陕北侏罗纪煤田以煤层厚度大、埋藏浅、煤质优闻名于世，是我国重要的能源化工基地。然而，陕北地区生态环境脆弱，伴随煤炭资源地下大规模开采会导致覆岩变形破坏，产生的垮落带和导水裂隙带（简称“两带”），破坏地下水资源，威胁地面建筑安全，也使原本脆弱的生态环境遭到破坏。因此，准确预测“两带”发育高度，探索其发育规律和影响因素，进而开展突水模式研究，对保证煤矿安全生产、保护矿区生态环境具有十分重要意义。

本文基于陕北侏罗纪煤田典型煤矿实测数据，参照前人相关研究成果，按照埋深和基载比不同，将导水裂隙带高度分为三种类型（典型浅埋、近浅埋、中深埋）。采用多元非线性回归、机器学习等方法对采集样本进行了“两带”高度预测分析，得出不同煤层赋存条件下所对应的最佳“两带”高度预测方法。将预测分析结果与红柳林煤矿15207工作面实测“两带”高度进行了对比分析，进一步验证了本研究数值预测方法的可行性。在此研究基础上进行了突水模式研究，为煤矿地下水害防治提供了可靠依据。论文主要研究方法和取得成果如下：

（1）在获取并研究陕北侏罗纪煤层的186组导水裂隙带高度的基础上，将导水裂隙带高度划分为三种类型，即典型浅埋、近浅埋、中深埋。采用多元非线性回归方法对研究区“两带”高度数据进行分析，随后将每组数据以7:3的比例随机划分为训练数据和验证数据，再利用每组的70%训练数据建立神经网络模型(BP)、随机森林模型(RF)、粒子群优化支持向量机模型(PSO-SVR)和，遗传算法优化支持向量机模型(GA-SVR)，最后用剩余的30%检验样本完成模型精度检验计算及泛化能力评估，据此建立了典型浅埋区、近浅埋区和中深埋区三个陕北侏罗纪煤层典型赋煤特征区域的最优预测模型。

（2）采用多元非线性回归和4种机器学习方法，对三种不同埋深下的煤矿进行了“两带”高度预测，其结果表明：典型浅埋煤层导高适合用多元非线性回归方法来确定，精度为0.64；近浅埋煤层导高适合采用采高、埋深双因素PSO-SVR计算模型来预计，预测精度为0.84；中深埋煤层应运用采高一元非线性回归模型来确定导高，精度为0.79。垮落带高度可参照采高一元非线性回归方法来确定，精度为0.61，经与实测结果对比分析，验证了所选方法的合理性。

（3）综合研究区的“两带”高度、防水和防砂安全煤岩柱保护层厚度，将黄土沟壑径流下采动突水模式划分为三类，即突水溃砂模式、突水模式及渗漏模式。以红柳林矿井为例，采用本研究得出的预测模型进行“两带”高度预测，并与实测结果对比分析，进一步验证了预测模型的有效性。另外，使用PFC数值模拟平台构建了颗粒流数值采煤模型，对突水模型进行了研究，得出了红柳林矿井区域的三种采动突水模式，进而探讨了从裂隙贯通成缝到产生突水溃沙灾害的转化机理。

研究成果可为陕北侏罗纪煤层的绿色安全开发及突水溃沙灾害的预测和防治提供理论依据和科学依据。

The Jurassic coalfield in northern Shaanxi is famous for its large seam thickness, shallow burial and excellent coal quality, and is an important energy and chemical base in China. However, the ecological environment in northern Shaanxi is fragile, and along with the large-scale underground exploitation of coal resources, it will lead to the deformation and destruction of overburden, and the collapse zone and water-conducting fissure zone (referred to as the "two belts") will destroy groundwater resources, threaten the safety of ground buildings, and also destroy the originally fragile ecological environment. Therefore, accurately predicting the development height of the "two belts", exploring its development laws and influencing factors, and then carrying out research on water inrush patterns are of great significance to ensuring the safe production of coal mines and protecting the ecological environment of mining areas.

