机器学习驱动聚氨酯构效关系研究进展

许书铭; 史有强; 王予喆; 赵现伟; 李卫平

doi:10.7527/S1000-6893.2026.34327

航空学报 >

0 1 - 0

DOI: https://doi.org/10.7527/S1000-6893.2026.34327

机器学习驱动聚氨酯构效关系研究进展

许书铭 ,
史有强 ,
王予喆 ,
赵现伟 ,
李卫平

展开

1. 北京航空航天大学材料科学与工程学院
2. 中国航发北京航空材料研究院

收稿日期: 2026-07-31

修回日期: 2026-09-10

网络出版日期: 2026-09-17

基金资助

国家重点研发计划项目;国家自然科学基金项目;北京市自然科学基金项目

收起

Advances in Machine Learning-Driven Structure-Property Relationships of Polyurethane

XU Shu-Ming ,
SHI You-Qiang ,
WANG Yu-Zhe ,
ZHAO Xian-Wei ,
LI Wei-Ping

Expand

Received date: 2026-07-31

Revised date: 2026-09-10

Online published: 2026-09-17

Fold

摘要

航空航天器在服役过程中所承受的高低温交替、强紫外辐射、原子氧侵蚀以及微流星体撞击等极端环境，对聚氨酯（PU）防护材料的耐温、耐辐射与自修复等性能提出了更高的要求。由于聚氨酯性能可设计性强，软硬段组成与宏观性能间呈高度非线性关联，传统的试错法研究周期长、成本高，且难以从分子层面解析性能演化机理。近年来，机器学习（ML）凭借强大的数据特征提取与建模能力，正在改变材料研发的既有模式。本文系统综述了机器学习在聚氨酯软硬段构效关系及其在航空航天涂层、隔热、结构功能一体化等领域的研究进展，概述了聚氨酯软硬段的化学特征与机器学习预测模型的基本逻辑，系统总结了机器学习对聚氨酯热行为、应力-应变行为、表面性能以及其他性能的预测和机理解释的代表性工作，讨论了机器学习在聚氨酯航空航天特色性能预测领域的现状与空白。在此基础上，进一步介绍机器学习辅助聚氨酯基涂料、泡沫及纤维等材料设计的实用化进展，重点涉及泡沫材料的微观结构识别、阻燃性能优化、热导率预测和主动学习逆设计，并探讨了其在航空航天轻质隔热领域的应用潜力，最后指出当前研究存在的不足与发展方向。

关键词： 聚氨酯; 机器学习; 构效关系; 智能设计; 性能预测

本文引用格式

许书铭 , 史有强 , 王予喆 , 赵现伟 , 李卫平 . 机器学习驱动聚氨酯构效关系研究进展[J]. 航空学报, 0 : 1 -0 . DOI: 10.7527/S1000-6893.2026.34327

Abstract

During the service process of aerospace vehicles, they are subjected to extreme environments such as alternating high and low temperatures, strong ultraviolet radiation, atomic oxygen erosion, and micrometeoroid impacts, which impose higher requirements on the temperature resistance, radiation resistance, and self-healing properties of polyurethane (PU) protective materials. Due to the strong designability of PU properties and the highly nonlinear correlation between soft and hard segment composition and macroscopic properties, traditional trial-and-error research methods are time-consuming and costly, and it is difficult to analyze the performance evolution mechanism at the molecular level. In recent years, machine learning (ML) is changing the existing paradigm of material research and development with its powerful data feature extraction and modeling capabilities. This paper systematically reviews the research progress of ML in the structure-property relationship of PU soft and hard segments, as well as its application in aerospace coatings, thermal insulation, structural and functional integration, etc. It outlines the chemical characteristics of PU soft and hard segments and the basic logic of ML prediction models, and systematically summarizes representative works on the prediction and mechanistic interpretation of PU thermal behavior, stress-strain behavior, surface properties, and other properties by ML. Additionally, it discusses the current status and gaps of machine learning in predicting the characteristic properties of polyurethane for aerospace applications. Based on this, it further introduces the practical progress of ML-assisted material design for PU-based coatings, foams, and fibers, focusing on the microstructure identification of foam materials, flame retardant performance optimization, thermal conductivity prediction, and active learning inverse design. It also discusses the potential applications in the field of lightweight thermal insulation for aerospace, and finally points out the current research deficiencies and development directions.

