Feature extraction from non-stationary ampltitude modulated broad-band signal using the Hilbert-Huang Transform

The traditional DEMON(Dectecion of Envelope Modulation on Noise) method is difficult to obtain the modulation characteristics when the signal modulation frequency or modulation depth changes with time. For non-stationary modulation characteristics, using Hilbert-Huang Transform (HHT) data adaptive features and high-resolution time-frequency analysis capabilities as well as mathematical morphology filter (MMF) on the signal noise reduction performance, a modified method “MH_DEMON” based on HHT and MMF is given. Simulation data and real acoustic radiated noise experimental data analysis shows that, MH_DEMON can extract non-stationary broadband amplitude modulation characteristics, which could suport both modulation source for the target classification and motion analysis.