高模量玻璃纤维的研究进展

doi:10.16552/j.cnki.issn1001-1625.2025.1102

摘要/Abstract

摘要：

高模量玻璃纤维具有高强度、耐高温、耐腐蚀及优异的抗冲击和电绝缘等特性，是风电叶片、新能源汽车和航空航天等领域的核心增强材料。本文综述了高模量玻璃纤维的发展历程、典型产品及其性能特点，探讨了其微观结构特征与宏观模量之间的内在关联机制；阐述了高模量玻璃纤维组成的演变趋势，铝硅酸盐玻璃体系中避铝原理的适用条件，以及高配位铝、三簇氧等结构单元对弹性模量的影响；梳理了弹性模量的理论基础与多种计算模型，包括经典的Makishima-Mackenzie（M-M）模型、分子动力学模拟、拓扑约束理论及其修正方法，并引入机器学习探讨了其在成分设计与性能预测中的作用。

关键词: 玻璃纤维, 高模量, 分子动力学模拟, 机器学习, 高配位铝, 三簇氧

Abstract:

High-modulus glass fibers possess high strength， high-temperature resistance， corrosion resistance， excellent impact resistance， and electrical insulation. They serve as key reinforcement materials in wind turbine blades， new energy vehicles， aerospace， and other fields. This review summarizes the development history， typical products， and performance characteristics of high-modulus glass fibers， and discusses the intrinsic relationship between their microstructural features and macroscopic modulus. This paper expounds the evolution of chemical composition of high-modulus glass fibers， the applicable conditions of the aluminum avoidance principle in the aluminosilicate glass system， and the influence of structural units such as high-coordination aluminum and tri-cluster oxygen on elasticity modulus. The theoretical foundations and various computational models for predicting elasticity modulus are systematically outlined， including the classical Makishima-Mackenzie （M-M） model， molecular dynamics simulation， topological constraint theory and its modifications. The machine learning is introduced to explore its role in composition design and property prediction.

Key words: glass fiber, high-modulus, molecular dynamics simulation, machine learning, high-coordination aluminum, tri-cluster oxygen

中图分类号:

TQ171

魏耀澎, 伏建浩, 钱兆青, 何冉, 曹毅, 卢亚东, 马天琛, 康俊峰. 高模量玻璃纤维的研究进展[J]. 硅酸盐通报, 2026, 45(3): 961-973.

WEI Yaopeng, FU Jianhao, QIAN Zhaoqing, HE Ran, CAO Yi, LU Yadong, MA Tianchen, KANG Junfeng. Research Progress of High-Modulus Glass Fibers[J]. BULLETIN OF THE CHINESE CERAMIC SOCIETY, 2026, 45(3): 961-973.

图/表 14

表1 国内外企业高模量玻璃纤维产品及关键性能

Table 1 High-modulus glass fibers products and key properties from domestic and international enterprises

Enterprise	Main product/code name	Elasticity modulus/GPa	Main application field
AGY	S-3	~95	Aerospace， national defense and high-end sports equipment
Owens Corning	H³	95	Large wind turbine blades
China Jushi Co.， Ltd.	E9	>95	Wind turbine blades and new energy vehicles
Taishan Fiberglass Inc.	THM^®	≥95	Wind power， transportation and other fields

表2 E玻璃纤维和高模量玻璃纤维（HM）组成及模量

Table 2 Composition and modulus of E-glass fibers and high-modulus （HM） glass fibers

Glass fiber	Chemical constitution （mass fraction）/%								Modulus/GPa
Glass fiber	SiO₂	Al₂O₃	MgO	CaO	B₂O₃	M _x O _y	RE₂O₃	TiO₂	Modulus/GPa
E-glass fibers	52~62	12~16	0~5	16~25	0~10		0~2	0~1.5	80~85
First-generation HM	60~62	14~17	6~<8.75	14~17.5	Free	≤0.09（Na₂O）		0~1	86~89
Second-generation HM	56~68	11~≤20	10~16	<12	0~3	≤4（ZnO）	＞0.5	0~2	88~93
Third-generation HM	48~61	22~27	5~20	1~11		≤2.5（Li₂O）	≤5.5（Y₂O₃）		90~104

图1 在碱不足和碱过量情况下不同体系［AlO4］和［AlO5］的含量［10-12］

Fig.1 Content of ［AlO4］ and ［AlO5］ in different systems under conditions of insufficient alkali and excessive alkali［10-12］

