Study on Microstructure, Properties and Processing Technology of Rare Earth Wrought Magnesium Alloy Based on Neural Network

Abstract

In the past decades, magnesium alloys have attracted great attention. As the lightest structural metal, magnesium alloy is widely used in 3C and aerospace fields. In this paper, the as-cast commercial Mg-Y-Zr alloy was used as the initial material, and the fine-grained Mg-Y-Zr alloy bar with a diameter of 5mm was obtained by two times accumulative large-ratio extrusion process. The hot working diagram of the alloy based on neural network model and Muthy instability criterion is established. Considering the influence of strain on hot working diagram, a three-dimensional working diagram of the alloy is established. The microstructure evolution, mechanical properties and work hardening behavior of mg-y-Zr and Mg-Zn-Zr-xY during extrusion were studied by optical microscope and mechanical property test. The results show that the mechanical properties models of Mg-Y-Zr and Mg-Zn-Zr-xY alloys are constructed based on BP neural network, and the tensile strength, yield strength, elongation and area shrinkage of the alloys are predicted. The prediction correlation is 0.99978, and the errors are all within 3%.