This paper explores the machining of aluminum alloy, focusing on optimizing and predicting surface roughness through advanced methods. The study investigates the optimization and prediction of surface roughness in milling aluminum alloy using cryogenically treated and non-cryogenically treated cutting tools.
A Taguchi L18 orthogonal array was implemented to determine the effects of cutting tool type, cutting speed, feed rate and depth of cut on surface roughness. Additionally, an artificial neural network model was developed to predict surface roughness. The back-propagation algorithm was used for training, and various architectures and learning algorithms, including Levenberg–Marquardt (LM), scaled conjugate gradient (SCG), quasi-Newton back propagation (BFGS), resilient back propagation (RP) and conjugate gradient back propagation (CGP), were evaluated.
Optimization results indicated that feed rate was the most significant factor affecting surface quality, contributing 36.41% according to analysis of variance. In the ANN the best predictive performance was achieved using the BFGS algorithm with a 4-13-1 network structure, yielding correlation coefficients (R2) values above 0.97 and low root mean square error (RMSE) and mean error percentage (MEP) values for both training and testing datasets.
The study concludes that the Taguchi method effectively optimized machining parameters, and the artificial neural network model demonstrated strong predictive accuracy, confirming its suitability for estimating surface roughness in milling processes.
