• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文期刊>Journal of Materials Research >Accelerated design of multicomponent metallic glasses using machine learning
【24h】

Accelerated design of multicomponent metallic glasses using machine learning

机译：Accelerated design of multicomponent metallic glasses using machine learning

获取原文
获取原文并翻译 | 示例
           
页面导航
  • 摘要
  • 著录项
  • 相关主题

摘要

The present study examines the role of important elemental, thermodynamic, structural and kinetic attributes in amorphous phase formation and proposes a near fool-proof design strategy for multicomponent metallic glasses (MMGs) using machine learning (ML) approach. The feature space was optimized using feature engineering and incorporating the scientific fundamentals of glass formation as the 'veto' method. The incorporation of the characteristic transformation temperatures to the feature space allowed viewing the glass formation phenomenon from previously unexplored dimensions. A multilayer perceptron neural network (MLPNN) with error backpropagation was used to classify MMGs and crystalline multicomponent alloys (CMAs). The trained model performed reasonably well based on various scoring metrics with a cross-validation accuracy of 90.35%. Further, several new MMGs were designed, synthesized and examined for their glass-forming ability (GFA). The analysis showed good agreement between the experimental results and model predictions, validating the efficacy of machine learning approach in steering the development of MMGs in future.

著录项

获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号