基于LSTM-PINN的MgH<sub>2</sub>吸放氢行为预测模型研究

雷佩霞; 王莹; 田国娟; 李欣怡; 吴乐; 王玉琪; 丁鑫; 郑岚

doi:10.11949/0438-1157.20260371

您当前的位置：

首页 >

文章列表页 >

基于LSTM-PINN的MgH₂吸放氢行为预测模型研究

智能过程工程 | 更新时间：2026-06-11

- 基于LSTM-PINN的MgH₂吸放氢行为预测模型研究
- Research on model of MgH₂ hydrogen absorption/desorption behavior based on LSTM-PINN
- 化工学报 2026年
- 作者机构：
  
  1.西北大学化工学院，陕西西安710127
  2.西安市绿色氢能制储用一体化技术重点实验室，陕西西安710127
- 作者简介：
  
  雷佩霞（2002—），女，硕士研究生，2833347672@qq.com
  吴乐（1990—），男，博士，教授，lewu@nwu.edu.cn
- 基金信息：
  
  国家自然科学基金面上项目(22378328);陕西省青年科技人才新星计划项目(2024ZC-KJXX-010)
- DOI：10.11949/0438-1157.20260371
  中图分类号： TQ 03.39
- 收稿：2026-03-23，
  
  修回：2026-06-11，
  
  录用：2026-06-11，
- 稿件说明：
移动端阅览
雷佩霞, 王莹, 田国娟, 等. 基于LSTM-PINN的MgH₂吸放氢行为预测模型研究[J/OL]. 化工学报, 2026. DOI: 10.11949/0438-1157.20260371.

LEI Peixia, WANG Ying, TIAN Guojuan, et al. Research on model of MgH₂ hydrogen absorption/desorption behavior based on LSTM-PINN[J/OL]. CIESC Journal, 2026. DOI: 10.11949/0438-1157.20260371.
雷佩霞, 王莹, 田国娟, 等. 基于LSTM-PINN的MgH₂吸放氢行为预测模型研究[J/OL]. 化工学报, 2026. DOI: 10.11949/0438-1157.20260371. DOI：

LEI Peixia, WANG Ying, TIAN Guojuan, et al. Research on model of MgH₂ hydrogen absorption/desorption behavior based on LSTM-PINN[J/OL]. CIESC Journal, 2026. DOI: 10.11949/0438-1157.20260371. DOI：

摘要

针对MgH

吸放氢过程热力学与动力学复杂、传统模型依赖假设及纯数据驱动模型物理一致性差的问题，本研究首次提出一种融合长短期记忆神经网络（LSTM）与物理信息神经网络（PINN）的混合预测模型。以LSTM提取时序特征，嵌入范特霍夫热力学方程与JDM动力学方程作为双物理约束，设计复合损失函数实现端到端的训练。通过参数灵敏度分析，确定LSTM网络神经元数64/32、学习率0.0010、范特霍夫权重0.10、JDM权重0.05为设定参数范围内综合最优配置。研究结果表明模型性能优异，吸氢过程的决定系数R

最高达0.9998，放氢过程最高达0.9997；与LSTM、LSTM-Vanthoff和LSTM-JDM等模型相比，预测精度较高且收敛时间较短；模型训练过程稳定收敛且泛化性好。LSTM-PINN模型解决了传统物理建模依赖假设的问题，实现时序精准与物理一致的预测目标，为MgH

吸放氢动态过程预测提供新思路。

Abstract

Aiming at the thermodynamic and kinetic complexity of the MgH

hydrogen absorption/desorption process

the reliance of traditional models on assumptions

and the poor physical consistency of purely data-driven models

this study proposes for the first time a hybrid prediction model that integrates Long Short-Term Memory (LSTM) networks with Physics-Informed Neural Networks (PINNs). The LSTM extracts temporal features

while the van’t Hoff thermodynamic equation and the JDM kinetic equation are embedded as dual physical constraints. A composite loss function is designed to enable end-to-end training. Through parameter sensitivity analysis

the optimal configuration within the defined parameter range is determined as follows: LSTM network neurons: 64/32

