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
Abstract
The hot deformation behavior of AZ42 alloy was observed using thermal compression tests at a temperature scope of 250-400 oC 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 oC, 0.02-1 s-1, and region B: 300-400 oC, 0.01-0.1 s-1), high-power dissipation coefficient region (region C: 350-400 oC, 0.001-0.02 s-1, and region D: 300-350 oC, 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 processing map, the optimal process domains were determined.