Dynamic grinding wheel wear-informed tooth surface roughness prediction for high-speed gear generating grinding

High-speed gears, serving as critical transmission components in new energy vehicles, operate at speeds exceeding 30000rpm. Meeting stringent specifications for tooth surface roughness is crucial for achieving low-noise and long-life performance of high-speed gears. However, accurate prediction of this roughness remains challenging. Existing physics-driven tooth surface roughness prediction models generally fail to consider the dynamic wear status of grinding wheels, while data-driven models struggle to adapt to diverse machining conditions. To address these limitations, this study proposes a hybrid-driven tooth surface roughness prediction method that accounts for the dynamic grinding wheel wear status. A geometric grinding worm model was constructed based on the enveloping principle and the random distribution characteristics of grains. By simulating the dynamic material removal behavior on the tooth surface, a micro-scale tooth surface topography model was further established. According to the grinding wheel wear status evaluation method, the actual grain size was modified by incorporating the time-varying characteristics of grinding parameters, and the dynamic prediction of grinding wheel wear status was realized based on these parameters. The physics-driven grinding worm model was optimized using the modified grain size, and a hybrid-driven model was developed to accurately simulate the tooth surface roughness during gear grinding. Verification via gear grinding experiments shows that the average prediction error of the proposed hybrid model is 5.491% lower than that of the traditional physics-driven model, demonstrating its superior effectiveness. In the batch production of high-speed gears, the hybrid model can be used to improve the fatigue performance and noise by optimize gear grinding process parameters.