations. Experimental data further validate its effectiveness.