Surface Tribological Properties Enhancement Using Multivariate Linear Regression Optimization of Surface Micro-Texture

Abstract

This work aims to provide a comprehensive understanding of the structural impact of micro-texture on the properties of bearing capacity and friction coefficient through numerical simulation and theoretical calculation. Compared to the traditional optimization method of single-factor analysis (SFA) and orthogonal experiment, the multivariate linear regression (MLA) algorithm can optimize the structure parameters of the micro-texture within a wider range and analyze the coupling effect of the parameters. Therefore, in this work, micro-textures with varying texture size, area ratio, depth, and geometry were designed, and their impact on the bearing capacity and friction coefficient was investigated using SFA and MLA algorithms. Both methods obtained the optimal structures, and their properties were compared. It was found that the MLA algorithm can further improve the friction coefficient based on the SFA results. The optimal friction coefficient of 0.070409 can be obtained using the SFA method with a size of 500 µm, an area ratio of 40%, a depth of 5 µm, and a geometry of the slit, having a 10.7% reduction compared with the texture-free surface. In comparison, the friction coefficient can be further reduced to 0.067844 by the MLA algorithm under the parameters of size of 600 µm, area ratio of 50%, depth of 9 µm, and geometry of the slit. The final optimal micro-texture surface shows a 15.6% reduction in the friction coefficient compared to the texture-free surfaces and a 4.9% reduction compared to the optimal surfaces obtained by SFA.