Revealing the Hot Deformation Behavior of AZ42 Mg Alloy by Using 3D Hot Processing Map Based on a Novel NGO-ANN Model

Mengtao Ning,Xiaomin Chen,Yongcheng Lin,Hongwei Hu,Xiaojie Zhou,Jian Zhang,Xianzheng Lu,You Wu,Jian Chen,Qiang Shen
DOI: https://doi.org/10.1016/j.jmrt.2023.10.073
IF: 6.267
2023-01-01
Journal of Materials Research and Technology
Abstract:The hot deformation behavior of AZ42 alloy was observed using thermal compression tests at a temperature scope of 250-400 degrees C and strain rate scope of 0.001-1 s-1. True stress-strain curves exhibited a combination of work hardening and dynamic softening features. A Northern Goshawk algorithm (NGO)-optimized artificial neural network (ANN) model was proposed. The established NGO-ANN model demonstrated impressive prediction accuracy, achieving a high determination coefficient of 0.991, a mean absolute percentage error of 3.51 %, and a root mean square error of 2.73. Subsequently, three-dimensional (3D) hot processing map based on the dynamic material model (DMM) theory was created. There were three different regions within the processing maps: the flow instability region (region A: 250-260 degrees C, 0.02-1 s-1, and region B: 300-400 degrees C, 0.01-0.1 s-1), high-power dissipation coefficient region (region C: 350-400 degrees C, 0.001-0.02 s-1, and region D: 300-350 degrees C, 0.5-1 s-1), and low power dissipation efficiency safety region (region E: the rest ones). Microstructural analysis revealed significant local plastic flow features in the flow instability region and a combination of coarse initial deformation grains and fine dynamic recrystallization (DRX) grains in the low power dissipation efficiency safety region. Fine and uniform grains were observed in the high-power dissipation efficiency region with DRX degree VDRX as high as 85.6 %, resulting in the best mechanical properties. Based on the established 3D hot pro-cessing map, the optimal process domains were determined. (c) 2023 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
What problem does this paper attempt to address?