AI-based modeling and multi-objective optimization of ultraviolet nanosecond laser-machined sapphire

This study presents a two-step methodology, integrating support vector machine (SVM) and non-dominated sorting genetic algorithm II (NSGA-II) to model and optimize machining characteristics in laser-machined sapphire. In the first step, SVM was employed to predict the machined depth and surface roughness of sapphire substrates exposed to ultraviolet (UV) nanosecond laser pulses. A set of 27 systematic experiments, covering various levels of laser pulse energy, scanning speed, and hatching distance, was conducted to train SVM models. Both established SVM models showed remarkably low error values for the prediction of machined depth and surface roughness. An additional set of five experiments underscored the reliability of these SVM models. The laser machining outputs were extended by using the proposed SVM models to unveil the profound impact of processing parameters on machining characteristics. In the second step, the validated SVM models from the previous step were utilized as objective functions within the NSGA-II algorithm to maximize machined depth and minimize surface roughness simultaneously. This approach yielded a range of optimal solutions tailored to specific design requirements. It is found that by leveraging SVM in the first step to predict machined depth and surface roughness with exceptional accuracy, followed by the integration of NSGA-II for multi-objective optimization in the second step, the proposed approach offers a pioneering solution to unravel the intricate interplay of processing parameters and optimizing laser machining outcomes. These findings underscore the remarkable potential of the proposed two-step SVM–NSGA-II method in addressing a diverse array of engineering challenges.