Key words： polyurethane; machine learning; structure-property relationship; intelligent design; performance prediction

参考文献

[1] 魏堃, 汪赛琳, 包诗曼, 等. 单组分聚氨酯胶黏剂研究进展 [J]. 材料工程, 2026: 1-14.
WEI K, WANG S L, BAO S M, et al. Research progress of one-component polyurethane adhesives [J]. Journal of Materials Engineering, 2026: 1-14.
[2] WU S, MA S, ZHANG Q, et al. A comprehensive review of polyurethane: Properties, applications and future perspectives [J]. Polymer, 2025, 327: 128361.
[3] SHIN E J, PRASAD C, CHOI H Y. Recent advances in thermoplastic polyurethane-based composites, properties, synthesis and its applications [J]. Journal of Industrial and Engineering Chemistry, 2025, 156: 150-191.
[4] YANG C, WANG G, ZHANG A, et al. High-elastic and strong hexamethylene diisocyanate (HDI)-based thermoplastic polyurethane foams derived by microcellular foaming with co-blowing agents [J]. Journal of CO2 Utilization, 2023, 74: 102543.
[5] JUNG S H, CHOI S, PARK J, et al. Strategic design of degradable polyurethanes: Site-specific depolymerization, film-forming behavior, and reprocessability [J]. Chemical Engineering Journal, 2024, 495: 153289.
[6] XU J, CHEN J, ZHANG Y, et al. A Fast Room‐Temperature Self‐Healing Glassy Polyurethane [J]. Angewandte Chemie International Edition, 2021, 60(14): 7947-7955.
[7] LIBLIKAS I, BONJOUR O, SAVEST N, et al. Bis-spirocyclic diol monomers and polyurethanes derived from citric acid: Synthesis, properties, electrospinnability, and evaluation of chemical recyclability [J]. Chemical Engineering Journal, 2025, 515(163525).
[8] BAO C, ZHANG X, YU P, et al. Facile fabrication of degradable polyurethane thermosets with high mechanical strength and toughness via the cross-linking of triple boron–urethane bonds [J]. Journal of Materials Chemistry A, 2021, 9(39): 22410-22417.
[9] HUANG R, ZHAO F, LI Y, et al. GO-enhanced polyurethane phase change materials with triple dynamic bonds: Oxime-urethane, coordination and hydrogen bonds [J]. Chemical Engineering Journal, 2025, 522: 168010.
[10] TRINH B M, GUPTA A, OWEN P, et al. Compostable lignin grafted poly(ε-caprolactone) polyurethane biomedical materials: Shape memory, foaming capabilities, and biocompatibility [J]. Chemical Engineering Journal, 2024, 485: 149845.
[11] WU H, PAN W, HE X, et al. Enhanced mechanical properties and tunable degradation of shape memory polyurethanes through oligo(glycolic acid) chain extender design [J]. Chemical Engineering Journal, 2025, 520: 166222.
[12] SAITO K, SCHARA P, EISENREICH F, et al. One-Pot Upcycling of Polycarbonate into BPA-Poly(ether-carbonate) Polyols for Polyurethane Applications [J]. ACS Sustainable Chemistry & Engineering, 2025, 13(9): 3766–3773.
[13] WANG X, MENG F, QIN L, et al. Beyond Mechanical Limitations: A Millable PDMS-Polyurethane Elastomer Integrating Superior Strength, Wide-Temperature Tolerance, and Multifunctional Properties [J]. ACS applied materials & interfaces, 2026, 18(11): 17002–17013.
[14] SHEN Q, WANG Z, XU H, et al. Preparation and characterization of halogen-free flame retardants waterborne polyurethane co-modified with soft and hard segments [J]. Progress in Organic Coatings, 2025, 204: 109251.
[15] LI Z, MA X, GENG Y, et al. Colorless and transparent self-healing polyurethane urea with superior tensile strength for protective coating [J]. European Polymer Journal, 2025, 228: 113827.
[16] CHENG C, LI J, YANG F, et al. Renewable eugenol-based functional polymers with self-healing and high temperature resistance properties [J]. Journal of Polymer Research, 2018, 25(2).
[17] HA Z, LEI L, ZHOU M, et al. Bio-Based Waterborne Polyurethane Coatings with High Transparency, Antismudge and Anticorrosive Properties [J]. ACS applied materials & interfaces, 2023, 15(5): 7427–7441.
