Please wait a minute...
金属学报  2026, Vol. 62 Issue (9): 1615-1626    DOI: 10.11900/0412.1961.2024.00332
  研究论文 本期目录 | 过刊浏览 |
基于机器学习的生物可降解锌合金力学性能预测
郭传平1, 石尘尘1, 刘鹏1, 高冬芳2, 赵洋洋3, 乔阳1()
1 济南大学 机械工程学院 济南 250022
2 山东大学齐鲁第二医院 基础医学研究所 济南 250033
3 山东大学第二医院 创伤骨科 济南 250031
Prediction of Mechanical Properties of Biodegradable Zinc Alloys Based on Machine Learning
GUO Chuanping1, SHI Chenchen1, LIU Peng1, GAO Dongfang2, ZHAO Yangyang3, QIAO Yang1()
1 School of Mechanical Engineering, University of Jinan, Jinan 250022, China
2 Institute of Medical Sciences, The Second Qilu Hospital of Shandong University, Jinan 250033, China
3 Trauma Orthopedics, Second Hospital of Shandong University, Jinan 250031, China
引用本文:

郭传平, 石尘尘, 刘鹏, 高冬芳, 赵洋洋, 乔阳. 基于机器学习的生物可降解锌合金力学性能预测[J]. 金属学报, 2026, 62(9): 1615-1626.
Chuanping GUO, Chenchen SHI, Peng LIU, Dongfang GAO, Yangyang ZHAO, Yang QIAO. Prediction of Mechanical Properties of Biodegradable Zinc Alloys Based on Machine Learning[J]. Acta Metall Sin, 2026, 62(9): 1615-1626.

全文: PDF(3456 KB)   HTML
摘要: 

为了明确满足可降解医用植入器械力学性能要求的锌合金的设计准则,本工作通过实验研究与文献搜集获取了Zn-Mg-Mn合金的力学性能数据,采用面向性能的机器学习(ML)模型,预测了Zn-Mg-Mn合金在不同元素种类与含量、不同合金制备工艺下的抗压屈服强度(CYS)与硬度,进而探究了元素种类与含量对材料微观结构与宏观力学性能的影响规律。基于现有数据集通过对比7种不同ML算法,确定了最小近邻(KNN)算法的预测能力最佳。为进一步验证模型预测的准确性,采用随机法选取数据集以外的数据与模型结果进行对比分析。同时,应用SHapley additive explanations (SHAP)方法定量研究了Zn-Mg-Mn合金中两种合金元素、制备工艺与CYS、硬度之间的相关性。结果表明,Mg元素对合金CYS和硬度的影响最大。添加Mg元素后,合金中析出Mg2Zn11和MgZn13,对力学性能产生显著影响。当Zn-Mg-Mn合金成分配比(质量分数)为Mg 2.25%~2.50%、Mn 2.50%~3.50%时,其CYS和硬度达到医用植入物标准。

关键词 : 机器学习,  锌合金,  可生物降解,  粉末冶金,  力学性能    
Abstract:

Recent studies indicate that zinc alloys are preferred in biodegradable metal materials owing to their unique biodegradability and biocompatibility. However, their mechanical properties are relatively insufficient; thus, it is crucial to design zinc alloys that meet the mechanical performance implantation standards required for application in biomedicine. The traditional alloy design method depends on experience and trial and error resulting in low efficiency and high cost. In recent years, the rapid development of artificial intelligence has provided new tools and methods for material science. Machine learning (ML), a subset of artificial intelligence, offers new ideas for material design and prediction. This study obtained the mechanical property data of the Zn-Mg-Mn alloy through experimental investigation and literature review. A performance-oriented ML model was used to predict the compressive yield strength (CYS) and hardness of the Zn-Mg-Mn alloy, considering various element types and contents and different alloy preparation processes. Furthermore, the influence of the element types and contents on the microstructure and macroscopic mechanical properties of the material was explored. Based on the existing dataset, a comparative evaluation of seven machine learning algorithms was conducted, and the k-nearest neighbors (KNN) algorithm exhibited the highest predictive performance. To further validate the accuracy of the model prediction, a random method was used to select data beyond the dataset for comparative analysis with the model results. Simultaneously, the Shapley additive explanation method was applied to quantitatively examine the correlation between the two alloying elements, the preparation process, the CYS, and the hardness in the Zn-Mg-Mn alloy. The Mg element was determined to significantly impact the alloy's CYS and hardness. Finally, the influence of individual elements on the mechanical properties of the materials was analyzed through microstructure characterization. The results showed that the formation of new phases (Mg2Zn11 and MgZn13) due to adding alloy elements significantly affected the mechanical properties. Based on the research results, this study proposed a composition ratio range for the Zn-Mg-Mn alloy to satisfy the mechanical performance standards required for medical implant materials. When Mg content is between 2.25% and 2.50% (mass fraction, the same below) and Mn content is between 2.50% and 3.50%, the CYS and hardness of the alloy comply with the implant standards.