图2 ［AlO5］和［O3］含量对弹性模量的影响［23-28］

Fig.2 Influence of ［AlO5］ and ［O3］ content on elasticity modulus［23-28］

表3 部分氧化物的单位体积解离能Gi 和离子堆积密度因子Ci［30］

Table 3 Unit volume dissociation energy Gi and ion bulk density factor Ciof some oxides［30］

Oxide	SiO₂	Al₂O₃	CaO	MgO	B₂O₃	Li₂O	ZrO₂	Y₂O₃	TiO₂	La₂O₃
C_i	0.541	0.832	0.596	0.731	0.731	0.539	0.846	0.720	0.779	0.539
G_i /（kJ·cm^-3）	64.7	134.4	65.1	84.0	78.1	80.6	97.4	74.3	86.9	68.0

图3 ［BO4］、［SiO4］和［AlO4］四面体的RUPF （a）、（b）、（c）和APF （d）、（e）、（f）的二维示意图［31］

Fig.3 Two-dimensional schematic diagrams of RUPF （a），（b），（c）， and APF （d），（e），（f） of ［BO4］，［SiO4］， and ［AlO4］ tetrahedra［31］

图4 修正前和修正后M-M模型与实验值的相关性［31］

Fig.4 Correlation between M-M model and experimental values before and after correction［31］

表4 部分原子的配位和单键强度［39］

Table 4 Coordination and single bond strength of some atoms［39］

Atom	Valence	Coordination number	SBS/（kcal•mol^-1）
B	3	3	119
B	3	4	89
Si	4	4	106
Al	3	4	101~79
Al	3	6	53~67
La	3	7	58
Y	3	8	50
Mg	2	6	37
Li	1	4	36
Na	1	6	20
Ca	2	8	32

图5 修正前和修正后Fnet与弹性模量之间的相关性［10，32，41-46］

Fig.5 Correlation between Fnet and elasticity modulus before and after correction［10，32，41-46］

图6 不同组分实验和模拟测得弹性模量值的对比［22，27，47-49］

Fig.6 Comparison of elasticity modulus values measured by experiments and simulations of different components［22，27，47-49］

表5 不同模型对比

Table 5 Comparison of different models

Model	Main input	Advantage	Limitation	Range of application	Application scenario
M-M model	$G i, V t$	Low computing cost and no need for a large amount of training data	High-coordination aluminum was not taken into consideration	Early rapid screening of the formula	Initial screening of the formula
Corrected M-M model	$η R U P F, G i, V t$	Better quantitative accuracy	The local structure unit cannot be directly parsed	More reliable modulus evaluation and comparison	Fine-tuning of the formula
MD+QSPR	$C N X O,$ $S B S X O$	Be capable of conducting feature importance analysis	Sensitive to potential functions and sample preparation paths	Mechanism analysis and medium-scale screening	Mechanism clarification
Stress strain method	Stress， strain	Verify other models	The quenching rate and loading method are very sensitive	Verification and mechanism research	Formula verification
TCT	$n 线$ ， $n 角$	Fast computing speed and strong interpretability	Accuracy depends on the quality of the structural input	Design rules and trend prediction	Adjustment and control of structure

表5 不同模型对比

Table 5 Comparison of different models

Model	Main input	Advantage	Limitation	Range of application	Application scenario
M-M model	$G i, V t$	Low computing cost and no need for a large amount of training data	High-coordination aluminum was not taken into consideration	Early rapid screening of the formula	Initial screening of the formula
Corrected M-M model	$η R U P F, G i, V t$	Better quantitative accuracy	The local structure unit cannot be directly parsed	More reliable modulus evaluation and comparison	Fine-tuning of the formula
MD+QSPR	$C N X O,$ $S B S X O$	Be capable of conducting feature importance analysis	Sensitive to potential functions and sample preparation paths	Mechanism analysis and medium-scale screening	Mechanism clarification
Stress strain method	Stress， strain	Verify other models	The quenching rate and loading method are very sensitive	Verification and mechanism research	Formula verification
TCT	$n 线$ ， $n 角$	Fast computing speed and strong interpretability	Accuracy depends on the quality of the structural input	Design rules and trend prediction	Adjustment and control of structure

图7 平均配位数和三簇氧含量对弹性模量的影响［64］

Fig.7 Influences of average coordination number and content of tri-cluster oxygen on elasticity modulus［64］

