Diff-CS-AE: A Dual Reassignment Differential Cyclostationary Analysis Scheme for Acoustic Emission-Based Machine Condition Monitoring

The acoustic emission (AE) technique serves as a robust alternative in machine condition monitoring, demonstrating particular sensitivity to incipient failures. Traditional cyclostationary analysis relies on prior knowledge of repetitive transients, limiting its use when the system’s dynamic model is unknown. A cyclostationary method is also urgently needed to handle the AE signal’s more broadband and abundant information characteristics. In light of this, a tailored dual reassignment differential cyclostationary analysis scheme for AE (Diff-CS-AE) signals is proposed in this study. Employing readily available historical data from healthy operational stage as the baseline, it fully exploits the discrepancy between real-time and baseline data across two dimensions of the cyclic spectral coherence (CSCoh) matrix and achieves early but on-time warning of incipient faults. The efficacy and superiority of the scheme are verified in a planetary gearbox diagnosis scenario and a case of bearing continuous monitoring.