Identifying the occurrence of gear contact fatigue failure as early as possible is essential for condition-based maintenance (CBM). Vibration signals can be used to identify gear contact fatigue. However, the use of vibration signals can be challenging due to its complexity, compounded by lower levels of vibration during the initial stages of contact fatigue. The present study details a new algorithm that integrates stand-alone features to correlate the vibrational signal with early failure occurrence. The study aim is to identify the failure in the early stages, before reaching the ISO 6336-5 stopping criterion of 4% damaged area. A damage induction on the flank of helical gears is applied to simulate and characterize the failure occurrence. Damping characteristics with impact evaluation, Kurtosis analysis and the monitoring of the Gear Meshing Frequency are applied to characterize the failure signature. This strategy stands out by the integration of these stand-alone features and their behavior. The algorithm's capacity is verified through durability tests, promoting the natural evolution of this failure mode. Results show a success rate of above 80% at identifying the failure presence before the stopping criterion limit.
Read full abstract