Adaptive product cyclic spectrum and its application for fault diagnosis of rotating machinery

Rotating machinery (RM) serves as a critical component in modern industry, with widespread applications in diverse sectors ranging from energy production to manufacturing processes. Fault diagnosis of RM is essential for preventing unexpected failures and enhancing overall equipment reliability. However, strong interference sources and heavy environmental noise pose significant challenges for RM fault diagnosis in Industry 4.0. Therefore, a novel adaptive product cyclic spectrum (APCS) method is proposed for fault pattern (FP) extraction from signals containing complex interference components. In this study, the cyclic spectral coherence (CSCoh) is first applied to raw signals to reveal their cyclostationary characteristics. Subsequently, a modified fault to noise ratio (MFNR) indicator is introduced to assess the FP in the cyclic frequency spectrum generated from the CSCoh. Finally, the PCS is proposed to enhance FP by leveraging multiple cyclic frequency spectra, where the optimal PCS is adaptively determined using the MFNR. To validate the effectiveness of the APCS, a simulated signal, a planetary gearbox signal, and a wind turbine bearing signal are analyzed. The results demonstrate the superior performance of the APCS in extracting FPs from signals with complex interference, offering a robust method for RM fault diagnosis under challenging conditions.