数据驱动的高温结构强度与寿命评估：进展与挑战

Data-Driven Structural Strength and Life Assessment of High Temperature Structure: Progresses and Challenges

关键词:  高温结构  数据驱动  图像处理  寿命预测  本构建模  微观组织  综述

Key words:  High-temperature structure  Data driving  Image process  Life prediction  Constitutive modelling  Microstructure  Review

摘要:

本文旨在分析数据驱动思维方法和研究范式在高温结构强度与寿命评估中的应用现状、发展趋势以及存在的问题和挑战。重点关注高温结构服役微观组织演化的图像数据处理和定量识别、多因素耦合作用下的材料/结构寿命预测和本构建模方法。总结了数据驱动方法典型的4种应用方式，即完全代替现有理论方法、耦合驱动已有方法的改进、挖掘变量之间潜在规律和定性认识定量化。最后，指出了数据驱动方法面临的诸如泛化能力不强、外推能力差以及与物理机制关联弱的问题，探讨了未来研究的潜在方向，以期能够推进数据驱动新范式在高温结构完整性领域的交叉应用。

Abstract:

The paper aims to analyze the current application status， development trends， and existing problems and challenges of data-driven methodologies and investigation paradigms in high-temperature structural strength and life assessment. In particular， it emphasizes on data processing and quantitative identification of images， the prediction of material/structure life under the coupling effect of multiple factors and the constitutive modeling method for microstructure evolution of high-temperature structures in service. It summarizes four typical application modes of data-driven methods， namely， the complete replacement of existing theoretical methods， the improvement of coupling-driven existing methods， the exploitation of potential regularities between variables and the quantification of qualitative understanding. Lastly， it points out the current shortcomings of data-driven methods， including weak generalization ability， poor extrapolation ability and weak correlation with physical mechanisms. Furthermore， the potential directions for future research are explored with a view of promoting the cross-application of data-driven new paradigms in the field of high-temperature structural integrity.