Gear Meshing Impact Characteristics Extraction Based on In-Situ Signal Perception

The early fault signal of mechanical equipment is often weak, and substantial energy loss occurs when this signal propagates through multiple interfaces to traditional external sensors. This transmission loss complicates the extraction of early fault characteristics using conventional external vibration signal analysis methods. To address this issue, this study proposes an in-situ measurement method for extracting gear meshing impact with torsional vibration compensation. Initially, four acceleration sensors are evenly distributed on the gear shaft near the gear for in-situ measurements. Subsequently, torsional vibration signals are extracted to compensate for each sensor's signals. The gear's low-frequency rotational speed signal is obtained through filtering using a filter bank, and the gear's instantaneous phase is extracted through the Hilbert transform. Finally, the instantaneous gear meshing impact signal is obtained by performing a projection transformation and phase alignment based on the instantaneous phase of each sensor. The effectiveness of this method is validated through an early crack failure experiment on a gearbox. The results demonstrate that the meshing impact extraction method is more sensitive to fault signals and possesses higher practical value for the early fault diagnosis of gearboxes.