Scholarlink

<p>This study proposes an innovative method for predicting the mechanical properties and microstructure of dual-phase (DP) steel through deep learning (DL). Based on the dataset, an artificial neural network (ANN) model is a key component of DL, with a coefficient of determination of 0.93, indicating high accuracy between experimental and predicted values. At the same time, the content of Mn, C, Al, and Si, as well as the intercritical annealing (IA) conditions, shows higher feature importance scores, accounting for 18.54% and 47.99%, respectively. Furthermore, under experimental validation of the ANN predictions, the deviations between experimental and predicted values for ultimate tensile strength (UTS), total elongation (TE), and martensite volume fraction (FM) were within 1.5%, 1.8%, and 10%, respectively. These results demonstrate the reasonably high accuracy of the artificial neural network model. This paper highlights the potential of using the ANN model as a method for designing and optimizing the mechanical properties and microstructure of DP steel based on heat treatment. </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

Prediction of mechanical properties and microstructure of dual-phase steel based on deep learning method

  • 期刊论文

  • 发表日期：

    2025年7月20日

查看全文
  • Ya Xingwei,
  • Haijun Pan,
  • Wenyu Tao,
  • Yewei Tang,
  • Jinsheng Yuan,
  • Feifan Zang,
  • Lanlan Cai,
  • Bin Zhu

摘要

This study proposes an innovative method for predicting the mechanical properties and microstructure of dual-phase (DP) steel through deep learning (DL). Based on the dataset, an artificial neural network (ANN) model is a key component of DL, with a coefficient of determination of 0.93, indicating high accuracy between experimental and predicted values. At the same time, the content of Mn, C, Al, and Si, as well as the intercritical annealing (IA) conditions, shows higher feature importance scores, accounting for 18.54% and 47.99%, respectively. Furthermore, under experimental validation of the ANN predictions, the deviations between experimental and predicted values for ultimate tensile strength (UTS), total elongation (TE), and martensite volume fraction (FM) were within 1.5%, 1.8%, and 10%, respectively. These results demonstrate the reasonably high accuracy of the artificial neural network model. This paper highlights the potential of using the ANN model as a method for designing and optimizing the mechanical properties and microstructure of DP steel based on heat treatment.