Dynamic Modeling of Distributed Wear-Like Faults in Spur Gears: Simplified Approach with Experimental Validation

Dynamic models of gears are recognized for offering a promising platform for gaining a profound understanding of the dynamic response, particularly the vibration signature. Wear is considered among the most common and concerning fault mechanisms in gears, yet its recognition and subsequent diagnosis remain challenging. In this study, we introduce an existing dynamic model of spur gear vibrations and extend its validation for distributed wear-like faults. The novelty of this work lies in addressing the complexities associated with modeling distributed faults using simplified yet sophisticated approaches. These involve variance among defected teeth, calculation of time-variant gear mesh stiffness, and consideration of the forces induced by the fault. The model is validated through pioneering controlled experiments, analyzing dozens of degrading distributed wear-like faults. This comparison verifies our capability to generate realistic simulations of vibration signals from worn gears. To bridge the discrepancy between the induced and simulated faults, the model first constructs the healthy profile of the inspected gear, incorporating manufacturing errors and tooth modifications. Subsequently, meticulous photography enables the replication of faults in the model with a high resemblance to the experiment. The results demonstrate a strong correlation between measured and simulated signals, as verified through trend analysis of features extracted from synchronous average signals in both the cycle and order domains. This study lays the groundwork for in-depth investigation into the physics of gear wear, paving the way for potential applications such as fault severity estimation and intelligent fault diagnosis in future studies.