An Improved Second-Order Multi-Synchrosqueezing Transform for the Analysis of Non-Stationary Signals

Kewen Wang; Yajun Shang; Yongzheng Lu; Tianran Lin

doi:10.37965/jdmd.2023.207

Authors

Kewen Wang Qingdao Key Rail Transportation Laboratory for Noise and Vibration Control and Automated Fault Diagnostic, Qingdao University of Technology, Qingdao 266525, China https://orcid.org/0000-0003-0258-6976
Yajun Shang Qingdao Key Rail Transportation Laboratory for Noise and Vibration Control and Automated Fault Diagnostic, Qingdao University of Technology, Qingdao 266525, China
Yongzheng Lu Qingdao Key Rail Transportation Laboratory for Noise and Vibration Control and Automated Fault Diagnostic, Qingdao University of Technology, Qingdao 266525, China https://orcid.org/0000-0002-9619-5396
Tianran Lin Qingdao Key Rail Transportation Laboratory for Noise and Vibration Control and Automated Fault Diagnostic, Qingdao University of Technology, Qingdao 266525, China https://orcid.org/0000-0001-6869-4617

DOI:

https://doi.org/10.37965/jdmd.2023.207

Keywords:

Synchrosqueezing transform, Non-stationary signals, Time-frequency operator, Fault diagnosis

Abstract

Second-order multi-synchrosqueezing transform (SMSST), an effective tool for the analysis of non-stationary signals, can significantly improve the time-frequency resolution of a non-stationary signal. Though, the noise energy in the signal can also be enhanced in the transform which can largely affect the characteristic frequency component identification for an accurate fault diagnostic. An improved algorithm termed as an improved second-order multi-synchrosqueezing transform (ISMSST) is then proposed in this study to alleviate the problem of noise interference in the analysis of non-stationary signals. In the study, the time-frequency (TF) distribution of a non-stationary signal is calculated first using SMSST, and then a δ function is constructed based on a newly proposed time-frequency operator (TFO) which is then substituted back into SMSST to produce a noise-free time frequency result. The effectiveness of the technique is validated by comparing the TF results obtained using the proposed algorithm and those using other TFA techniques in the analysis of a simulated signal and an experimental data. The result shows that the current technique can render the most accurate TFA result within the TFA techniques employed in this study.

Conflict of Interest Statement

The authors declare no conflicts of interest.