摘要
随着全球能源转型的加速,氢能因其高能量密度和清洁特性受到广泛关注。在储氢技术中,固态储氢以安全性高和体积能量密度大被视为最具潜力的途径。然而,固态储氢材料在储氢密度与操作温度2个重要指标面临着“鱼与熊掌”不可兼得的困局,严重制约了其实际应用。近年来,数据驱动技术在材料设计、性能预测和催化剂优化方面展现出巨大潜力,为新型储氢材料的开发提供了新思路。本文系统综述了数据驱动技术在固态储氢领域的研究进展,重点包括3方面:首先,高质量数据库的构建与应用为模型训练提供可靠支撑;其次,基于机器学习的合金正向与逆向设计实现了材料性能的高效预测与优化;最后,多智能体平台(如Cat-Advisor)通过多模态文献信息处理,推动镁基脱氢催化剂的智能筛选与优化。文章还讨论了催化剂微观结构表征不足、逆向设计能力有限及多源数据提取困难等挑战,并展望了通过AI、多模态智能体和数据库质量提升,推动固态储氢材料研发向系统化与智能化发展的前景。
With the acceleration of global energy transition,hydrogen energy has received widespread attention due to its high energy density and clean characteristics.Among hydrogen storage technologies,solid-state hydrogen storage is considered the most promising approach because of its high safety and large volumetric energy density.However,solid-state hydrogen storage materials face a dilemma between achieving high hydrogen density and maintaining suitable operating temperatures,a trade-off that severely limits their practical applications.In recent years,data-driven technologies have shown significant potential in material design,performance prediction,and catalyst optimization,providing new avenues for the development of novel hydrogen storage materials.This paper systematically reviews the research progress of data-driven technologies in the field of solid-state hydrogen storage,focusing on three key aspects:First,the construction and application of high-quality databases to provide reliable support for model training;second,forward and inverse design of alloys based on machine learning,achieving efficient prediction and optimization of material properties;and third,the use of multi-agent platforms such as Cat-Advisor for intelligent screening and optimization of magnesium-based dehydrogenation catalysts through multimodal processing of literature information.The article also discusses challenges such as inadequate characterization of catalyst microstructures,limited inverse design capabilities,and difficulties in extracting high-quality data from multiple sources.It envisions the prospects of advancing solid-state hydrogen storage material research and development towards systematization and intelligence through the integration of AI,multimodal intelligent agents,and improvements in database quality.
作者
杨维结
YANG Weijie(Department of Power Engineering,North China Electric Power University,Baoding 071003,Hebei,China)
出处
《金属功能材料》
2025年第5期100-108,共9页
Metallic Functional Materials
基金
河北省自然科学基金项目(E2023502006)
中央高校基本科研业务费专项资金项目(2025MS131)
中央高校基本科研业务费专项资金项目(2025JC008)。
关键词
固态储氢
数据驱动设计
储氢合金
脱氢催化剂
多智能体平台
solid-state hydrogen storage
data-driven design
hydrogen storage alloy
dehydrogenation catalyst
multi-agent platform