Based on the measured data of typical coal mines in the Jurassic coalfield in northern Shaanxi, and with reference to the relevant research results of predecessors, the height of the water-conducting fissure zone is divided into three types (typical shallow burial, near shallow burial, and medium and deep burial) according to the different burial depth and bedrock load ratio. Numerical methods such as multiple nonlinear regression and machine learning were used to analyze the "two belts" height prediction of the collected samples, and the best "two belts" height prediction method corresponding to different coal seam storage conditions was obtained. The prediction and analysis results are compared with the measured "two belts" height of the 15207 working surface of Hongliulin Coal Mine, which further verifies the feasibility of the numerical prediction method of this study. On the basis of this study, the water inrush pattern was studied, which provided a reliable basis for the prevention and control of groundwater damage in coal mines. The main research methods and achievements of the thesis are as follows:

(1) On the basis of obtaining and studying the height of the water-conducting fissure zone of 186 groups of Jurassic coal seams in northern Shaanxi, the height of the water-conducting fissure zone is divided into three types according to the burial depth and base load ratio, namely typical shallow burial, near shallow burial and medium and deep burial. The multivariate nonlinear regression method was used to analyze the "two belts" height data in the study area, and each set of data was randomly divided into training data and verification data in a 7:3 ratio, and then the neural network model (BP), random forest model (RF), particle swarm optimization support vector machine model (PSO-SVR) and genetic algorithm were established using 70% of the training data of each group, and the support vector machine model (GA-SVR) was optimized, and finally the remaining 30% was used. The test samples completed the model accuracy test calculation and generalization ability assessment, and based on this, the optimal prediction model of typical coal-enrichment characteristics in three mining areas in northern Shaanxi, namely shallow burial area, near shallow burial area and medium buried deep area, was established.

(2) Using multiple nonlinear regression and four machine learning methods, the "two belts" height prediction of coal mines under three different burial depths is carried out, and the results show that the typical shallow buried coal seam conductivity is suitable for multiple nonlinear regression method to determine the conductivity with an accuracy of 0.64; the near shallow buried coal seam conduction height is suitable for using the mining height and buried depth two-factor PSO-SVR calculation model to predict the prediction accuracy is 0.84; the medium and deep buried coal seam should use the mining height univariate nonlinear regression model to determine the conduction height, with an accuracy of 0.79. The height of the collapse zone can be determined by reference to the univariate nonlinear regression method of cessation of height, and the accuracy is 0.61, and the rationality of the selected method is verified by comparative analysis with the measured results.

(3) The height of the "two belts" in the comprehensive study area, the thickness of the protective layer of the waterproof and sandproof safe coal rock column divide the mining inrush mode under the loess gully runoff into three categories, namely the sand inrush type, the water inrush type and the leakage type. Taking the Red Willow Forest Mine as an example, the prediction model obtained in this study is used to predict the height of the "two belts", and the actual measurement results are compared and analyzed, which further verifies the effectiveness of the prediction model. In addition, a numerical coal mining model of particle flow was constructed using the PFC numerical simulation platform, and the water inrush model was studied, and three mining inrush modes in the Red Willow Forest mine area were obtained, and the transformation mechanism from fracture penetration to sand inrush disaster was discussed.

The research results can provide theoretical and scientific basis for the green and safe development of mining areas in northern Shaanxi and the prediction and prevention of water and sand inrush disasters.

[1] 刘晓. 煤炭矿区生态系统健康评价研究[D].西安科技大学, 2009.

[2] 张金锁, 张伟, 宋世杰, 等. 区域煤炭资源安全绿色高效开采评价指标体系与标准——以陕北侏罗纪煤田为例[J].中国煤炭, 2015, 41(06):49-54

[3] 余学义, 施文刚, 张平, 等. 黄土沟壑区地表移动变形特征分析[J].矿山测量, 2010, (02):38-40.

[4] 王启庆. 西北沟壑下垫层N2红土釆动破坏灾害演化机理研究[D].中国矿业大学, 2017.

[5] 闫朝波. 张家峁煤矿煤层顶板涌（突）水危险性分区预测研究[D].西安科技大学, 2013.

[6] 邵红旗, 王建文, 徐拴海, 等. 侏罗纪煤田顶板疏放水钻探设计方法探讨[J].矿业工程, 2014, 12(06): 59-61

[7] Ren G , Xie Y M . Optimization of underground excavation in rock masses using ESO techniques[J]. 2012.

[8] D., ELSWORTH, J., et al. Topographic Influence of Longwall Mining on Ground-Water Supplies[J]. Ground Water, 1995, 33(5): 786-793.

[9] BOOTH, J. C. The effects of longwall coal mining on overlying aquifers[J]. Geological Society London Special Publications, 2002, 198(1): 17-45.