Key words: machine learning    zinc alloy    biodegradable    powder metallurgy    mechanical property
收稿日期: 2024-09-27     
ZTFLH:  TG146.2  
基金资助:国家自然科学基金项目(52175408);山东省自然科学基金项目(ZR2023ME077);山东省自然科学基金项目(ZR2023MC140)
通讯作者: 乔 阳,me_qiaoy@ujn.edu.cn,主要从事生物医用材料的制备及高性能加工研究
Corresponding author: QIAO Yang, associate professor, Tel: 13791051675, E-mail: me_qiaoy@ujn.edu.cn
作者简介: 郭传平,男,2000年生,硕士
PowderShapeParticle size / μmPurity %Melting point / oC
ZnSphericity< 3099.9419.5
MgSphericity≤ 4099.9648.9
MnIrregular shape< 3099.91244.0
表1  用于制备Zn-Mg-Mn合金的粉末参数
图1  Zn、Mg和Mn混合粉末的DSC曲线及烧结工艺曲线
StatisticInputOutput
Mg %Mn %CYS MPaHardness HV
Minimum0.00.081.038.0
Maximum3.04.2325.0177.6
Average1.10.5217.287.3
Standard deviation0.90.748.425.3
表2  输入和输出变量范围
图2  使用平均值和标准差扩展数据流程图
图3  Zn-Mg-Mn合金抗压屈服强度预测值及相应实验数据
图4  Zn-Mg-Mn合金硬度的预测值及相应实验数据
图5  基于7种算法得到的决定系数(R2)、均方根误差(RMSE)及平均绝对误差(MAE)对比图
No.Processing techniqueMgMnPredicted valueExperimental valueRef.
%%