learning rate: 0.0010

van’t Hoff weight: 0.10

and JDM weight: 0.05. The results demonstrate excellent model performance

with the coefficient of determination (R

) reaching up to 0.9998 for the hydrogen absorption process and 0.9997 for the desorption pro

cess. Compared with models such as LSTM

LSTM-Vanthoff

and LSTM-JDM

the proposed model achieves higher prediction accuracy and shorter convergence time. The model training process is stable

convergent

and exhibits good generalizability. The LSTM-PINN model overcomes the reliance of traditional physical modeling on assumptions

achieving both temporal accuracy and physical consistency

thereby offering a new approach for predicting the dynamic behavior of MgH

hydrogen absorption/desorption.

关键词

Keywords

references

Liu Y F , Guo Y S , Jiang Y R , et al . Recent progress in thermodynamic and kinetics modification of magnesium hydride hydrogen storage materials [J ] . Materials Reports: Energy , 2024 , 4 ( 1 ): 100252 .

王法正 , 隋璘 , 熊伟丽 . 面向多采样率数据的TTPA-LSTM软测量建模 [J ] . 化工学报 , 2025 , 76 ( 4 ): 1635 - 1646 .

Wang F Z , Sui L , Xiong W L . TTPA-LSTM soft sensor modeling for multi-sampling rate data [J ] . CIESC Journal , 2025 , 76 ( 4 ): 1635 - 1646 .

李文华 , 叶洪涛 , 罗文广 , 等 . 基于MHSA-LSTM的软测量建模及其在化工过程中的应用 [J ] . 化工学报 , 2024 , 75 ( 12 ): 4654 - 4665 .

Li W H , Ye H T , Luo W G , et al . Soft sensor modeling based on MHSA-LSTM and its application in chemical process [J ] . CIESC Journal , 2024 , 75 ( 12 ): 4654 - 4665 .

Li Q , Lin X , Luo Q , et al . Kinetics of the hydrogen absorption and desorption processes of hydrogen storage alloys: a review [J ] . International Journal of Minerals, Metallurgy and Materials , 2022 , 29 ( 1 ): 32 - 48 .

吴选军 , 王超 , 曹子健 , 等 . 数据与物理信息混合驱动的固定床吸附穿透深度学习模型 [J ] . 化工学报 , 2023 , 74 ( 3 ): 1145 - 1160 .

Wu X J , Wang C , Cao Z J , et al . Deep learning model of fixed bed adsorption breakthrough curve hybrid-driven by data and physical information [J ] . CIESC Journal , 2023 , 74 ( 3 ): 1145 - 1160 .

Kitagawa Y , Tanabe K . Development of a kinetic model of hydrogen absorption and desorption in magnesium and analysis of the rate-determining step [J ] . Chemical Physics Letters , 2018 , 699 : 132 - 138 .

Ren L , Li Y H , Zhang N , et al . Nanostructuring of Mg-based hydrogen storage materials: recent advances for promoting key applications [J ] . Nano-Micro Letters , 2023 , 15 ( 1 ): 93 .

史柯柯 , 刘木子 , 赵强 , 等 . 镁基储氢材料的性能及研究进展 [J ] . 化工进展 , 2023 , 42 ( 9 ): 4731 - 4745 .

Shi K K , Liu M Z , Zhao Q , et al . Properties and research progress of magnesium based hydrogen storage materials [J ] . Chemical Industry and Engineering Progress , 2023 , 42 ( 9 ): 4731 - 4745 .

Li Q , Lu Y F , Luo Q , et al . Thermodynamics and kinetics of hydriding and dehydriding reactions in Mg-based hydrogen storage materials [J ] . Journal of Magnesium and Alloys , 2021 , 9 ( 6 ): 1922 - 1941 .

Li A J , Fang Q , Sun W , et al . Catalytic mechanism of hydrogen absorption and desorption reactions in Mg-based hydrogen storage materials [J ] . Journal of Alloys and Compounds , 2025 , 1038 : 182814 .

范夏雨 , 孙建辰 , 李可莹 , 等 . 机器学习驱动液态有机储氢技术的系统优化 [J ] . 化工学报 , 2025 , 76 ( 8 ): 3805 - 3821 .

