基于机器学习的岩石动态断裂韧度预测方法

A machine learning-based prediction method for dynamic fracture toughness of rocks

摘要

岩石动态断裂韧度是评估岩石在动态载荷下抗裂纹扩展能力的关键参数，对深部资源开采、地质灾害防治等工程领域具有重要意义。传统实验方法面临成本高、操作复杂等问题，而机器学习技术为动态断裂韧度的预测提供了高效、低成本的解决方案。本研究基于210组切槽半圆盘弯曲岩石试样实验数据，选取密度、弹性模量、抗压强度、抗拉强度和加载率作为输入变量，采用反向传播（back propagation，BP）神经网络、随机森林（random forest，RF）、支持向量机（support vector machine，SVM）、极限学习机（extreme learning machine，ELM）、长短期记忆网络（long short-term memory network，LSTM）和卷积神经网络（convolutional neural network，CNN）6种机器学习模型，预测岩石的动态断裂韧度。通过决定系数（R²）、平均绝对误差（R_MAE）、平均偏差误差（R_MBE）和均方根误差（R_RMSE）评估模型性能。结果表明，ELM和BP模型表现最优，R²分别达到0.99和0.986 5，预测精度高且稳定性好。

Abstract

Rock dynamic fracture toughness is a key parameter for evaluating the ability of rocks to resist crack propagation under dynamic loads，holding significant importance for engineering fields such as deep resource extraction and geological hazard prevention. Traditional experimental methods face problems such as high costs and complex operations，while machine learning technology provides an efficient and low-cost solution for predicting dynamic fracture toughness. Based on the data of 210 groups of rock specimens for notched semi-circular bending （NSCB） experiments，this paper selects density，elastic modulus，compressive strength，tensile strength，and loading rate as input variables，and uses six machine learning models-back propagation （BP） neural network，random forest （RF），support vector machine （SVM），extreme learning machine （ELM），long short-term memory network （LSTM），and convolutional neural network （CNN）-to predict the dynamic fracture toughness of rocks. The model performance is evaluated by indicators such as coefficient of determination （R ²），mean absolute error （R_MAE），mean bias error （R_MBE），and root mean square error （R_RMSE）. The results show that the ELM and BP models perform the best，with R ² reaching 0.99 and 0.986 5，respectively，showing high prediction accuracy and good stability.

关键词

机器学习 / 岩石动态断裂韧度 / 加载率 / 神经网络 / 预测模型

Key words

machine learning / rock dynamic fracture toughness / loading rate / neural network / prediction model

参考文献

[1] 周科平, 吴家有, 周诗彤, 等. 裂缝长度和角度对红砂岩半圆试件断裂特性的影响[J]. 黄金科学技术, 2025， 33（3）: 528-539.

[2] ZHOU Keping， WU Jiayou， ZHOU Shitong， et al. Impact of crack length and angle on the fracture behavior of red sandstone in semi-circular bend（SCB）specimens[J]. Gold science and technology, 2025， 33（3）: 528-539（in Chinese）.

[3] 肖晓冬, 周磊, 梁贤, 等. 加热速率对中心直裂纹巴西圆盘试样断裂行为影响规律研究[J]. 矿业科学学报, 2024， 9（4）: 586-596.

[4] XIAO Xiaodong， ZHOU Lei， LIANG Xian， et al. The effect of heating rate on the fracture behavior of cracked straight-through Brazilian disc specimen[J]. Journal of mining science and technology, 2024， 9（4）: 586-596（in Chinese）.

[5] MARUVANCHERY V， KIM E. Effects of water on rock fracture properties:studies of mode Ⅰ fracture toughness，crack propagation velocity，and consumed energy in calcite-cemented sandstone[J]. Geomechanics and engineering, 2019， 17（1）: 57-67.

[6] 夏开文, 姚伟, 付岩, 等. 深部岩石动力学SHPB实验研究进展[J]. 工程爆破, 2024， 30（5）: 29-44.

[7] XIA Kaiwen， YAO Wei， FU Yan， et al. Advances on dynamic behaviors of deep rocks using SHPB[J]. Engineering blasting, 2024， 30（5）: 29-44（in Chinese）.

[8] DU K， LI X F， TAO M， et al. Experimental study on acoustic emission （AE） characteristics and crack classification during rock fracture in several basic lab tests[J]. International journal of rock mechanics and mining sciences, 2020， 133: 104411.

[9] BAHRAMI B， NEJATI M， AYATOLLAHI M R， et al. Theory and experiment on true mode Ⅱ fracturing of rocks[J]. Engineering fracture mechanics, 2020， 240: 107314.

