|
Abstract: The
nonstationary vibration signal of a hydroturbine rotor is pre-processed
by means of frequency modulation and multi-rate filter banks firstly,
and then is analyzed based on adaptive Chirplet decomposition and
adaptive Gaussian Chirplet spectrogram. The optimization of the Chirplet
parameters is modified according to the vibration characteristics of
rotating machinery. Compared with the other time-frequency analysis
methods, it is demonstrated that the adaptive Gaussian Chirplet
spectrogram features high time-frequency resolution, immunity of cross
term interference, and applicability to analysis of nonstationary
vibration signal.
Key
words: Nonstationary
signal Chirplet transform Adaptive signal decomposition
Time-frequency analysis Hydroturbine Transients
CLC No: TH165.3
国家自然科学基金(50375076)、国家杰出青年科学基金(50425516)和教育部“跨世纪优秀人才培养计划”基金资助项目.
Received 20040702, received in revised form 20041220
|