Fan X Y , Sun J C , Li K Y , et al . Machine learning drives system optimization of liquid organic hydrogen storage technology [J ] . CIESC Journal , 2025 , 76 ( 8 ): 3805 - 3821 .

Wang J , Hu Z W , Yan X , et al . Gradient-boosted spatiotemporal neural network for simulating underground hydrogen storage in aquifers [J ] . Journal of Computational Physics , 2025 , 521 : 113557 .

Ahmadi M . Physics-informed neural modeling of interfacial tension in hydrogen-rich systems using attention-based learning [J ] . International Journal of Hydrogen Energy , 2025 , 163 : 150829 .

Vo Thanh H . Physics informed machine learning for prediction hydrogen solubility in aqueous solution to implication for underground hydrogen storage formations [J ] . Renewable Energy , 2026 , 256 : 124460 .

Zhou P P , Zhou Q W , Xiao X Z , et al . Machine learning in solid-state hydrogen storage materials: challenges and perspectives [J ] . Advanced Materials , 2025 , 37 ( 6 ): 2413430 .

Yang X L , Li W X , Zhang J Q , et al . Hydrogen storage performance of Mg/MgH 2 and its improvement measures: research progress and trends [J ] . Materials , 2023 , 16 ( 4 ): 1587 .

郑欣雨 , 任泽华 , 周利 , 等 . 基于晶体图卷积神经网络的晶格能回归模型 [J ] . 化工学报 , 2025 , 76 ( 3 ): 1084 - 1092, 封 4 .

Zheng X Y , Ren Z H , Zhou L , et al . Lattice energy regression model based on crystal graph convolutional neural networks [J ] . CIESC Journal , 2025 , 76 ( 3 ): 1084 - 1092, 封 4 .

Morrison O , Uteva E , Walker G S , et al . Long time scale molecular dynamics simulation of magnesium hydride dehydrogenation enabled by machine learning interatomic potentials [J ] . ACS Applied Energy Materials , 2025 , 8 ( 1 ): 492 - 502 .

Maior C B S , Silva S P , Lins I D , et al . Machine learning for hydrogen storage applications: an exploration of the current literature [J ] . International Journal of Hydrogen Energy , 2025 , 151 : 150219 .

Duan C W , et al . A phase-transition catalyst bidirectionally enhances the hydrogen absorption/desorption kinetics of Mg/MgH 2 [J ] . Energy & Fuels , 2025 , 39 ( 23 ): 11437 - 11454 .

Allal Z , Noura H N , Vernier F , et al . Predicting hydrogen storage capacity of metal hydrides using novel imputation techniques and tree-based machine learning models [J ] . Computational Materials Science , 2026 , 262 : 114390 .

Dong S Y , Wang Y Y , Li J Y , et al . Exploration and design of Mg alloys for hydrogen storage with supervised machine learning [J ] . International Journal of Hydrogen Energy , 2023 , 48 ( 97 ): 38412 - 38424 .

Meng Z , Qian Q C , Xu M Q , et al . PINN-FORM: a new physics-informed neural network for reliability analysis with partial differential equation [J ] . Computer Methods in Applied Mechanics and Engineering , 2023 , 414 : 116172 .

Beyazit N İ . Comparative study of hydrogen storage and metal hydride systems: future energy storage solutions [J ] . Processes , 2025 , 13 ( 5 ): 1506 .

陆至彬 , 瞿景辉 , 刘桦 , 等 . 基于物理信息神经网络的传热过程物理场代理模型的构建 [J ] . 化工学报 , 2021 , 72 ( 3 ): 1496 - 1503 .

Lu Z B , Qu J H , Liu H , et al . Surrogate modeling for physical fields of heat transfer processes based on physics-informed neural network [J ] . CIESC Journal , 2021 , 72 ( 3 ): 1496 - 1503 .

Jang S H , Zhang D , Tran H B , et al . Physically interpretable descriptors drive the materials design of metal hydrides for hydrogen storage [J ] . Chemical Science , 2025 , 16 ( 48 ): 23111 - 23120 .