图8 模型训练过程

Fig.8 Model training process

图9 预测和解释玻璃性质的机器学习［70］

Fig.9 Machine learning for predicting and interpreting properties of glass［70］

参考文献 71

[1]	韩利雄，曾庆文. 高强度高模量玻璃纤维特性与应用［J］. 化工新型材料， 2011， 39（8）： 1-3.
	HAN L X， ZENG Q W. Characteristic and application of high-strength and high-modulus glass fibers［J］. New Chemical Materials， 2011， 39（8）： 1-3 （in Chinese）.
[2]	LI H， DEMIROK G， ATILGAN S， et al. High-modulus glass fiber for wind renewable energy generation： selective review on the recent research and development［J］. International Journal of Applied Glass Science， 2024， 15（4）： 367-380.
[3]	TANG X H， YOU J L， WANG J， et al. Quantitative studies on the microstructure of ternary CaO-Al₂O₃-SiO₂ glasses by Raman spectroscopy， ²⁷Al MAS NMR and quantum chemistry ab initio calculation［J］. Ceramics International， 2023， 49（22）： 34397-34408.
[4]	ZHANG S， HOU Y， GUO J， et al. Flowability and structural evolution of CaO-SiO₂-MgO-Al₂O₃ slag with varying MgO/CaO ratios： experiments and MD simulations［J］. Journal of Non-Crystalline Solids， 2024， 634： 122979.
[5]	PHILIPPS K， STOFFEL R P， DRONSKOWSKI R， et al. Experimental and theoretical investigation of the elastic moduli of silicate glasses and crystals［J］. Frontiers in Materials， 2017， 4： 2.
[6]	KALAHE J， STONE M P， DRAGIC P D， et al. The structures and properties of yttrium aluminosilicate glasses with low， medium， and high silica contents［J］. Journal of Non-Crystalline Solids， 2023， 614： 122394.
[7]	LOEWENSTEIN W. The distribution of aluminum in the tetrahedra of silicates and aluminates［J］.American Mineralogist， 1954， 39： 92-96
[8]	GIULI G， BONAZZI P， MENCHETTI S. Al-Fe disorder in synthetic epidotes； a single-crystal X-ray diffraction study［J］. American Mineralogist， 1999， 84（5/6）： 933-936.
[9]	ZHANG R， WANG Z， MENG Y F， et al. Quantitative insight into aluminum structures in CaO-Al₂O₃-SiO₂ system via Raman and ²⁷Al MAS-NMR spectroscopies［J］. Journal of Non-Crystalline Solids， 2021， 573： 121116.
[10]	XIANG Y， DU J C， SMEDSKJAER M M， et al. Structure and properties of sodium aluminosilicate glasses from molecular dynamics simulations［J］. The Journal of Chemical Physics， 2013， 139（4）： 044507.
[11]	ALLU A R， BALAJI S， TULYAGANOV D U， et al. Understanding the formation of CaAl₂Si₂O₈ in melilite-based glass-ceramics： combined diffraction and spectroscopic studies［J］. ACS Omega， 2017， 2（9）： 6233-6243.
[12]	ZHANG L L， WU W X， SUN C X， et al. Molecular dynamics simulations study on structure and elastic property of alkali-free aluminoborosilicate glasses with different Al₂O₃/（MgO+CaO） ratio［J］. Journal of Non-Crystalline Solids， 2023， 622： 122689.
[13]	WELCH R S， LEE K H， WILKINSON C J， et al. Topological hardening through oxygen triclusters in calcium aluminosilicate glasses［J］. Journal of the American Ceramic Society， 2021， 104（12）： 6183-6193.
[14]	CHEN Y， PAN W J， JIA B R， et al. Effects of the amphoteric behavior of Al₂O₃ on the structure and properties of CaO-SiO₂-Al₂O₃ melts by molecular dynamics［J］. Journal of Non-Crystalline Solids， 2021， 552： 120435.
[15]	GU G X， LIU X， ZHANG L L， et al. Mixed alkaline earth effect on the structure and elastic modulus of CaO-MgO-Al₂O₃-SiO₂ glasses： a molecular dynamics simulation［J］. Journal of Non-Crystalline Solids， 2023， 600： 122027.
[16]	LOGRADO M， YOUNGMAN R E， AITKEN B G， et al. Network modifier effects in multiple network former glasses： NMR results on the system 70 SiO₂-7.5 P₂O₅-（22.5-x）Al₂O₃-xNa₂O （0≤x≤17.5）［J］. The Journal of Physical Chemistry C， 2023， 127（34）： 17269-17284.
