航空科学技术2025，Vol.36Issue(6)：58-69,12.DOI:10.19452/j.issn1007-5453.2025.06.007
深度学习辅助的纳米薄膜材料压痕力学性能反演与预测
Deep Learning-Assisted Inverse Extraction and Prediction of Nanoindentation Mechanical Properties of Nanofilm Materials
摘要
Abstract
Accurate determination of the mechanical properties of nanometallic thin films is crucial for reliability evaluation.The advancement of artificial intelligence,or deep learning,has been utilized widely to strengthen and help to find more unknown solutions for the science and technology fields.This paper used nanoindentation testing and finite element inverse extraction to characterize the mechanical properties of nickel-coated multi-walled carbon nanotubes reinforced sintered nano-silver.Bayesian tuning algorithm was used for hyperparameters determination in ANN and CNN models in predicting yield stress,highly accurate prediction of the indentation mechanical properties of nanofilm materials has been successfully realized.The CNN model,despite with lower training efficiency,showed higher prediction accuracy and robustness with a coefficient of determination of 0.99.The ANN model is not that well performed,due to the lack of more input features.This could provide a versatile method for determining the mechanical properties of nanometallic thin films in the aviation industry and also give some hints on the applications of deep learning methods in predicting mechanical properties for other materials.关键词
卷积神经网络/纳米压痕/反演预测/纳米烧结银/包镍碳纳米管Key words
CNN/nanoindentation/inverse prediction/sintered nano-silver/nickel modified multiwall carbon nanotubes引用本文复制引用
于鹏举,隗嘉慧,代岩伟,秦飞..深度学习辅助的纳米薄膜材料压痕力学性能反演与预测[J].航空科学技术,2025,36(6):58-69,12.基金项目
国家自然科学基金(12272012) （12272012）
航空科学基金(2022Z057075001) National Natural Science Foundation of China(12272012) （2022Z057075001）
Aeronautical Science Foundation of China(2022Z057075001) （2022Z057075001）