[10] HAMLIN S N, ALPERS C N. Hydrogeology and geochemistry of acid mine drainage in ground water in the vicinity of Penn Mine and Camanche Reservoir, Calaveras County, California. Summary report, 1993-1995[J]. California, 1996.

[11] 钱鸣高, 缪协兴, 许家林. 岩层控制中的关键层理论研究[J]. 煤炭学报, 1996, (03): 2-7

[12] 许家林, 钱鸣高. 岩层采动裂隙分布在绿色开采中的应用[J]. 中国矿业大学学报, 2004, (02): 17-20.

[13] 许家林, 钱鸣高. 覆岩采动裂隙分布特征的研究[J]. 矿山压力与顶板管理, 1997, (Z1): 213-215.

[14] 许家林, 朱卫兵, 王晓振. 基于关键层位置的导水裂隙带高度预计方法[J]. 煤炭学报, 2012, 37(05): 762-769.

[15] GB/T 12719-2021, 矿区水文地质工程地质勘查规范[S].

[16] 吕广罗, 杨磊, 田刚军, 等. 深埋特厚煤层综放开采顶板导水裂隙带发育高度探查分析[J]. 中国煤炭, 2016, 42(11): 53-57.

[17] PALCHIK, V. Formation of fractured zones in overburden due to longwall mining[J]. Environmental Geology, 2003, 44(1).

[18] PALCHIK, V. Localization of mining-induced horizontal fractures along rock layer interfaces in overburden: field measurements and prediction[J]. Environmental Geology, 2005, 48(1).

[19] 李星亮, 雷武林. 应用井下仰孔注水测漏法探测导水裂隙带高度[J]. 内蒙古煤炭经济, 2018, (23): 117-128.

[20] 汪华君. 覆岩导水裂隙带井下微地震监测研究[J]. 矿业快报, 2006, (03): 27-29

[21] 张峰, 题正义, 秦洪岩, 等. 基于MATLAB多元非线性特厚煤层导水裂隙带高度预计[J]. 煤炭技术, 2021, 40(01): 56-61

[22] 胡小娟, 李文平, 曹丁涛, 等. 综采导水裂隙带多因素影响指标研究与高度预计[J].煤炭学报, 2012, 37(04): 613-620

[23] 娄高中, 郭文兵, 高金龙, 等. 非充分采动导水裂缝带高度影响因素敏感性分析[J].河南理工大学学报(自然科学版), 2019, 38(03): 24-31

[24] Dekang Zhao, Qiang Wu. An approach to predict the height of fractured water-conducting zone of coal roof strata using random forest regression [J]. Scientific Reports, 2018, 8(13).

[25] 邵良杉, 周玉. QGA-RFR模型在导水裂隙带高度预测中的应用[J]. 中国安全科学学报, 2018, 28(03): 19-24

[26] 藺哲渊. 基于支持向量机的导水裂缝带高度预测模型[J]. 矿山测量, 2013, (01): 51-55.

[27] 柴华彬, 张俊鹏, 严超. 基于GA-SVR的采动覆岩导水裂隙带高度预测[J]. 采矿与安全工程学报, 2018, 35(02): 359-365.

[28] 陈佩佩, 刘鸿泉, 朱在兴, 等. 基于人工神经网络技术的综放导水断裂带高度预计[J]. 煤炭学报, 2005, (04): 438-442.

[29 谢晓锋, 李夕兵, 尚雪义, 等. PCA-BP神经网络模型预测导水裂隙带高度[J]. 中国安全科学学报, 2017, 27(03): 100-105.

[30] 张新盈. K-means和QGA优化RBF神经网络模型在导水裂缝带高度预测方面的应用[J]. 中国矿业, 2018, 27(08): 164-167

[31] 刘国发, 王玉振. PSO-RBF神经网络在导水裂缝带高度预测中的应用[J]. 中国矿业, 2018, 27(05): 128-131

[32] 左建平, 孙运江, 王金涛, 等. 充分采动覆岩“类双曲线” 破坏移动机理及模拟分析[J]. 采矿与安全工程学报, 2018, 35(1): 71-77.

[33] 石磊. 基于PFC的覆岩破坏高度数值模拟及实测分析[J]. 矿业安全与环保, 2021, 48(03): 43-49.

[34] 侯恩科, 从通, 谢晓深, 等. 基于颗粒流的浅埋双煤层斜交开采地表裂缝发育特征[J]. 采矿与岩层控制工程学报, 2020, 2(01): 20-28.

