A novel integrated degree of cyclostationarity indicator for gear micro-pitting monitoring of offshore wind turbines under marine non-Gaussian impulsive noise

Micro-pitting is a common and critical surface fatigue failure in offshore wind turbine (OWT) gearboxes. Given the high operation and maintenance costs of offshore installations, reliable early monitoring of micro-pitting propagation is essential. However, vibration-based monitoring of micro-pitting remains challenging, as the signals associated with micro-pitting are weak and easily obscured by background noise and other interference; moreover, the complex aerodynamic and hydrodynamic loads — such as wind gusts and wave impacts — frequently introduce strong non-Gaussian impulsive noise. These heavy-tailed transients can skew conventional metrics and mask the subtle damage features. This study presents a cyclostationarity-based monitoring indicator to characterize the propagation of gear micro-pitting using vibration signals. The indicator is based on the fact that micro-pitting-related surface damage generates cyclostationary random vibration components. To extract these components under impulsive contamination, cyclic correntropy is introduced to improve robustness to non-Gaussian impulses while preserving the micro-pitting information. The proposed indicator is validated using run-to-failure gear degradation data under different speeds and synthesized marine impulsive noise conditions. The results demonstrate that the proposed indicator consistently outperforms conventional vibration-based indicators across all tested conditions, with a particularly clear advantage under non-Gaussian impulsive noise. These findings demonstrate its potential for robust micro-pitting monitoring in OWT gearboxes.