Local maximum matching synchrosqueezing transform and its application in fault diagnosis of rotating machinery
Abstract Time-frequency analysis algorithms are widely used to characterize nonstationary vibration signals in the joint time-frequency domain. Synchrosqueezing transform (SST) can improve the energy concentration of a conventional time-frequency representation (TFR), but its TFR may exhibit spectral spreading and amplitude distortion for frequency-modulated (FM) signals whose instantaneous frequencies vary rapidly with time. To address this problem, this paper proposes a local maximum matching synchrosqueezing transform (LM-MSST). MLCT is used to obtain a matched TFR through chirp-rate compensation and kurtosis-guided demodulation-angle selection. A local-maximum-based synchrosqueezing operator then redistributes the matched MLCT coefficients to local-maximum frequencies while preserving signal reconstructability. The proposed method improves energy concentration, preserves amplitude information, and enhances noise robustness for FM signals. In this paper, the effectiveness of the proposed algorithm is verified using simulated and experimental signals.
Authors
- Zhu Yan (ORCID: https://orcid.org/0000-0002-4939-6312)
- Ke Ye (ORCID: https://orcid.org/0009-0004-5758-9669)
- Liang Sun (ORCID: https://orcid.org/0000-0003-2667-2932)
- Guangyu Liu
- Bin Guo
Institutions
- Beijing University of Technology (CN)
- China Railway Signal & Communication (China) (CN)
Publication Details
- Journal
- Engineering Research Express
- Published
- 2026-11-17
- DOI
- https://doi.org/10.1088/2631-8695/aec35e
- Primary Topic
- Machine Fault Diagnosis Techniques
- Type
- article
- Field-Weighted Citation Impact
- 0.00