Machine-learning approach to predict work hardening behavior of pearlitic steel

作者：Qiao, Ling; Liu, Yong; Zhu, Jingchuan^*; Wang, Zibo

来源：Materials Letters, 2021, 289: 129384.

DOI：10.1016/j.matlet.2021.129384

• Metals and alloys
• Simulation and modelling
• GRNN model
• Pearlitic steel
• Alloying elements
• Work hardening behavior

Determining the mechanical properties with complex composition is a key issue in many applications. This study has examined the work hardening characteristics of the developed pearlitic steel. The generalized regression neural network (GRNN) optimized by four-folds cross validation technology was applied to predict the hardening exponent correlated with the alloying elements. The optimized GRNN model exhibited an excellent generalization and predictive performance when dealing with the small number of experimental data. As a result, the Pearson correlation coefficient analysis shows strong positive correlation of C and Si and negative correlation of Mn and V. This work has confirmed the practicability of GRNN approach to predict the hardening exponent for assisting industrial application design.

• 单位
  harbin inst technol