Research on Magnetic Rollers for Recovering Non-Ferrous Metals from End-of-Life Vehicles Employing Machine Learning

Abstract

Recovering copper foil and crushed aluminum from end-of-life vehicles (ELVs) is a significant issue in the recycling industry. As a key technology for sorting aluminum, copper, and other non-ferrous metals, eddy current separation (ECS) is efficient in isolating the non-ferrous metals according to their different electrical conductivity and density. However, further research is still needed in the separation of large-size copper foil and crushed aluminum from scrapped vehicles. In this study, support vector regression (SVR) and the sparrow search algorithm (SSA) are exploited for the first time to be used in optimizing the Halbach magnetic roller. Firstly, the numerical simulation results are based on the response surface methodology (RSM). Then, the accuracy of four kernel functions employing SVR is compared to select a kernel function. The sparrow search algorithm (SSA) is proposed to optimize the structural parameters of the Halbach magnetic roller, concentrating on the above-selected kernel function. Meanwhile, the parameters are confirmed. Numerical simulation results indicate that machine learning for magnetic roller optimization is feasible.