[17]	WU J S， GROSS T M， HUANG L P， et al. Composition and pressure effects on the structure， elastic properties and hardness of aluminoborosilicate glass［J］. Journal of Non-Crystalline Solids， 2020， 530： 119797.
[18]	WEBER R， SEN S， YOUNGMAN R E， et al. Structure of high alumina content Al₂O₃-SiO₂ composition glasses［J］. The Journal of Physical Chemistry B， 2008， 112（51）： 16726-16733.
[19]	BRADTMÜLLER H， UESBECK T， ECKERT H， et al. Structural origins of crack resistance on magnesium aluminoborosilicate glasses studied by solid-state NMR［J］. The Journal of Physical Chemistry C， 2019， 123（24）： 14941-14954.
[20]	LEE S K， STEBBINS J F. The structure of aluminosilicate glasses： high-resolution ¹⁷O and ²⁷Al MAS and 3QMAS NMR study［J］. The Journal of Physical Chemistry B， 2000， 104（17）： 4091-4100.
[21]	REDDY A A， GOEL A， TULYAGANOV D U， et al. Study of calcium-magnesium-aluminum-silicate （CMAS） glass and glass-ceramic sealant for solid oxide fuel cells［J］. Journal of Power Sources， 2013， 231： 203-212.
[22]	ZHANG L L， GAO W K， QU Y， et al. Molecular dynamics simulations study on structure and mechanical properties of Na₂O-CaO-Al₂O₃-B₂O₃-SiO₂ glasses with different Al₂O₃/B₂O₃ ratio［J］. Journal of Non-Crystalline Solids， 2024， 641： 123154.
[23]	ŘEHÁČKOVÁ L， NOVÁK V， TOKARSKÝ J， et al. Rheological behaviour of CaO-MgO-SiO₂-Al₂O₃-B₂O₃ system with varying B₂O₃ content up to 30 wt% at basicity of 0.4［J］. Ceramics International， 2024， 50（1）： 1389-1397.
[24]	PU Z D， HUANG J G， LI J F， et al. Effect of F content on the structure， viscosity and dielectric properties of SiO₂-Al₂O₃-B₂O₃-RO-TiO₂ glasses［J］. Journal of Non-Crystalline Solids， 2021， 563： 120817.
[25]	MANSOUR E. Semi-quantitative analysis for FTIR spectra of Al₂O₃-PbO-B₂O₃-SiO₂ glasses［J］. Journal of Non-Crystalline Solids， 2012， 358（3）： 454-460.
[26]	GUI H， LI C， LIN C W， et al. Glass forming， crystallization， and physical properties of MgO-Al₂O₃-SiO₂-B₂O₃ glass-ceramics modified by ZnO replacing MgO［J］. Journal of the European Ceramic Society， 2019， 39（4）： 1397-1410.
[27]	ZHANG L L， QIAN Z Q， FU J H， et al. Structure， elastic modulus， and thermal expansion of MgO-Al₂O₃-B₂O₃-SiO₂ glasses with Al₂O₃ substitution for B₂O₃ based on experiments and molecular dynamics simulations［J］. Ceramics International， 2025， 51（26）： 48944-48954.
[28]	SHIH Y T， SUNDARARAMAN S， ISPAS S， et al. New interaction potentials for alkaline earth silicate and borate glasses［J］. Journal of Non-Crystalline Solids， 2021， 565： 120853.
[29]	MAKISHIMA A， MACKENZIE J D. Direct calculation of Young’s moidulus of glass［J］. Journal of Non-Crystalline Solids， 1973， 12（1）： 35-45.
[30]	於秋霞，高捷敏. 高模量玻璃纤维现状及发展趋势［J］. 纤维复合材料， 2025， 42（2）： 59-63.
	YU Q X， GAO J M. Development status and future trend for high modulus glass fiber［J］. Fiber Composites， 2025， 42（2）： 59-63 （in Chinese）.
[31]	SHI Y， TANDIA A， DENG B H， et al. Revisiting the Makishima-Mackenzie model for predicting the Young’s modulus of oxide glasses［J］. Acta Materialia， 2020， 195： 252-262.
[32]	LU X N， DU J C. Quantitative structure-property relationship （QSPR） analysis of calcium aluminosilicate glasses based on molecular dynamics simulations［J］. Journal of Non-Crystalline Solids， 2020， 530： 119772.
[33]	MASSOBRIO C， DU J C， BERNASCONI M， et al. Molecular dynamics simulations of disordered materials： from network glasses to phase-change memory alloys［M］. Cham： Springer International Publishing， 2015.
[34]	DU J A， WILLIAMS C. Innovative projects between MNE subsidiaries and local partners in China： exploring locations and inter-organizational trust［J］. Journal of International Management， 2017， 23（1）： 16-31.