[35] WANG C, ZHANG C, ZHAO X, et al. Dynamic structural evolution of overlying strata during shallow coal seam longwall mining [J]. 2018, 103: 20-32.

[36] 刘天泉. “三下一上”采煤技术的现状及展望[J]. 煤炭科学技术, 1995, (01): 5-7.

[37] 王琳琳, 魏久传, 尹会永, 等. 新安煤矿16煤导水裂隙带高度研究[J]. 山东科技大学学报(自然科学版), 2014, 33(01): 40-45

[38] 陈蓥, 张宏伟, 朱志洁, 等. 双系煤层开采相互影响下的覆岩运动与破坏规律分析[J]. 中国地质灾害与防治学报, 2014, 25(03): 67-73.

[39] 沙猛猛. 敏东一矿综放开采覆岩导水裂隙带演化规律研究[D].中国矿业大学, 2018.

[40] 徐智敏, 孙亚军, 高尚, 等. 干旱矿区采动顶板导水裂隙的演化规律及保水采煤意义[J]. 煤炭学报, 2019, 44(03): 767-776.

[41] 冯国财, 徐白山, 王东. 三台子水库下压煤综放开采覆岩破坏充水特征[J]. 采矿与安全工程学报, 2014, 31(01): 108-114.

[42] WENPING L, QIQING W, SHILIANG L, et al. Study on the creep permeability of mining-cracked N 2 laterite as the key aquifuge for preserving water resources in Northwestern China [J]. International Journal of Coal Science & Technology, 2018, 5(3).

[43] 李东, 刘生优, 张光德, 等. 鄂尔多斯盆地北部典型顶板水害特征及其防治技术[J]. 煤炭学报, 2017, 42(12): 3249-3254.

[44] 张杰, 杨涛, 索永录, 等. 基于隔水土层失稳模型的顶板突水致灾预测研究[J]. 煤炭学报, 2017, 42(10): 2718-2724.

[45] 吕玉广, 肖庆华, 程久龙, 等. 弱富水软岩水-沙混合型突水机制与防治技术——以上海庙矿区为例[J]. 煤炭学报, 2019, 44(10): 3154-3163.

[46] 吕玉广, 肖庆华, 韩港. 软岩矿区顶板弱含水层高强度携沙突水机理研究[J]. 煤矿安全, 2019, 50(01): 38-42.

[47] 孟召平, 高延法, 卢爱红, 等. 第四系松散含水层下煤层开采突水危险性及防水煤柱确定方法[J]. 采矿与安全工程学报, 2013, 30(01): 23-29.

[48] 陈伟. 陕北黄土沟壑径流下采动水害机理与防控技术研究[D].中国矿业大学, 2015.

[49] 黎灵, 舒宗运, 冯宇锋. 特厚煤层综放开采覆岩离层水突水机理分析及防治[J]. 煤炭科学技术, 2018, 46(01): 175-182.

[50] TANG C A, THAM L G, LEE P K K, et al. Coupled analysis of flow, stress and damage (FSD) in rock failure [J]. International Journal of Rock Mechanics and Mining Sciences, 2002, 39(4).

[51] 隋旺华, 蔡光桃, 董青红. 近松散层采煤覆岩采动裂缝水砂突涌临界水力坡度试验[J]. 岩石力学与工程学报, 2007, (10): 2084-2089.

[52] 孙亚军, 徐智敏, 董青红. 小浪底水库下采煤导水裂隙发育监测与模拟研究[J]. 岩石力学与工程学报, 2009, 28(02): 238-245

[53] 武强, 许珂, 张维, 等. 再论煤层顶板涌(突)水危险性预测评价的“三图-双预测法”[J]. 煤炭学报, 2016, 41(06): 1341-1347

[54] 伊茂森, 朱卫兵, 李林, 等. 补连塔煤矿四盘区顶板突水机理及防治[J]. 煤炭学报, 2008, (03): 241-245

[55] 孙宗海. 支持向量机及其在控制中的应用研究 [D].浙江大学, 2003.

[56] 薛建坤, 王皓, 赵春虎, 等. 鄂尔多斯盆地侏罗系煤田导水裂隙带高度预测及顶板充水模式[J]. 采矿与安全工程学报, 2020, 37(6): 1222-1230.

[57] 魏安邦. 薄基岩厚松散层覆岩破断及导水裂隙发育规律[D].西安科技大学,2020.

