摘要: We analyze the gravitational wave signals with a model-independent
time-frequency analysis, which is improved from the Hilbert-Huang transform
(HHT) and optimized for characterizing the frequency variability on the
time-frequency map. Except for the regular HHT algorithm, i.e., obtaining
intrinsic mode functions with ensemble empirical mode decomposition and
yielding the instantaneous frequencies, we propose an alternative algorithm
that operates the ensemble mean on the time-frequency map. We systematically
analyze the known gravitational wave events of the compact binary coalescence
observed in LIGO O1 and O2, and in the simulated gravitational wave signals
from core-collapse supernovae (CCSNe) with our method. The time-frequency maps
of the binary black hole coalescence cases show much better details compared to
those wavelet spectra. Moreover, the oscillation in the instantaneous frequency
caused by mode-mixing could be reduced with our algorithm. For the CCSNe data,
the oscillation from the proto-neutron star and the radiation from the standing
accretion shock instability can be precisely determined with the HHT in great
detail. More importantly, the initial stage of different modes of oscillations
can be clearly separated. These results provide new hints for further
establishment of the detecting algorithm, and new probes to investigate the
underlying physical mechanisms.