OPTIMIZATION OF WEDM PARAMETERS USING MACHINE LEARNING: A COMPARATIVE ANALYSIS OF SELECTED REGRESSION MODELS

This paper discusses the influence of selected parameters (ON, OFF and SV) on surface quality and machining time during WEDM. The study used a multivariate ANOVA to test the statistical significance of the factors considered. In addition, machine learning algorithms such as Support Vector Regression (SVR) and polynomial regression were applied to further investigate the influence of the parameters. A comparative analysis of the algorithms showed that models based on the Support Vector Regression algorithm perform the best in predicting the roughness parameter Ra and machining time separately. SVR performs better with nonlinear data, making it more accurate in predicting machining time values than simpler regression models. The study showed that all ON, OFF and SV parameters have a significant effect on surface roughness and machining time. Further analysis of the experimental results showed that the application of machine learning methods to optimise the EDM process, by selecting appropriate operating parameters for the WEDM machine, resulted in better surface quality and reduced machining time. These results have practical significance for the metalworking industry and indicate potential opportunities to improve the performance of the WEDM process.