[10] MENG Y Y， JING H W， ZHOU Z F， et al. Experimental investigation on the mixed-mode fracture behavior of rock-like material with bedding plane[J]. Theoretical and applied fracture mechanics, 2022， 117: 103159.

[11] 张盛, 李小军, 李大伟. 岩石Ⅰ型断裂韧度测试技术和理论研究综述[J]. 河南理工大学学报（自然科学版）, 2009， 28（1）: 33-38.

[12] ZHANG Sheng， LI Xiaojun， LI Dawei. A review of test technique and theory of mode Ⅰ rock fracture toughness[J]. Journal of Henan Polytechnic University（natural science）, 2009， 28（1）: 33-38（in Chinese）.

[13] FRANKLIN J A， SUN Z Q， ATKINSON B K， et al. Suggested methods for determining the fracture toughness of rock[J]. International journal of rock mechanics and mining sciences, 1988， 25（2）: 71-96.

[14] DAI F， XIA K， ZHENG H， et al. Determination of dynamic rock mode-I fracture parameters using cracked chevron notched semi-circular bend specimen[J]. Engineering fracture mechanics, 2011， 78（15）: 2633-2644.

[15] ZHOU Y X， XIA K， LI X B， et al. Suggested methods for determining the dynamic strength parameters and mode-I fracture toughness of rock materials[J]. International journal of rock mechanics and mining sciences, 2012， 49: 105-112.

[16] YAO W， XIA K W. Dynamic notched semi-circle bend （NSCB） method for measuring fracture properties of rocks:fundamentals and applications[J]. Journal of rock mechanics and geotechnical engineering, 2019， 11（5）: 1066-1093.

[17] PEI C H， LI X P， MA R Q， et al. Research on the dynamic fracture toughness of reef limestone[J]. Ocean engineering, 2022， 264: 112387.

[18] LIU Z D， LI D Y， ZHANG Z X， et al. An experimental study of fracture mechanism and morphology of granite specimens under various dynamic loading rates[J]. Theoretical and applied fracture mechanics, 2024， 134，Part A: 104687.

[19] SONG Y， YUE Z W. Experimental study on dynamic fracture behaviors of Beishan NSCB and CCNSCB granite specimens under different loading rates[J]. Soil dynamics and earthquake engineering, 2021， 141: 106512.

[20] XING H Z， XIE F， WANG M Y， et al. Experimental investigation of fracture process zone of rock in dynamic mode Ⅰ fracturing and its effect on dynamic crack initiation toughness[J]. Engineering fracture mechanics, 2022， 275: 108828.

[21] 满轲, 刘晓丽, 宋志飞, 等. 岩石静态与动态断裂韧性的宏细观试验[J]. 清华大学学报（自然科学版）, 2021， 61（8）: 799-808.

[22] MAN Ke， LIU Xiaoli， SONG Zhifei， et al. Macro-micro experimental study of rock static and dynamic fracture toughness[J]. Journal of Tsinghua University（science and technology）, 2021， 61（8）: 799-808（in Chinese）.

[23] WU Y， YIN T B， LIU X L， et al. Determination of dynamic mode Ⅰ fracture toughness of rock at ambient high temperatures using notched semi-circular bend method[J]. Transactions of Nonferrous Metals Society of China, 2022， 32（9）: 3036-3050.

[24] LIU C J， ZHANG Q J， ZHU Q， et al. Effect of cyclic freeze-thaw treatments on the dynamic fracture characteristics of granite:laboratory testing[J]. Engineering fracture mechanics, 2022， 276，Part B: 108915.

[25] HAN Z Y， LI D Y， LI X B. Effects of axial pre-force and loading rate on mode I fracture behavior of granite[J]. International journal of rock mechanics and mining sciences, 2022， 157: 105172.

[26] 陈进帆, 尚德磊, 赵志宏, 等. 基于机器学习的含充填裂隙岩样断裂行为预测方法[J]. 岩石力学与工程学报, 2023， 42（增刊1）: 3458-3472.

[27] CHEN Jinfan， SHANG Delei， ZHAO Zhihong， et al. Prediction method of fracture behaviors of rock samples with an infilled fracture based on machine learning[J]. Chinese journal of rock mechanics and engineering, 2023， 42（S1）: 3458-3472（in Chinese）.

[28] YANG Y Z， SHAO Z S， WU K， et al. Machine learning approaches for predicting rock mode Ⅰ fracture toughness:insights from ISRM suggested CCNBD and SCB tests[J]. Engineering fracture mechanics, 2025， 318: 110949.

[29] RUMELHART D E， HINTON G E， WILLIAMS R J. Learning representations by back-propagating errors[J]. Nature, 1986， 323（6088）: 533-536.