[58] XIN N, GU X, WU H, et al. Application of genetic algorithm‐support vector regression (GA‐SVR) for quantitative analysis of herbal medicines [J]. 2012, 26(7): 353-360.

[59] 曹庆奎, 赵斐. 基于遗传-支持向量回归的煤层底板突水量预测研究[J]. 煤炭学报, 2011, 36(12): 2097-2101.

[60] ZHAO DEKANG A W Q. An approach to predict the height of fractured water-conducting zone of coal roof strata using random forest regression [J]. Scientific Reports%J %@ 2045-2322, 2018, 8(1).

[61] CUNDALL P A, STRACK O D J G. A discrete numerical model for granular assemblies [J]. 1979, 29(1): 47-65.

[62] CHO N A, MARTIN C, SEGO D J I J O R M, et al. A clumped particle model for rock [J]. 2007, 44(7): 997-1010.

[63] 严由吉. 深埋煤层采动覆岩导水裂隙带发育高度研究[D].西安科技大学 2021.

[64] 代革联, 薛小渊, 牛超, 等. 煤炭开采对相邻区域生态潜水流场扰动特征[J]. 煤炭学报, 2019, 44(03): 701-708.

[65] 刘瑜. 陕北侏罗系煤层开采导水裂缝带动态演化规律研究及应用 [D].中国矿业大学, 2018.

[66] 申涛. 榆神矿区2~(-2)煤层开采导水裂隙带发育规律研究 [D].西安科技大学, 2019.

[67] 刘少伟. 离石黄土覆盖下综放首采面采动覆岩破坏规律与水害分区研究[D].中国矿业大学, 2018.

[68] 李申. 基于样本优化与支持向量回归的导水裂隙带高度预测模型应用研究[D].中国矿业大学, 2019.

[69] 范继超, 谭二民. 基于微震事件能量-密度导水裂隙带高度研究[J]. 内蒙古煤炭经济, 2020, (03): 206-207.

[70] 杨俊哲, 胡博文, 王振荣. 8.8 m大采高工作面覆岩三带分布特征及分层沉降研究[J]. 煤炭科学技术, 2020, 48(06): 42-48.

[71] 王永国, 王明, 许蓬. 巴彦高勒煤矿3-1煤层顶板垮落裂缝带发育特征[J]. 煤田地质与勘探, 2019, 47(S1): 37-42.

[72] 田灵涛. 察哈素煤矿采空区覆岩“两带”高度研究 [J]. 河南理工大学学报(自然科学版), 2019, 38(05): 22-27.

[73] 李海军. 红柳林煤矿浅埋煤层群开采覆岩导水裂隙带发育规律研究[D]; 西安科技大学 2019.

[74] 刘士亮. 陕北侏罗系煤田开采环境工程地质模式研究[D].中国矿业大学, 2019.

[75] 袁喜东. 榆神矿区导水裂隙带发育规律研究[D].西安科技大学 2017.

[76] 范立民, 马雄德, 蒋泽泉, 等. 保水采煤研究30年回顾与展望[J]. 煤炭科学技术, 2019, 47(07): 1-30.

[77] 顾世乾. 杭来湾煤矿涌水危险性分析及其防治技术研究 [D].中国矿业大学 2019.

[78] 肖华. 榆神矿区侏罗—白垩系含水介质结构与采动破坏特征 [D].中国矿业大学 2017.

[79] 马亚杰, 武强, 章之燕, 等. 煤层开采顶板导水裂隙带高度预测研究[J]. 煤炭科学技术, 2008, 36(5): 59-62.

[80] 王永国, 王明, 许蓬. 巴彦高勒煤矿3-1煤层顶板垮落裂缝带发育特征[J]. 煤田地质与勘探, 2019, 47(z1): 37-42.

[81] 谢朋. 石拉乌素煤矿综放开采导水裂缝带高度及涌水量预计[D].中国矿业大学 2018.

[82] 霍晓斌. 浅埋大采高工作面覆岩破坏规律及导水裂隙带发育高度研究[D].西安科技大学, 2018.

[83] 曹丁涛, 李文平, 中国矿业大学. 煤矿导水裂隙带高度计算方法研究[J]. 中国地质灾害与防治学报, 2014, 25(01): 63-69.

[84] 张云, 曹胜根, 郭帅, 等. 含水层下短壁块段式采煤导水裂隙带高度发育规律研究[J]. 采矿与安全工程学报, 2018, 35(01): 106-111.