[18] TIAN C, DAI T, TANG L, et al. Enhancing fatigue resistance of polyether-based polyurethanes via targeted modulation of microphase-separated structure [J]. International Journal of Fatigue, 2026, 211: 109724.
[19] MEHRYAR MOHRI, AFSHIN ROSTAMIZADEH, TALWALKAR A. Foundations of Machine Learning [M]. Cambridge, MA: MIT Press, 2018.
[20] PUGAR J A, GANG C, HUANG C, et al. Predicting Young’s Modulus of Linear Polyurethane and Polyurethane–Polyurea Elastomers: Bridging Length Scales with Physicochemical Modeling and Machine Learning [J]. ACS Applied Materials & Interfaces, 2022, 14(14): 16568-16581.
[21] ACARU S F, COMí M, FALIREAS P, et al. Glass transition temperature prediction in lignin polyurethanes using machine learning on small experimental dataset [J]. Materials & Design, 2026, 267: 116265.
[22] PUGAR J A, CHILDS C M, HUANG C, et al. Elucidating the Physicochemical Basis of the Glass Transition Temperature in Linear Polyurethane Elastomers with Machine Learning [J]. The Journal of Physical Chemistry B, 2020, 124(43): 9722–9733.
[23] QIN Y, MA Z, LI X, et al. Machine Learning-Driven Prediction and Interpretation of Glass Transition Temperature in Polyurethanes [J]. ACS Applied Polymer Materials, 2026, 8(8): 5471-5484.
[24] LI C, ZHAO H, ZHANG W, et al. Correlation‐Driven Feature Selection of RDKit Descriptors and Molecular Fingerprints for Predicting Polyurethane Glass Transition Temperature [J]. Journal of Polymer Science, 2026, 64(12): 2597-2609.
[25] 丁芳. 基于机器学习的聚氨酯弹性体结构性能关系研究 [D]. 合肥: 中国科学技术大学, 2022.
DING F. Study on structure-property relationship of polyurethane elastomer based on machine learning [D]. Hefei: University of Science and Technology of China, 2022.
[26] LI R, LV Y, XIE C, et al. Explore Thermal and Mechanical Properties of Biobased Polyurethane Elastomers Through Machine Learning Models [J]. Macromolecular Rapid Communications, 2026, 47(11): e00963.
[27] ORNAGHI H L, NOHALES A, ASENSIO M, et al. Effect of chain extenders on the thermal and thermodegradation behavior of carbonatodiol thermoplastic polyurethane [J]. Polymer Bulletin, 2023, 81(3): 2267-2286.
[28] DE L. FERREIRA B D, ARAúJO N R S, LIGóRIO R F, et al. Comparative Kinetic Study of Automotive Polyurethane Degradation in non-isothermal and isothermal conditions using Artificial Neural Network [J]. Thermochimica Acta, 2018, 666: 116-123.
[29] MURAVYEV N V, LUCIANO G, ORNAGHI H L, et al. Artificial Neural Networks for Pyrolysis, Thermal Analysis, and Thermokinetic Studies: The Status Quo [J]. Molecules, 2021, 26(12): 3727.
[30] 滕鑫, 唐颂超, 徐世爱, 等. 人工神经网络在聚氨酯配方设计中的应用研究 [J]. 实验科学与技术, 2015, 13(01): 38-43.
TENG X, TANG S C, XU S A, et al. Application of artificial neural network in polyurethane formula design [J]. Experiment Science and Technology, 2015, 13(1): 38-43.
[31] DING F, LIU L-Y, LIU T-L, et al. Predicting the Mechanical Properties of Polyurethane Elastomers Using Machine Learning [J]. Chinese Journal of Polymer Science, 2022, 41(3): 422-431.
[32] MENON A, THOMPSON-COLóN J A, WASHBURN N R. Hierarchical Machine Learning Model for Mechanical Property Predictions of Polyurethane Elastomers From Small Datasets [J]. Frontiers in Materials, 2019, 6.
[33] MENG Y, LIN Y, ZHANG A. Prediction and Explanation of Properties in Multicomponent Polyurethane Elastomers: Integrating Molecular Dynamics and Machine Learning [J]. Macromolecules, 2024, 57(23): 10912–10925.
[34] TALAPATRA A, DATTA D. A molecular dynamics-based investigation on tribological properties of functionalized graphene reinforced thermoplastic polyurethane nanocomposites [J]. Proceedings of the Institution of Mechanical Engineers, Part J: Journal of Engineering Tribology, 2020, 235(1): 61-78.
[35] ZHOU L, WANG M-F, HUANG C-K, et al. Prediction of Stress-strain Behavior for Polyurethane Elastomers Based on Machine Learning [J]. Chinese Journal of Polymer Science, 2026, 44(5): 1562-1573.
