基于深度学习的多孔材料渗透率预测研究进展

Research progress of porous material permeability prediction based on deep learning

渗透率作为描述多孔材料流动性能的重要指标,在机械、能源等领域有着广泛的应用。经验关系式作为传统的多孔材料渗透率预测方法,往往通过实验统计法建立,在通用性、预测精度方面存在不足。近年来,基于深度学习构建渗透率预测模型的方法受到了众多学者的关注,在解决经验关系式存在的不足方面表现出很好的前景。为此,围绕基于深度学习的多孔材料渗透率预测建模方法,首先阐述了深度学习技术在结构模型重构中的应用及发展趋势,然后综述了结构参数-渗透率、图像-渗透率以及图像-流场-渗透率快速预测建模方法的基本原理和研究进展,最后展望了该领域的研究方向,以及对多孔材料制造系统性能提升方面的前景。

As an important parameter to describe the flow characteristics of porous materials,permeability is wildly used in various fields such as mechanics and energy.Empirical formulas is the traditional methods for predicting permeability in porous materials,and have some limitations in terms of generality and prediction accuracy.In recent years,there has been growing interest among researchers in utilizing deep learning to construct permeability prediction models,which shows promising prospects in addressing the shortcomings of empirical formulas.Focusing on the modeling approach for permeability prediction in porous materials based on deep learning,the application and development trends of deep learning techniques in porous material reconstruction were discussed.An overview of the fundamental principles and research progress was addressed for fast prediction modeling methods of permeability based on the relationships between permeability and structural parameters,permeability and images,as well as permeability and flow field.Finally,future research directions in this field and the potential for enhancing the performance of porous material manufacturing systems were discussed.