Applied Mathematics & Information Sciences
Abstract
An important research trend in metal forming is to predict high temperature flow stress of materials during deformation. In the conventional models, there exist difficulties in the regression analysis based on the experimental results to obtain the model constants. Support vector machine (SVM) is a new technology for solving classification and regression. In this study, a novel accurate and rapid prediction of high temperature flow stress of AZ80 magnesium alloy with particle swarm optimization-based support vector regression (PSOe-SVR) was developed. Datasets were established based on compression tests in the temperature range of 350-450◦C and strain rate range of 0:01−50s−1. Meanwhile, the datasets were corrected for deformation heating and unbalance. The maximum relative errors between the experimental and predicted flow stress with PSOe-SVR, Back propagation neural network (BPNN) and constitutive equation was compared and analyzed. The results show the lower the strain rate, the greater the predicting accuracy of testing samples using PSOe-SVR. Meanwhile, the PSOe-SVR model has the most accurate prediction ability to those of BPNN and constitutive equation. The sample dependence of PSOe-SVR is also lower.
Suggested Reviewers
N/A
Recommended Citation
Lou, Yan; Ke, ChangXing; and Li, Luoxing
(2013)
"Accurately Predicting High Temperature Flow Stress of AZ80 Magnesium Alloy with Particle Swarm Optimization-based Support Vector Regression,"
Applied Mathematics & Information Sciences: Vol. 07:
Iss.
3, Article 29.
Available at:
https://digitalcommons.aaru.edu.jo/amis/vol07/iss3/29