Using a neural network model and regression analysis to predict the wear and impact energy of Al-Si alloys modified with various modifiers and grain refiners using its hardness

Abstract

Measuring the wear resistance and impact energy of materials, especially aluminium alloys, is more complex and costly than hardness tests. This study used hardness tests as a low-cost, easy method for estimating wear rate and impact energy. The study examined the correlation between Si content and hardness, wear rate, and impact energy on six different case studies of Al-Si alloys with varying Si content (unmodified (1), modified with strontium (2), Na2SiF6 (3), Al-3B(4), Al-3TI-3B (5), and a combination of Na2SiF6 + Al-3TI-3B (6)). The measured Brinell hardness, wear, and impact energy for these alloys ranged from 48 to 98 HB, 385.6–36.8 mg/hr, and 15.2–2.9 J, respectively. The best results for predicting Si content, wear rate, and impact energy based on hardness were obtained using an artificial neural network (ANN) model and regression analysis on the combination of Al-3Ti-3B + Na2SiF6 (Case 6). The results showed high precision in predicting the silicon contents and wear rate/impact energy of Al-Si alloys based on hardness by ANN.