Dominance of Neural Networks in Additive Manufacturing Surface Roughness Control

This article examines the prediction of surface roughness control in additive manufacturing by comparing Linear Regression and Neural Network models. Through K-means clustering, we identify four clusters in the data, each associated with specific 3D printing parameters. Within each cluster, we analyze the factors influencing surface roughness and power efficiency. The primary objective is to forecast a target variable, considering metrics such as R-squared, Mean Squared Error, and correlation coefficient. Our research demonstrates that both Linear Regression and Neural Network models exhibit strong predictive capabilities, with the Neural Network outperforming the Linear Regression model. The Neural Network demonstrates superior performance metrics, reduced Mean Squared Error, and greater resilience to outliers. These results underscore the effectiveness of Neural Networks in achieving precision and identifying nonlinear patterns in surface roughness prediction. Thus, emphasizing the significance of employing advanced machine learning techniques, such as Neural Networks, in similar domains.