.png){kind=icon}
.png){kind=icon}
Return
Feature Extraction Using Parameterized Multisynchrosqueezing Transform
.png){kind=icon}
PRE
AI
DOI：10.1109/JSEN.2022.3179165.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Parametrized time-frequency analysis (PTFA) can effectively improve time-frequency energy aggregation of non-stationary signal and immunity of cross term interference, but it exists the energy diffusion near the real instantaneous frequency. The improved multi-synchrosqueezing transform (IMSST) can improve the time-frequency energy aggregation, but it still has defects in processing strong FM and AM signals under noise interference. Therefore, in order to make use of their advantages and overcome their disadvantages, a novel parametrized multi-synchrosqueezing transform method based on weighted least square, IMSST and PTFA, namely PMSST is proposed in this paper. In the PMSST, the IMSST is designed to obtain the signal time-frequency representation with high energy aggregation. Then the ridge extraction algorithm is employed to extract the instantaneous frequency ridges of each mono-component signal. The weighted least square method is used to estimate the parameters of parameterized transform kernel. Finally, time-frequency spectrum is superimposed to obtain the time-frequency energy representation of the enhanced signal. The experiment results show that the PMSST can effectively process non-stationary signals with varying instantaneous frequency by the simulated signal and actual fault signals.
Keywords:
Time-frequency analysis
Transforms
Kernel
Sensors
Feature extraction
Interference
Frequency modulation
Parametrized time-frequency analysis
synchrosqueezing transform
weighted least squares
instantaneous frequency
feature extraction
Journal
IF:
4.5
Papers:
2.1W
Citations:
7.3W
Organization
View more organizations