Scholarlink

<p>In order to study the hot deformation behavior of a novel metastable β titanium alloy Ti-3Al-5Mo-4Nb-4Cr-2Zr, hot compression experiments were in progress on the specimens at deformation temperatures of 700-850&#xa0;°C and strain rates of 0.001-1&#xa0;s<sup>-1</sup>. Two phenomenological constitutive models (the Arrhenius model and modified Johnson-Cook model), a physically-based constitutive model, and an optimized BP artificial neural network based on the firefly algorithm (FA-BP) were developed to predict the flow behavior of the alloy. After comparing the prediction accuracies of the above models, it was found that the FA-BP model outperformed the others due to its exceptional self-learning and adaptive capabilities. Specifically, the mean square correlation coefficient (<i>R</i><sup>2</sup>), root mean square error (RMSE), average absolute relative error (AARE), and mean absolute value error (MAE) of the FA-BP model are 0.9995, 1.6839 MPa, 1.074%, and 1.1240&#xa0;MPa, respectively. These metrics collectively demonstrate the high prediction accuracy of the FA-BP model. The hot processing map of the alloy was developed using the dynamic material model, and the optimal processing window was determined to be within a deformation temperature range of 820-850&#xa0;°C and a strain rate of 0.001-0.0025&#xa0;s<sup>-1</sup>. The results of EBSD and TEM analysis showed that the extent of dynamic recrystallization (DRX) increases with the increase of the power dissipation factor, and thus, the hot processing map has a high accuracy.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Investigating the Hot Deformation Behavior of Ti-3Al-5Mo-4Nb-4Cr-2Zr Alloy Utilizing Constitutive Models and the Firefly BP-ANN Model

  • 期刊论文

  • 发表日期：

    2025年9月8日

查看全文
  • Chuan Wang,
  • Haoyu Zhang,
  • Shuai Zhang,
  • Lijia Chen,
  • Shumin Wang,
  • Chao Li,
  • Jun Cheng,
  • Yingjie Wu

摘要

In order to study the hot deformation behavior of a novel metastable β titanium alloy Ti-3Al-5Mo-4Nb-4Cr-2Zr, hot compression experiments were in progress on the specimens at deformation temperatures of 700-850 °C and strain rates of 0.001-1 s-1. Two phenomenological constitutive models (the Arrhenius model and modified Johnson-Cook model), a physically-based constitutive model, and an optimized BP artificial neural network based on the firefly algorithm (FA-BP) were developed to predict the flow behavior of the alloy. After comparing the prediction accuracies of the above models, it was found that the FA-BP model outperformed the others due to its exceptional self-learning and adaptive capabilities. Specifically, the mean square correlation coefficient (R2), root mean square error (RMSE), average absolute relative error (AARE), and mean absolute value error (MAE) of the FA-BP model are 0.9995, 1.6839 MPa, 1.074%, and 1.1240 MPa, respectively. These metrics collectively demonstrate the high prediction accuracy of the FA-BP model. The hot processing map of the alloy was developed using the dynamic material model, and the optimal processing window was determined to be within a deformation temperature range of 820-850 °C and a strain rate of 0.001-0.0025 s-1. The results of EBSD and TEM analysis showed that the extent of dynamic recrystallization (DRX) increases with the increase of the power dissipation factor, and thus, the hot processing map has a high accuracy.