Chemometric Approach for Mechanical Properties Prediction during the Electromagnetic Casting Process

Abstract

In this study the mechanical properties (reduction of area, S-0, tensile strength, R-m, yield strength, R-p, and elongation, A) of EN AW 7075 aluminum alloy obtained by electromagnetic casting were investigated at different operating parameters: frequency (V), field strength (T) and current intensity (I). The predictive mathematical models using Response Surface Methodology, with second order polynomial (SOP) regression models, and Artificial Neural Network model (ANN), were afterwards compared to obtained experimental results. Analysis of variance and post-hoc Tukey's HSD test at 95% confidence limit ("honestly significant differences") have been utilised to show significant differences between various samples. SOP models showed good prediction capabilities, with high coefficients of determination (r(2)), 0.531-0.977, while ANN model performed even better prediction accuracy: 0.800-0.992. The optimal samples were chosen depending on mechanical properties of the product (S-0 = 50.49mm(2), R-m = 405.75Nmm(-2), R-p = 302.49Nmm(-2), A = 6.86%), using optimal operating parameters (V = 30 Hz, I = 250 A, T = 18 x 10(-3) At).