CYS

MPa

Hardness

HV

CYS

MPa

Hardness

HV

1PM1.100.852007920381This work
2PM2.751.02201120198120This work
3PM1.570.421908418984This work
4PM0.520.15221118216115This work
5PM0.220.521906319364This work
6Cast1.000.10-92-98[19]
7Cast1.500.10-154-149
8HE00.401365613658[26]
9Cast1.200-85-93[27]
表3  KNN算法验证集的预测值与实验值对比[19,26,27]
图6  KNN算法得到的抗压屈服强度、硬度预测值与实验值对比图
图7  特征相关性热图及统计图
图8  特征值对抗压屈服强度与硬度的综合相关性分析
图9  合金元素对抗压屈服强度和硬度影响的预测值
图10  不同Mg含量下Zn-3.5Mn-xMg合金的力学性能、XRD谱及OM和SEM像
图11  不同Mn含量下Zn-2.5Mg-xMn合金的力学性能、XRD谱和SEM像
[1] Li G N, Yang H T, Zheng Y F, et al. Challenges in the use of zinc and its alloys as biodegradable metals: Perspective from biomechanical compatibility [J]. Acta Biomater., 2019, 97: 23
doi: S1742-7061(19)30527-6 pmid: 31349057
[2] Yang D H, Chen J Q, Chen W P, et al. Fabrication of cellular Zn-Mg alloy foam by gas release reaction via powder metallurgical approach [J]. J. Mater. Sci. Technol., 2017, 33: 1141
doi: 10.1016/j.jmst.2017.03.019
[3] Tong X, Shi Z M, Xu L C, et al. Degradation behavior, cytotoxicity, hemolysis, and antibacterial properties of electro-deposited Zn-Cu metal foams as potential biodegradable bone implants [J]. Acta Biomater., 2020, 102: 481
doi: S1742-7061(19)30781-0 pmid: 31740321
[4] Sikora-Jasinska M, Mostaed E, Mostaed A, et al. Fabrication, mechanical properties and in vitro degradation behavior of newly developed Zn-Ag alloys for degradable implant applications [J]. Mater. Sci. Eng., 2017, C77: 1170
[5] Saris N E L, Mervaala E, Karppanen H, et al. Magnesium: An update on physiological, clinical and analytical aspects [J]. Clin. Chim. Acta, 2000, 294: 1
doi: 10.1016/s0009-8981(99)00258-2 pmid: 10727669
[6] Bagha P S, Khaleghpanah S, Sheibani S, et al. Characterization of nanostructured biodegradable Zn-Mn alloy synthesized by mechanical alloying [J]. J. Alloys Compd., 2018, 735: 1319
doi: 10.1016/j.jallcom.2017.11.155
[7] Shi Z Z, Li X M, Yao S L, et al. 300 MPa grade biodegradable high-strength ductile low-alloy (BHSDLA) Zn-Mn-Mg alloys: An in vitro study [J]. J. Mater. Sci. Technol., 2023, 138: 233
doi: 10.1016/j.jmst.2022.08.015
[8] Cho D H, Lee B W, Park J Y, et al. Effect of Mn addition on corrosion properties of biodegradable Mg-4Zn-0.5Ca-xMn alloys [J]. J. Alloys Compd., 2017, 695: 1166
doi: 10.1016/j.jallcom.2016.10.244
[9] Li M, Shi Z Z, Wang Q, et al. Zn-0.8 Mn alloy for degradable structural applications: Hot compression behaviors, four dynamic recrystallization mechanisms, and better elevated-temperature strength [J]. J. Mater. Sci. Technol., 2023, 137: 159
doi: 10.1016/j.jmst.2022.08.002
[10] Sun S N, Ren Y P, Wang L Q, et al. Abnormal effect of Mn addition on the mechanical properties of as-extruded Zn alloys [J]. Mater. Sci. Eng., 2017, A701: 129
[11] Li M, Xue Z, Shi Z Z, et al. High temperature strengthening mechanism of pure Zn with trace Mn addition [J]. J. Mater. Res. Technol., 2023, 25: 6764
doi: 10.1016/j.jmrt.2023.07.144
