边坡工程是露天煤矿中的重点工程，边坡的稳定性关系着煤矿的安全生产。边坡稳定性预测是边坡防治工作的前提，针对煤矿边坡工程稳定性预测的复杂性，为了快速、有效地判别煤矿边坡稳定性，利用随机森林算法建立煤矿边坡稳定性预测模型。通过选取与煤矿边坡工程密切相关的岩石重度、黏聚力、内摩擦角、边坡角、边坡高度、孔隙水压力6 个指标作为边坡稳定性的影响因素，即为随机森林预测模型的输入，边坡稳定性状态作为随机森林预测模型的输出，通过随机森林算法建立边坡稳定性影响因素与边坡稳定状态之间的非线性关系。利用煤矿实测30 组边坡稳定性数据作为随机森林预测模型的训练数据集，进行模型的学习训练；另用12 组边坡稳定性数据作为预测模型的测试数据，通过训练好的边坡稳定性预测模型进行测试；为了验证随机森林预测模型的准确率，同时与SVM 和BP 神经网络的测试数据进行比较。结果说明，选取煤矿边坡稳定性的6 个指标建立的随机森林预测模型，人工控制参数较少、结构简单、容易实现，且具有较高的准确度，边坡稳定状态预测结果与煤矿边坡工程实际状态相吻合，能有效预测边坡稳定性状态，指导煤矿边坡防治工作的开展。
Slope engineering is a key project in open-pit coal mines. The stability of the slope is closely related to safety production of coal mines. Slope stability prediction is a prerequisite in slope control, faced with complexities. To quickly and effectively determine the coal mine slope stability, this paper establishes a prediction model using the random forest algorithm. Six factors influencing the slop stability were selected as input of the prediction model, including the gravity density of rocks, cohesive force, internal friction angle, slope angle, slope height and pore water pressure, and slope stability status was selected as output of the prediction model. The random forest algorithm was used to establish the nonlinear relationship between slope stability factors and stability status. The 30 sets of measured data were used as training data set to learn and train the random forest slope stability prediction model. In addition, 12 groups of data as slope stability test data were used to test the trained prediction models. In the meantime, the accuracy of the random forest prediction models was tested by comparing them with the SVM and BP neural network prediction models. The results show that the random forest prediction model based on the selected six factors has less manual control parameters, simple structure and high accuracy. The predictive results coincide with the actual state of the slope project, indicating that the prediction model is able to predict the slope stability effectively and provide guidance to coal mine slope prevention work.
[1] Sharma R K, Mehta B S, Jamwal C S. Cut slope stability evaluation of NH-21 along Nalayan-Gambhrola section, Bilaspur district, Himachal Pradesh, India[J]. Nat Hazards, 2013, 66(6): 249-270.
[2] 张豪, 罗亦泳. 基于人工免疫算法的边坡稳定性预测模型[J]. 煤炭学 报, 2012, 37(6): 911-917. Zhang Hao, Luo Yiyong. Prediction model for slope stability based on artificial immune algorithm[J]. Journal of China Coal Society, 2012, 37 (6): 911-917.
[3] 何方维, 朱明, 刘文生, 等. BP网络在露天矿边坡角优化中的应用[J]. 金属矿山, 2011(1): 35-38. He Fangwei, Zhu Ming, Liu Wensheng, et al. Application of BP artificial neural network in optimization of open-pit slope angle[J]. Metal Mine, 2011(1): 35-38.
[4] 张均锋, 丁烨. 边坡稳定性分析的三维极限平衡法及应用[J]. 岩石力 学与工程学报, 2005, 24(3): 365-370. Zhang Junfeng, Ding Ye. Generalized 3D limit-equilibrium method for slope stability analysis and its application[J]. Chinese Journal of Rock Mechanics and Engineering, 2005, 24(3): 365-370.
[5] Chen Z, Wang X, Haberfield C, et al. A three- dimensional slope stability analysis method using the upper bound theorem Part I theory and methods[J]. International Journal of Rock Mechanics & Mining Sciences, 2001, 38(3): 369-378.
[6] 程纬华, 乔登攀, 张磊, 等. BP神经网络在露天矿边坡稳定性分析中 的应用[J]. 矿冶, 2012, 21(2): 10-15. Cheng Weihua, Qiao Dengpan, Zhang Lei, et al. Application of BP networks in the stability analysis of slopes in the open-pit mine[J]. Mining & Metallurgy, 2012, 21(2): 10-15.
[7] 乔金丽, 刘波, 李艳艳, 等. 基于遗传规划的边坡稳定安全系数预测[J]. 煤炭学报, 2010, 35(9): 1466-1469. Qiao Jinli, Liu Bo, Li Yanyan, et al. The prediction of the safety factor of the slope stability based on genetic programming[J]. Journal of China Coal Society, 2010, 35(9): 1466-1469.
[8] 马海兴, 张刚. 基于LS-SVM的边坡稳定性预测研究[J]. 宁夏大学学 报: 自然科学版, 2012, 33(3): 250-253. Ma Haixing, Zhang Gang. Study on slope stability prediction based on LS- SVM[J]. Journal of Ningxia University: Natural Science Edition, 2012, 33(3): 250-253.
[9] Breiman L. Random forests[J]. Machine Learning, 2001, 45(1): 5-32.
[10] 马昕, 郭静, 孙啸. 蛋白质中RNA-结合残基预测的随机森林模型[J]. 东南大学学报: 自然科学版, 2012, 42(1): 50-54. Ma Xin, Guo Jing, Sun Xiao. Prediction of RNA-binding residues in proteins using random forest[J]. Journal of Southwest University: Natural Science Edition, 2012, 42(1): 50-54.
[11] 赵小欢, 夏靖波, 李明辉. 基于随机森林算法的网络流量分类方法[J]. 中国电子科学研究院学报, 2013, 8(2): 184-190. Zhao Xiaohuan, Xia Jingbo, Li Minghui. Research on classification of network traffic based on random forests algorithm[J]. Journal of China Academy of Electronics and Information Technology, 2013, 8(2): 184- 190.
[12] 董师师, 黄哲学. 随机森林理论浅析[J]. 集成技术, 2013, 2(1): 1-7. Dong Shishi, Huang Zhexue. A brief theoretical overview of random forests[J]. Journal of Integration Technology, 2013, 2(1): 1-7.
[13] 杨帆, 林琛, 周绮凤, 等. 基于随机森林的潜在k 近邻算法及其在基 因表达数据分类中的应用[J]. 系统工程理论与实践, 2012, 32(4): 815-825. Yang Fan, Lin Chen, Zhou Qifeng, et al. Random forest based potential k nearest neighbor classifier and its application in gene expression data[J]. Systems Engineering Theory & Practice, 2012, 32 (4): 815-825.
[14] 庄进发, 罗键, 彭彦卿, 等. 基于改进随机森林的故障诊断方法研究[J]. 计算机集成制造系统, 2009, 15(4): 777-785. Zhuang Jinfa, Luo Jian, Peng Yanqing, et al. Fault diagnosis method based on modified random forests[J]. Computer Integrated Manufacturing Systems, 2009, 15(4): 777-785.
[15] Gorog P, Torok A. Slope stability assessment of weathered clay by using field data and computer modeling: a case study from Budapest[J]. Natural Hazards and Earth System Sciences, 2007, 7(3): 417-422.