[30] BREIMAN L. Random forests[J]. Machine learning, 2001， 45（1）: 5-32.

[31] CORTES C， VAPNIK V. Support-vector networks[J]. Machine learning, 1995， 20（3）: 273-297.

[32] HUANG G B， ZHU Q Y， SIEW C K. Extreme learning machine:theory and applications[J]. Neurocomputing, 2006， 70（1/2/3）: 489-501.

[33] YUNITA A， PRATAMA M I， ALMUZAKKI M Z， et al. Performance analysis of neural network architectures for time series forecasting:a comparative study of RNN，LSTM，GRU，and hybrid models[J]. MethodsX, 2025， 15: 103462.

[34] 宋雨濛, 龚元丽, 任艳. 基于LSTM和CNN的对抗性跨站脚本攻击分析和检测方法研究[J]. 信息安全研究, 2025， 11（8）: 761-767.

[35] SONG Yumeng， GONG Yuanli， REN Yan. Research on analysis and detection methods of adversarial cross-site scripting attacks based on LSTM and CNN[J]. Journal of information security research, 2025， 11（8）: 761-767（in Chinese）.

[36] 余旭, 石克龙, 王宇, 等. 燃爆载荷冲击下五峰-龙马溪组页岩断裂韧性及破坏模式[J]. 煤炭学报, 2023， 48（12）: 4322-4335.

[37] YU Xu， SHI Kelong， WANG Yu， et al. Fracture toughness and failure mode of Wufeng-Longmaxi shale under explosion load impact[J]. Journal of China Coal Society, 2023， 48（12）: 4322-4335（in Chinese）.

[38] ZHOU Z L， CAI X， MA D， et al. Water saturation effects on dynamic fracture behavior of sandstone[J]. International journal of rock mechanics and mining sciences, 2019， 114: 46-61.

[39] CHEN R， XIA K， DAI F， et al. Determination of dynamic fracture parameters using a semi-circular bend technique in split Hopkinson pressure bar testing[J]. Engineering fracture mechanics, 2009， 76（9）: 1268-1276.

[40] YIN T B， LI X B， XIA K W， et al. Effect of thermal treatment on the dynamic fracture toughness of Laurentian granite[J]. Rock mechanics and rock engineering, 2012， 45（6）: 1087-1094.

[41] YAO W， XIA K W， ZHANG T. Dynamic fracture test of Laurentian granite subjected to hydrostatic pressure[J]. Experimental mechanics, 2019， 59（2）: 245-250.

[42] WANG Y B. Rock dynamic fracture characteristics based on NSCB impact method[J]. Shock and vibration, 2018， 2018（1）: 3105384.

[43] 方士正, 杨仁树, 李炜煜, 等. 基于NSCB方法的冻结红砂岩动态断裂特性试验[J]. 工程科学学报, 2023， 45（10）: 1704-1715.

[44] FANG Shizheng， YANG Renshu， LI Weiyu， et al. Investigation of dynamic fracture characteristics of frozen red sandstone using notched semi-circular bend method[J]. Chinese journal of engineering, 2023， 45（10）: 1704-1715（in Chinese）.

[45] ZHANG Q B， ZHAO J. Determination of mechanical properties and full-field strain measurements of rock material under dynamic loads[J]. International journal of rock mechanics and mining sciences, 2013， 60: 423-439.

[46] ZHANG Q B， ZHAO J. Quasi-static and dynamic fracture behaviour of rock materials: phenomena and mechanisms[J]. International journal of fracture, 2014， 189（ 1）: 1-32.

[47] CHEN R， LI K， XIA K W， et al. Dynamic fracture properties of rocks subjected to static pre-load using notched semi-circular bend method[J]. Rock mechanics and rock engineering, 2016， 49（10）: 3865-3872.

[48] BHAT H S， ROSAKIS A J， SAMMIS C G. A micromechanics based constitutive model for brittle failure at high strain rates[J]. Journal of applied mechanics, 2012， 79（3）: 031016.

[49] LI X Z， LUO Q L， YAN F Y， et al. Dynamic compressive micro-macro fracture mechanism with the water-saturated strengthening and weakening effect in brittle rocks[J]. Theoretical and applied fracture mechanics, 2025， 137: 104871.

[50] LI X Z， LUO Q L， QI C Z. Dynamic unsaturated effect of microcrack extension on brittle rock fracture under impact compression[J]. International journal of rock mechanics and mining sciences, 2025， 195: 106282.

基金资助

国家自然科学基金资助项目(52438007)

国家自然科学基金资助项目(51708016)

国家自然科学基金资助项目(12172036)