[85] 张世忠. 伊犁矿区弱胶结地层采动阻水性能演化规律及其控制机理[D]; 中国矿业大学, 2021.

[86] 范志忠. 大采高综采面围岩控制的尺度效应研究[D].中国矿业大学(北京), 2019.

[87] 施龙青, 吴洪斌, 李永雷, 等. 导水裂隙带发育高度预测的PCA-GA-Elman优化模型[J]. 河南理工大学学报(自然科学版), 2021, 40(04): 10-18.

[88] 孙学阳, 夏玉成. 煤矿区构造环境内涵及类型划分[J]. 煤田地质与勘探, 2015, 43(04): 79-84.

[89] 黄庆享. 浅埋煤层的矿压特征与浅埋煤层定义[J]. 岩石力学与工程学报, 2002, (08): 1174-1177.

[90] 曹丁涛, 李文平, 中国矿业大学. 煤矿导水裂隙带高度计算方法研究[J]. 中国地质灾害与防治学报, 2014, 25(01): 63-69.

[91] 张立琪. 调节系数的BP神经网络在字符识别中的研究[D].哈尔滨工程大学. 2010.

[92] 陈博, 欧阳竹. 中国科学院研究生院. 基于BP神经网络的冬小麦耗水预测[J]. 农业工程学报, 2010, 26(04): 81-86.

[93] 尤杰, 车轶, 仲伟秋. 基于BP神经网络的既有建筑混凝土强度预测[J]. 建筑科学与工程学报, 2011, 28(01): 70-75.

[94] 钱征文, 程礼, 赵兵兵, 等. 基于BP神经网络的叶片损伤度评估方法[J]. 航空动力学报, 2011, 26(04): 794-800.

[95] 周治平, 孙子文, 吴志健. 采用L-M算法的JPEG图像隐写分析[J]. 计算机工程与应用, 2009, 45(14): 113-115.

[96] SUN J, MAO H, LIU J, et al. The research of paddy rice moisture lossless detection based on LM BP neural network; proceedings of the International Conference on Computer and Computing Technologies in Agriculture, F, 2008 [C]. Springer.

[97] 汪马成. 基于PSO-SVM模型的建筑施工项目安全预警方法研究[D].上海工程技术大学. 2016.

[98] 赵春虎. 陕蒙煤炭开采对地下水环境系统扰动机理及评价研究[D].煤炭科学研究总院. 2016.

[99] 邢茂林, 李文平, 阴静慧. 侏罗系煤综放开采导水断裂带高度预计研究[J]. 煤矿安全, 2017, 48(09): 39-42.

[100] 沈佳. 山区煤矿采空区地表残余移动变形时空变化规律研究[D].江苏师范大学 2017.

[101] 皮希宇. 煤层群采动卸压煤与覆岩裂隙演化特征及其对瓦斯抽采的影响[D].北京科技大学, 2021.

[102] 许珂. 台格庙矿区顶板涌（突）水危险性评价与矿井涌水量预测[D].中国矿业大学(北京) 2016.

[103] 《煤炭工业部》. 建筑物、水体、铁路及主要井巷煤柱留设与压煤开采规程[M]. 煤炭工业出版社, 1986.

[104] 阿比尔的, 郑颖人, 冯夏庭, 等. 平行黏结模型宏细观力学参数相关性研究[J]. 岩土力学, 2018, 39(04): 1289-1301.

[105] B Yang, J Yue, S Lei. A study on the effects of microparameters on macroproperties for specimens created by bonded particles[J]. Engineering computations: International journal for computer-aided engineering and software, 2006, 23(5/6): 607-631.

[106] DR.P.A.CUNDALL. Numerical experiments on localization in frictional materials[J]. Ingenieur-Archiv, 1989, 59(2): 148-159.

[107] Hou Juan, Zhang Mengxi, Chen Qiang, et al. Failure-mode analysis of loose deposit slope in Ya'an-Kangding Expressway under seismic loading using particle flow code[J]. Granular matter, 2019, 21(1): 8.1-8.12.

[108] DOP A, PAC B. bonded-particle model for rock[J]. International journal of rock mechanics and mining sciences, 2004, 41(8): 1329-1364.

[109] SUN Q C, XIN H L, LIU J G, et al. Skeleton and force chain network in static granular material[J]. Rock and Soil Mechanics, 2009, 30(S1): 83-87.