CESIgram: A Fault Feature Extraction Method for Rolling Bearings in Wind Turbine Equipment Based on Collaborative Filtering Correlation Spectrum

To address the difficulty of extracting weak fault features of rolling bearings in wind turbines under strong background noise, a fault feature extraction method based on the collaborative filtering correlation spectrum, named CESIgram, is proposed. The collaborative filtering correlation spectrum (CFCS) based on Block Matching 3D is designed to suppress random noise while preserving cyclostationary structures, resulting in a clearer cyclic spectral representation. A projection method along the cyclic frequency axis is proposed to obtain the carrier-based enhanced envelope spectrum. An integrated envelope spectrum index combining harmonic significance and periodic impact is proposed to quantify fault feature enrichment in different enhanced envelope spectra. The method works in three stages: spectral representation via Fast-SC, reformulation of the spectral correlation via CFCS, and adaptive band selection via CESI. The method successfully extracted fault characteristic frequencies and their harmonics in simulation and experimental signals under various strong noise conditions, while Fast Kurtogram, Autogram, Infogram, and Fast Entrogram failed to detect any fault-related peaks. Comparative analysis shows that the proposed method has significant advantages in noise suppression and fault feature extraction. The effectiveness is verified using simulation and experimental signals of rolling bearing faults in wind power equipment.