[36] RAPP J L, ANSTINE D M, GUSEV F, et al. Design of Tough 3D Printable Elastomers with Human-in-the-Loop Reinforcement Learning [J]. Angewandte Chemie International Edition, 2025, 64(36): e202513147.
[37] LIU L, LI R, XIE C, et al. A big data approach to explore core properties of waterborne polyurethane coatings [J]. Progress in Organic Coatings, 2025, 211: 109739.
[38] DALL AGNOL L, ORNAGHI H L, MONTICELI F, et al. Polyurethanes synthetized with polyols of distinct molar masses: Use of the artificial neural network for prediction of degree of polymerization [J]. Polymer Engineering & Science, 2021, 61(6): 1810-1818.
[39] PRASAD T, PATRA D K, KUNDU D. Molecular Descriptors-Based Analysis for Computation of Polyurethane Melting Temperature [J]. Multiscale Science and Engineering, 2026, 8: 48–58.
[40] LIANG K, QI X, XIAO X, et al. Chemically-informed active learning enables data-efficient multi-objective optimization of self-healing polyurethanes [J]. Chemical Science, 2026, 17(7): 3627-3638.
[41] HUANG C, WANG X. Study on the Healing Efficiency of Hydrogen-Bonded Self-Healing Polyurethane Based on Machine Learning Techniques [J]. Journal of Applied Polymer Science, 2026, 143(21): e70689.
[42] WANG L, HARRIS J, MAMOLO S, et al. A Theory-Guided Machine Learning and Molecular Dynamics Approach for Characterizing Fast-Curing Polyurethane Systems [J]. Polymers, 2026, 18(6): 679.
[43] CRUZ J A d, ORNAGHI H L, AMICO S C, et al. Predicting viscosity in polyurethane polymerization for liquid composite molding using neural networks and surface methodology [J]. Polymer Bulletin, 2023, 81(9): 8341-8358.
[44] HAGE I S, SEIF C Y, QUINSAAT J E Q, et al. Neural network-optimized imaging for classifying lignin-based polyurethane foams: Linking molecular composition to cellular microstructure using advanced machine learning [J]. Polymer, 2025, 324: 128235.
[45] SUN X, FU J, HAO M, et al. Machine learning-guided optimization of flame retardancy in rigid polyurethane foams [J]. Polymer Degradation and Stability, 2026, 244: 111838.
[46] CELIK BAYAR C. Development of a Digital Image Processing- and Machine Learning-Based Approach to Predict the Morphology and Thermal Properties of Polyurethane Foams [J]. Polymers, 2025, 17(7): 928.
[47] GOODARZI B V, BAHRAMIAN A R. Applying machine learning for predicting thermal conductivity coefficient of polymeric aerogels [J]. Journal of Thermal Analysis and Calorimetry, 2021, 147(11): 6227-6238.
[48] HOFFMANN M, PAI S M, ALBUQUERQUE R Q, et al. Active Learning-Driven Inverse Design of Polyurethane Foams for EV Battery Applications [J]. Journal of Polymer Science, 2025, 63(21): 4621-4630.
[49] MA K. Integrated hybrid modeling and SHAP (SHapley Additive exPlanations) to predict and explain the adsorption properties of thermoplastic polyurethane (TPU) porous materials [J]. Rsc Advances, 2024, 14(15): 10348-10357.
[50] FIROOZI S, AMANI A, DERAKHSHAN M A, et al. Artificial Neural Networks Modeling of Electrospun Polyurethane Nanofibers from Chloroform/Methanol Solution [J]. Journal of Nano Research, 2016, 41: 18-30.
[51] RABBI A, NASOURI K, BAHRAMBEYGI H, et al. RSM and ANN approaches for modeling and optimizing of electrospun polyurethane nanofibers morphology [J]. Fibers and Polymers, 2012, 13(8): 1007-1014.
[52] RAHMAN S M, TAFRESHI H V, POURDEYHIMI B. Physics-based deep neural network model to guide electrospinning polyurethane fibers [J]. Journal of Applied Polymer Science, 2022, 139(45): e53108.
[53] SOHRABI M, RAZBIN M. Hybrid modeling for optimizing electrospun polyurethane nanofibrous membranes in air filtration applications [J]. Scientific Reports, 2025, 15: 27306.

Options

摘要页面

文章导航

模态框（Modal）标题

摘要

本文引用格式

Abstract

参考文献