Optimization of Laser Cutting Parameters for Mild Steel using Regression Analysis and Differential Evolution Algorithm

The primary objective in the production of parts is to optimize the manufacturing process. As the industry recognizes the roughness of the cut product as one of the key criteria, it becomes critical to select the correct laser settings with minimum trial, error and at the lowest possible cost while using reliable techniques to achieve the desired surface finish. Due to the nonlinear nature of laser cutting, statistical analysis is necessary to obtain a satisfactory surface finish. In this study, experimental data sourced from literature were subjected to analytical processes. In the experimental design, L25 orthogonal array was used. The optimization process for the laser cutting parameters (laser power, cutting speed, and assist gas pressure) was implemented using regression analysis and a differential evolution algorithm. The regression model, with an R2 value of 83.21%, accurately predicted roughness based on these parameters. The model's effectiveness was further supported by the high correlation (R2 = 86.6%) between the experimental and predicted results. Using the differential evolution optimization method, the minimum surface roughness was calculated as 0.442 µm. This study provides a method for identifying optimal laser settings to achieve the desired surface roughness based on the obtained results.