Antinoise Bearing Fault Diagnosis Using Time-Reassigned Multisynchrosqueezing Transform and Complex Sparse Learning Dictionary

Bearings are core components of rotating machinery, and their operating status monitoring and fault diagnosis are crucial for equipment health management. Accurately identifying fault impulse signatures in bearing signals under complex operating conditions is a key challenge in bearing fault diagnosis. Therefore, this article proposes a bearing fault time–frequency diagnosis method [time-frequency complex sparse coding K-SVD (TFCSCK)] based on the time-reassigned multisynchrosqueezing transform (TMSST) and the improved k-singular value decomposition (KSVD) algorithm [complex KSVD (CKSVD)]. First, TMSST is used to obtain a high-resolution time–frequency representation (TFR) to enhance the accuracy of impulse signature localization. Second, to address the problem that the traditional KSVD algorithm only works in the real domain and ignores time–frequency phase information, a CKSVD algorithm is proposed. This algorithm utilizes complex sparse coding and dictionary updating to preserve the time–frequency phase characteristics, improving the robustness of feature extraction under complex interference. Third, a transient component time–frequency mask decomposition (TFTCD) algorithm is proposed. This algorithm preserves the time-domain waveform details of the fault impulse through mask-weighted separation and inverse transform reconstruction. Finally, the effectiveness of the proposed method is verified using numerical simulations and real fault signals. The experimental results show that TFCSCK improves the accuracy of inner race fault frequency extraction by 2.53% compared to TMSST and KSVD on the inner race data of the TYS1-8 platform. Based on measured data from aircraft engine bearings, even when both TMSST and KSVD fail, TFCSCK still extracts high-speed rotational frequency and inner race fault frequency.