Multi-objective optimization and prediction of milling parameters for Al 6061 alloy using desirability analysis and artificial neural networks

Abstract This study focuses on developing a predictive model for surface roughness and material removal rate (MRR) in the machining of Al 6061 using a three-axis CNC vertical milling machine under dry-cutting conditions. The Taguchi design of experiments (L9 orthogonal array) and Artificial Neural Networks (ANNs) were employed to optimize and predict machining performance. The effects of milling parameters, including spindle speed, feed rate, and depth of cut, were assessed using Analysis of Variance (ANOVA), and a neural network model of two layers and fifteen neurons was constructed. ANOVA results identified spindle speed as the most influential parameter, contributing 88.76% to average surface roughness (Ra) and 65.16% to material removal rate. The experimental results indicated that best machining parameters delivered both a 1.52 μm Ra finishing and 179.857 mm 3 min −1 MRR maximum. The prediction accuracy of Ra and MRR through the feed-forward ANN illustrated mean squared errors of 0.003 and 1.1584 respectively. As a consequence, Taguchi’s method and ANN modeling have been combined to develop a comprehensive system for milling parameter optimization using a smaller number of experiments while at the same time delivering superior prediction precision. The study offers critical insights about improving Al 6061 milling quality and productivity that extends to precision manufacturing applications.