Rolling bearing fault features under actual operating conditions are easily masked by strong noise and complex interference, making periodic impulse extraction difficult. To address this problem, a Dual-Source Cross-Correlation Mode Matching Filtering (DSCCMMF) method is proposed. A dual-source framework combining Cuckoo Search-optimized Maximum Correlated Kurtosis Deconvolution (CS-MCKD) and Successive Jump and Mode Decomposition guided by Adaptive Correlated Kurtosis (SJMD-ACK) characterizes fault information from global periodic and local transient perspectives. Subsequently, an adaptive Cross-Correlation Impulse Atom (CCIA), guided by the minimum of the cross-correlation envelope, is constructed for signal reconstruction. By concentrating fault-related components while preserving the original temporal structure, DSCCMMF improves weak periodic impulse representation. Simulation and experimental results demonstrate effective interference suppression and clearer fault-related periodic impulses while maintaining the signal temporal structure.

Rolling bearing,Fault diagnosis,Periodic impulse feature enhancement,Waveform reconstruction