Zarzycki K , Ławryńczuk M . Long short-term memory neural networks for modeling dynamical processes and predictive control: a hybrid physics-informed approach [J ] . Sensors , 2023 , 23 ( 21 ): 8898 .

Wahyuddin E P , Caraka R E , Kurniawan R , et al . Improved LSTM hyperparameters alongside sentiment walk-forward validation for time series prediction [J ] . Journal of Open Innovation: Technology, Market, and Complexity , 2025 , 11 ( 1 ): 100458 .

Wang Q L , Han J Q , Chen F , et al . Real-time risk prediction of chemical processes based on attention-based Bi-LSTM [J ] . Chinese Journal of Chemical Engineering , 2024 , 75 : 131 - 141 .

Shi Y Y , Ye S H , Zheng Y G . Rapid forecasting of hydrogen concentration based on a multilayer CNN-LSTM network [J ] . Measurement Science and Technology , 2023 , 34 ( 6 ): 065101 .

Zou Z L , Yan X , Yuan Y S , et al . Attention mechanism enhanced LSTM networks for latency prediction in deterministic MEC networks [J ] . Intelligent Systems with Applications , 2024 , 23 : 200425 .

Alam M S , Ali M R , Rahman A . Detection and localization of ripe tomato in greenhouse environment using Keras-based deep learning models [J ] . Journal of Agriculture and Food Research , 2025 , 23 : 102182 .

Nathasarma R , Roy B K . Physics-informed long-short-term memory neural network for parameters estimation of nonlinear systems [J ] . IEEE Transactions on Industry Applications , 2023 , 59 ( 5 ): 5376 - 5384 .

Shirzadi M , Li Z , Yoshioka T , et al . Novel physics-informed optimization framework for complex multi-physics problems: Implementation for a sweeping gas membrane distillation module [J ] . Chemical Engineering Journal , 2024 , 498 : 155141 .

Wang T H , Liu G R , Li E , et al . Adaptive deep physics-informed neural network with dual-nested activation for solving complex partial differential equations [J ] . Computer Methods in Applied Mechanics and Engineering , 2025 , 444 : 118125 .

Liu Y , Hu H Y , Han Y H , et al . Accelerated discovery of highly active enzyme nanohybrids with parallelized Bayesian optimization in hybrid space [J ] . Nature Communications , 2026 , 17 : 3634 .

Han T , Wang Z , Meng H X . End-to-end capacity estimation of Lithium-ion batteries with an enhanced long short-term memory network considering domain adaptation [J ] . Journal of Power Sources , 2022 , 520 : 230823 .

Harris C R , Millman K J , van der Walt S J , et al . Array programming with NumPy [J ] . Nature , 2020 , 585 ( 7825 ): 357 - 362 .

Tran M K , Panchal S , Chauhan V , et al . Python-based scikit-learn machine learning models for thermal and electrical performance prediction of high-capacity lithium-ion battery [J ] . International Journal of Energy Research , 2022 , 46 ( 2 ): 786 - 794 .

Bolandi H , Sreekumar G , Li X Y , et al . Physics informed neural network for dynamic stress prediction [J ] . Applied Intelligence , 2023 , 53 ( 22 ): 26313 - 26328 .

Tao Z , Xu K , Liu F J . LSTM-PINN: an hybrid method for prediction of steady-state electrohydrodynamic flow [J ] . Journal of Computational Physics , 2026 , 548 : 114586 .

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

基于机器学习算法的生物质热解产物产率预测研究

稀疏回归方法挖掘有噪声均匀冷却粗糙颗粒流系统控制方程

机器学习辅助快速筛选高性能CO2吸附分子筛

机器学习驱动的木质素碳量子点性能预测与逆向设计

机器学习驱动分子筛基CO吸附剂设计与优化

基于LSTM-PINN的MgH2吸放氢行为预测模型研究

Research on model of MgH2 hydrogen absorption/desorption behavior based on LSTM-PINN

DOI：10.11949/0438-1157.20260371

摘要

Abstract

关键词

Keywords

references

基于LSTM-PINN的MgH₂吸放氢行为预测模型研究

Research on model of MgH₂ hydrogen absorption/desorption behavior based on LSTM-PINN