[12] Chang Y J, Jui C Y, Lee W J, et al. Prediction of the composition and hardness of high-entropy alloys by machine learning [J]. JOM, 2019, 71: 3433
doi: 10.1007/s11837-019-03704-4
[13] Suh J S, Suh B C, Bae J H, et al. Machine learning-based design of biodegradable Mg alloys for load-bearing implants [J]. Mater. Des., 2023, 225: 111442
doi: 10.1016/j.matdes.2022.111442
[14] Liu Y W, Wang L Y, Zhang H, et al. Accelerated development of high-strength magnesium alloys by machine learning [J]. Metall. Mater. Trans., 2021, 52A: 943
[15] Mazhnik E, Oganov A R. Application of machine learning methods for predicting new superhard materials [J]. J. Appl. Phys., 2020, 128: 075102
[16] Zhang S G. Pressureless sintered silicon carbide with enhanced mechanical properties obtained by a two-step sintering method [J]. Refract. Lime, 2015, 40(5): 48
[16] 张世国. 由两步烧结法获得的机械性能增强的无压烧结碳化硅 [J]. 耐火与石灰, 2015, 40(5): 48
[17] Xu H J, Li J, Shi R X. Preparation of ZrO2(3Y) nanoceramics under two-stage non-pressurized sintering [J]. China Powder Sci. Technol., 2006, 12(4): 11
[17] 徐海军, 李 嘉, 师瑞霞. 两段式无压烧结制备纳米二氧化锆(3Y)材料 [J]. 中国粉体技术, 2006, 12(4): 11
[18] Lóh N J, Simão L, Faller C A, et al. A review of two-step sintering for ceramics [J]. Ceram. Int., 2016, 42: 12556
doi: 10.1016/j.ceramint.2016.05.065
[19] Liu X W, Sun J K, Zhou F Y, et al. Micro-alloying with Mn in Zn-Mg alloy for future biodegradable metals application [J]. Mater. Des., 2016, 94: 95
doi: 10.1016/j.matdes.2015.12.128
[20] Ren T T, Gao X, Xu C, et al. Evaluation of as‐extruded ternary Zn-Mg-Zr alloys for biomedical implantation material: In vitro and in vivo behavior [J]. Mater. Corros., 2019, 70: 1056
[21] Liu X W, Sun J K, Qiu K J, et al. Effects of alloying elements (Ca and Sr) on microstructure, mechanical property and in vitro corrosion behavior of biodegradable Zn-1.5 Mg alloy [J]. J. Alloys Compd., 2016, 664: 444
doi: 10.1016/j.jallcom.2015.10.116
[22] Yang Y W, Yuan F L, Gao C D, et al. A combined strategy to enhance the properties of Zn by laser rapid solidification and laser alloying [J]. J. Mech. Behav. Biomed. Mater., 2018, 82: 51
doi: S1751-6161(18)30326-6 pmid: 29567530
[23] Yang H T, Jia B, Zhang Z C, et al. Alloying design of biodegradable zinc as promising bone implants for load-bearing applications [J]. Nat. Commun., 2020, 11: 401
doi: 10.1038/s41467-019-14153-7 pmid: 31964879
[24] Zhang K Q, Yin H Q, Jiang X, et al. A novel approach to predict green density by high-velocity compaction based on the materials informatics method [J]. Int. J. Miner. Metall. Mater., 2019, 26: 194
doi: 10.1007/s12613-019-1724-x
[25] Putcha C, Dutta S, Gupta S K. Reliability and Risk Analysis in Engineering and Medicine [M]. Cham: Springer, 2021: 1
[26] Jia B, Yang H T, Han Y, et al. In vitro and in vivo studies of Zn-Mn biodegradable metals designed for orthopedic applications [J]. Acta Biomater., 2020, 108: 358
doi: S1742-7061(20)30141-0 pmid: 32165194