[35]	LU X N， DENG L， GIN S， et al. Quantitative structure-property relationship （QSPR） analysis of ZrO₂-containing soda-lime borosilicate glasses［J］. The Journal of Physical Chemistry B， 2019， 123（6）： 1412-1422.
[36]	BOUBATA N， ROULA A， MOUSSAOUI I. Thermodynamic and relative approach to compute glass-forming ability of oxides［J］. Bulletin of Materials Science， 2013， 36（3）： 457-460.
[37]	SCHUSTIK S A， CRAVERO F， PONZONI I， et al. Polymer informatics： expert-in-the-loop in QSPR modeling of refractive index［J］. Computational Materials Science， 2021， 194： 110460.
[38]	DU J C， LU X N， GIN S， et al. Predicting the dissolution rate of borosilicate glasses using QSPR analysis based on molecular dynamics simulations［J］. Journal of the American Ceramic Society， 2021， 104（9）： 4445-4458.
[39]	SUN K H. Fundamental condition of glass formation［J］. Journal of the American Ceramic Society， 1947， 30（9）： 277-281.
[40]	KIEU L H， DELAYE J M， CORMIER L， et al. Development of empirical potentials for sodium borosilicate glass systems［J］. Journal of Non-Crystalline Solids， 2011， 357（18）： 3313-3321.
[41]	LU X N， DENG L， DU J C. Effect of ZrO₂ on the structure and properties of soda-lime silicate glasses from molecular dynamics simulations［J］. Journal of Non-Crystalline Solids， 2018， 491： 141-150.
[42]	LUSVARDI G， MALAVASI G， TARSITANO F， et al. Quantitative structure-property relationships of potentially bioactive fluoro phospho-silicate glasses［J］. The Journal of Physical Chemistry B， 2009， 113（30）： 10331-10338.
[43]	DUCHOWICZ P R， FIORESSI S E， BACELO D E， et al. QSPR studies on refractive indices of structurally heterogeneous polymers［J］. Chemometrics and Intelligent Laboratory Systems， 2015， 140： 86-91.
[44]	KHAN P M， RASULEV B， ROY K. QSPR modeling of the refractive index for diverse polymers using 2D descriptors［J］. ACS Omega， 2018， 3（10）： 13374-13386.
[45]	XU J， CHEN B， ZHANG Q J， et al. Prediction of refractive indices of linear polymers by a four-descriptor QSPR model［J］. Polymer， 2004， 45（26）： 8651-8659.
[46]	KALAHE J， MAHADEVAN T S， LU X N， et al. Insights on the structure and properties of sodium iron phosphate glasses from molecular dynamics simulations［J］. Journal of Nuclear Materials， 2024， 594： 155004.
[47]	KATRITZKY A R， SILD S， LOBANOV V， et al. Quantitative structure-property relationship （QSPR） correlation of glass transition temperatures of high molecular weight polymers［J］. Journal of Chemical Information and Computer Sciences， 1998， 38（2）： 300-304.
[48]	BI Z S， LI K J， JIANG C H， et al. Performance and transition mechanism from acidity to basicity of amphoteric oxides （Al₂O₃ and B₂O₃） in SiO₂-CaO-Al₂O₃-B₂O₃ system： a molecular dynamics study［J］. Ceramics International， 2021， 47（9）： 12252-12260.
[49]	DENG L， DU J C. Development of effective empirical potentials for molecular dynamics simulations of the structures and properties of boroaluminosilicate glasses［J］. Journal of Non-Crystalline Solids， 2016， 453： 177-194.
[50]	BAUCHY M. Deciphering the atomic genome of glasses by topological constraint theory and molecular dynamics： a review［J］. Computational Materials Science， 2019， 159： 95-102.
[51]	LU X N， DENG L， DU J C， et al. Predicting boron coordination in multicomponent borate and borosilicate glasses using analytical models and machine learning［J］. Journal of Non-Crystalline Solids， 2021， 553： 120490.
[52]	BISHNOI S， SINGH S， RAVINDER R， et al. Predicting Young’s modulus of oxide glasses with sparse datasets using machine learning［J］. Journal of Non-Crystalline Solids， 2019， 524： 119643.
