This study aims to investigate the effect of bionic crescent-shaped directional micro-textures on bearing surfaces to mitigate premature wear and catastrophic failure under starved lubrication conditions.
A coupled thermo-mechanical finite element model was developed to analyze the effects of key texture parameter – texture width (B), inter-texture spacing (M), transverse pitch (L), inclination angle (?) and texture length (W) – on bearing deformation, equivalent stress distribution and temperature fields. Furthermore, a Generalized Regression Neural Network (GRNN) surrogate model was constructed and integrated with the Non-dominated Sorting Genetic Algorithm II (NSGA-II) for multi-objective optimization of texture geometric parameters.
Under static loading, texture width (B) and transverse pitch (L) exerted significant influences on stress distribution; specifically, B = 0.1 mm increased the maximum equivalent stress by 27.4% compared with the untextured bearing. Under thermo-mechanical coupling, increasing L from 1 mm to 1.2 mm raised the maximum temperature from 47.98 °C to 72.43 °C and the maximum equivalent stress from 51.15 MPa to 94.68 MPa. The GRNN model predicted equivalent stress and temperature with high accuracy [R2 = 0.98, mean squared error (MSE) = 0.17; R2 = 0.99, MSE = 1.06]. Finite element simulations validated the optimized texture parameters, with relative errors being constrained within 6%, which effectively alleviates stress concentration and high-temperature zones.
The integrated GRNN-NSGA-II framework offers a novel computational approach for optimizing micro-texture parameters, enabling precise regulation of lubrication performance under suitable texture parameter ranges without extensive experimental testing.
