Prediction of wear amounts of AZ91 magnesium alloy matrix composites reinforced with ZnO-hBN nanocomposite particles by hybridized GA-SVR model

Abstract

In this study, pure and hexagonal boron nitride (hBN)-doped zinc oxide (ZnO) nanoparticles with different doping ratios (1, 5 and 10 wt%) were synthesized and their structural morphological properties were investigated. ZnO-hBN nanocomposite particles were used as reinforcement material in AZ91 magnesium alloy. AZ91 powder reinforced with ZnO-hBN nanocomposite particles was combined by powder metallurgy. Hardness of AZ91 matrix composites was measured at 5 different points on each sample and averaged. The specimens were subjected to wear tests in a pin-on-disk test apparatus. As a result of the tests, the hardness value of pure AZ91 alloy was found to be 62 HB, while the hardness value of AZ91-ZnO-10 hBN sample was found to be up to 85 HB. In the wear tests, it was observed that ZnO-hBN additives resulted in low weight losses in wear losses and also decreased the friction coefficient of the hBN additive in the friction coefficient graphs. A hybrid GA (genetic algorithm)-SVR (support vector regression) model was proposed for the prediction of wear quantities of composites. The purpose of using a hybrid system combining SVR and GA is to improve the wear prediction by optimizing the hyperparameters of t and GA. The results are observed using four different kernel functions in the SVR algorithm, and then, a hybrid GA-SVR structure is proposed and compared. With the proposed method, 98.80% prediction success of the wear quantities of the composite is achieved by the hybridized GA.