[53]	WILKINSON C J， ZHENG Q J， HUANG L P， et al. Topological constraint model for the elasticity of glass-forming systems［J］. Journal of Non-Crystalline Solids： X， 2019， 2： 100019.
[54]	LI X Z， WANG Y H， YANG P H， et al. Topological models of yttrium aluminosilicate glass based on molecular dynamics and structure characterization analysis［J］. Journal of the American Ceramic Society， 2025， 108（1）： e20118.
[55]	ZHENG Q J， ZENG H D. Progress in modeling of glass properties using topological constraint theory［J］. International Journal of Applied Glass Science， 2020， 11（3）： 432-441.
[56]	DENG B H. Machine learning on density and elastic property of oxide glasses driven by large dataset［J］. Journal of Non-Crystalline Solids， 2020， 529： 119768.
[57]	DENG L， DU J C. Development of boron oxide potentials for computer simulations of multicomponent oxide glasses［J］. Journal of the American Ceramic Society， 2019， 102（5）： 2482-2505.
[58]	SMITH J G， SIEGEL D J. Low-temperature paddlewheel effect in glassy solid electrolytes［J］. Nature Communications， 2020， 11： 1483.
[59]	BHATTOO R， BISHNOI S， ZAKI M， et al. Understanding the compositional control on electrical， mechanical， optical， and physical properties of inorganic glasses with interpretable machine learning［J］. Acta Materialia， 2023， 242： 118439.
[60]	CORMIER L， GHALEB D， NEUVILLE D R， et al. Chemical dependence of network topology of calcium aluminosilicate glasses： a computer simulation study［J］. Journal of Non-Crystalline Solids， 2003， 332（1/2/3）： 255-270.
[61]	HU Y J， ZHAO G， ZHANG M F， et al. Predicting densities and elastic moduli of SiO₂-based glasses by machine learning［J］. npj Computational Materials， 2020， 6： 25.
[62]	LIU X W， LONG Z L， PENG L. Prediction of Vickers hardness of amorphous alloys based on interpretable machine learning［J］. Journal of Non-Crystalline Solids， 2023， 602： 122095.
[63]	WANG M Y， ANOOP KRISHNAN N M， WANG B， et al. A new transferable interatomic potential for molecular dynamics simulations of borosilicate glasses［J］. Journal of Non-Crystalline Solids， 2018， 498： 294-304.
[64]	DU T， CHEN Z M， JOHANSEN S M， et al. Predicting stiffness and toughness of aluminosilicate glasses using an interpretable machine learning model［J］. Engineering Fracture Mechanics， 2025， 318： 110961.
[65]	SINGLA S， MANNAN S， ZAKI M， et al. Accelerated design of chalcogenide glasses through interpretable machine learning for composition-property relationships［J］. Journal of Physics： Materials， 2023， 6（2）： 024003.
[66]	ALCOBAÇA E， MASTELINI S M， BOTARI T， et al. Explainable machine learning algorithms for predicting glass transition temperatures［J］. Acta Materialia， 2020， 188： 92-100.
[67]	CASSAR D R， SANTOS G G， ZANOTTO E D. Designing optical glasses by machine learning coupled with a genetic algorithm［J］. Ceramics International， 2021， 47（8）： 10555-10564.
[68]	KANG Z Y， WANG L J， LI X Y， et al. Interpretable machine learning accelerates development of high-specific modulus glass［J］. Computational Materials Science， 2025， 246： 113482.
[69]	MANNAN S， ZAKI M， BISHNOI S， et al. Glass hardness： predicting composition and load effects via symbolic reasoning-informed machine learning［J］. Acta Materialia， 2023， 255： 119046.
[70]	ZAKI M， JAN A， ANOOP KRISHNAN N M， et al. Glassomics： an omics approach toward understanding glasses through modeling， simulations， and artificial intelligence［J］. MRS Bulletin， 2023， 48（10）： 1026-1039.
[71]	ZHU J Q， DING L F， SUN G H， et al. Accelerating design of glass substrates by machine learning using small-to-medium datasets［J］. Ceramics International， 2024， 50（2）： 3018-3025.