Equal channel angular deformation process and its
neuro-simulation for fine-grained magnesium alloy①
neuro-simulation for fine-grained magnesium alloy①
(1. School of Mechanical Engineering and Automation,
Guizhou University of Technology, Guiyang 550003, China;
2. School of Manufacturing Science and Engineering,
Sichuan University, Chengdu 610065, China;
3. Materials Group, Department of Mechanical and Manufacturing Engineering,
University of Melbourne, VIC 3010, Australia)
Guizhou University of Technology, Guiyang 550003, China;
2. School of Manufacturing Science and Engineering,
Sichuan University, Chengdu 610065, China;
3. Materials Group, Department of Mechanical and Manufacturing Engineering,
University of Melbourne, VIC 3010, Australia)
Abstract: Fine-grained structure of as-cast magnesium AM60 alloy was obtained by means of equal channel angular deformation(ECAD) technique. Through analyzing the relationship between the load and the displacement under different working conditions, it is demonstrated that employment of back-pressure, multi-passages of deformation, and speed of deformation are the main factors representing ECAD working condition. As for ECAD process, a network composed of non-
linear neuro-element based on error back-propagation learning algorithm is launched to set up a processing mapping module for dynamic forecasting of load summit under different working conditions. The experimental results show that back-pressure, multi-passages and deforming speed have strong correlation with ECAD processing characteristics. On the metallographs of AM60 alloy after multi-passes ECAD, a morphology that inter-metallic compound Mg17Al12precipites on magnesium matrix without discrepancy, which evolves from coarse casting ingot microstructure, is observed. And the grains are refined significantly under accumulated severe shear strain. The study demonstrates feasibility of ECAD by using as-cast magnesium alloy directly, and launches an intelligent neuro-simulation module for quantitative analysis of its process.
linear neuro-element based on error back-propagation learning algorithm is launched to set up a processing mapping module for dynamic forecasting of load summit under different working conditions. The experimental results show that back-pressure, multi-passages and deforming speed have strong correlation with ECAD processing characteristics. On the metallographs of AM60 alloy after multi-passes ECAD, a morphology that inter-metallic compound Mg17Al12precipites on magnesium matrix without discrepancy, which evolves from coarse casting ingot microstructure, is observed. And the grains are refined significantly under accumulated severe shear strain. The study demonstrates feasibility of ECAD by using as-cast magnesium alloy directly, and launches an intelligent neuro-simulation module for quantitative analysis of its process.
Key words: magnesium alloy; ECAD; fine-grained structure; neuro-simulation