[27] Shen C, Liu X W, Fan B, et al. Mechanical properties, in vitro degradation behavior, hemocompatibility and cytotoxicity evaluation of Zn-1.2Mg alloy for biodegradable implants [J]. RSC Adv., 2016, 6: 86410
doi: 10.1039/C6RA14300H
[28] Huang W J, Yin S C, Wang X J, et al. Grain growth and Hall-Petch relationship in Ti37V15Nb22Hf23W3 refractory high-entropy alloys [J]. J. Mater. Res., 2023, 38: 1719
doi: 10.1557/s43578-023-00924-y
[29] Hernández-Escobar D, Champagne S, Yilmazer H, et al. Current status and perspectives of zinc-based absorbable alloys for biomedical applications [J]. Acta Biomater., 2019, 97: 1
doi: S1742-7061(19)30523-9 pmid: 31351253
[30] Wan Y J, Zeng Y, Dou Y C, et al. Improved mechanical properties and strengthening mechanism with the altered precipitate orientation in magnesium alloys [J]. J. Magnes. Alloy., 2022, 10: 1256
doi: 10.1016/j.jma.2020.09.028
[31] Ren Y P, Lou D F, Zhang M D, et al. Fabrication, microstructures, and mechanical properties of Zn-0.1Mg-1Mn (wt.%) alloy tube [J]. J. Mater. Eng. Perform., 2023, 32: 793
doi: 10.1007/s11665-022-07122-7
[32] Wang L Q, Lou D F, Ren Y P, et al. Effect of Mn content on the microstructure and mechanical properties of as-extruded Zn-0.2Mg-Mn (wt%) alloys [J]. Mater. Charact., 2022, 194: 112400
doi: 10.1016/j.matchar.2022.112400
[33] Bowen P K, Drelich J, Goldman J. Zinc exhibits ideal physiological corrosion behavior for bioabsorbable stents [J]. Adv. Mater., 2013, 18: 2577
[34] Kong L Y, Heydari Z, Lami G H, et al. A comprehensive review of the current research status of biodegradable zinc alloys and composites for biomedical applications [J]. Materials, 2023, 16: 4797
doi: 10.3390/ma16134797
[1] 李帅雨, 刘泽栋, 杨劼人, 金伟, 彭博, 杜新, 周斯曼. TA18合金铸锭-管坯组织遗传行为及力学性能调控[J]. 金属学报, 2026, 62(9): 1487-1502.
[2] 刘泽田, 路伟昭, 徐新宇, 刘旭, 张少游, 王慧远. 均质化温度对3003铝合金微观组织和力学性能的影响规律[J]. 金属学报, 2026, 62(9): 1503-1516.
[3] 杨何, 侯自勇, 胡兴益, 程金骏, 王亚茹, 范国华, 黄晓旭. 超高强不锈钢微丝表征技术:进展与展望[J]. 金属学报, 2026, 62(7): 1189-1206.
[4] 徐青青, 闫震, 郭玉玉, 侯娟, 王皞, 黄爱军. 基于机器学习与多目标优化的304L不锈钢激光粉末床熔融工艺优化研究[J]. 金属学报, 2026, 62(7): 1228-1245.
[5] 王彬, 赵鹏, 刘家伟, 张春波, 秦志伟, 董红刚, 李鹏. 热处理对FGH96高温合金惯性摩擦焊接头微观组织演变及力学性能的影响[J]. 金属学报, 2026, 62(6): 1043-1058.
[6] 许鼎锋, 韩飞扬, 姜琦成, 王欢, 尚利媛, 卢一平. AlCrFeCoNi低膨胀合金的成分设计及组织和性能优化[J]. 金属学报, 2026, 62(6): 1021-1031.
[7] 吕昭平, 申耀祖, 王先珍, 陈强, 唐建国, 张晓宾, 于勇, 蒋虽合, 刘雄军, 王辉, 吴渊. 面向极端服役条件的高熵合金结构材料:性能优势、瓶颈与突破路径[J]. 金属学报, 2026, 62(5): 743-755.
[8] 陈星秋, 王建韬, 刘培涛. 矩张量机器学习势模型的发展与应用[J]. 金属学报, 2026, 62(5): 785-802.
[9] 康国政, 张旭, 胡冰晖, 双思垚, 于峻石, 熊宇凯, 宋世杰. 基于机器学习的金属材料多尺度塑性力学研究进展[J]. 金属学报, 2026, 62(5): 803-821.
[10] 郑玉峰, 谌雨农. 纯Mo及钼基可降解金属材料的研究现状与发展趋势[J]. 金属学报, 2026, 62(5): 905-922.
[11] 赵乃勤, 王振博, 戎旭东, 赵冬冬, 何春年. 原位反应粉末冶金铝基复合材料界面研究进展[J]. 金属学报, 2026, 62(5): 923-940.
[12] 陈光, 陈奉锐, 朱德民, 李贵忠, 李罗, 宋伟东, 王子特, 相恒高, 陈旸, 祁志祥. 聚片孪生功能基元序构的高性能TiAl合金[J]. 金属学报, 2026, 62(4): 541-549.
[13] 王建峰, 许珍木, 刘战, 高转妮, 占小红. 超声振动对TC4钛合金窄间隙激光焊接组织及强化机理的影响[J]. 金属学报, 2026, 62(3): 406-420.
[14] 蒋志达, 胥杨洋, 于佳新, 刘文才, 朱浩文, 吴国华, 尚郑平. 高温无氧轧制Cu/1060Al/Cu三层复合材料的力学和导电性能[J]. 金属学报, 2026, 62(3): 431-444.
[15] 江国龙, 周霞. 超声辅助下Mg/Mg-Zn-Al钎焊界面行为的分子动力学模拟[J]. 金属学报, 2026, 62(2): 372-382.