Data-driven approach to the optimization of laser-based microtextured PCD tools

Abstract This study focuses on the use of infrared nanosecond laser technology to prepare microgroove textures on the surface of polycrystalline diamond (PCD) tools, and systematically explores the effects of parameters such as laser power, scanning speed and defocus on the geometric characteristics of microgrooves. By constructing an experimental data set containing key processing parameters and microgroove morphology features, a machine learning model is introduced to model and predict the nonlinear relationship between complex parameters. The CatBoost model performs best, accurately describing the variation in depth and width of the groove, and the curvature of the fillet at the groove opening. Laser power has a significant positive correlation with microgroove depth and width, while defocus is a key factor affecting curvature, and there are complex interactive effects between parameters. Combining Gaussian process-based optimization and multi-objective genetic algorithm, the optimal laser parameter combination of microgroove depth and curvature is obtained, which provides a technical means to improve the friction reduction and wear resistance of microgroove texture tools in composite material processing.