img

QQ群聊

img

官方微信

高级检索

黄金科学技术 ›› 2023, Vol. 31 ›› Issue (3): 497-506.doi: 10.11872/j.issn.1005-2518.2023.03.122

• 采选技术与矿山管理 • 上一篇    下一篇

Adaboost集成学习优化的巷道围岩松动圈预测研究

方博扬(),赵国彦(),马举,陈立强,简筝   

  1. 中南大学资源与安全工程学院,湖南 长沙 410083
  • 收稿日期:2022-09-19 修回日期:2023-03-15 出版日期:2023-06-30 发布日期:2023-07-20
  • 通讯作者: 赵国彦 E-mail:591788294@qq.com;gy.zhao@263.net
  • 作者简介:方博扬(1998-),男,湖北孝感人,硕士研究生,从事地压智能监测及灾害控制研究工作。591788294@qq.com
  • 基金资助:
    ‘十三五’国家重点研发计划课题“深部金属矿绿色开采关键技术研发与示范”(2018YFC0604606)

Prediction Study on Loosening Ring of Surrounding Rock Around Roadways Using the Optimized Ensemble Learning Algorithms Based on Adaboost

Boyang FANG(),Guoyan ZHAO(),Ju MA,Liqiang CHEN,Zheng JIAN   

  1. School of Resources and Safety Engineering,Central South University,Changsha 410083,Hunan,China
  • Received:2022-09-19 Revised:2023-03-15 Online:2023-06-30 Published:2023-07-20
  • Contact: Guoyan ZHAO E-mail:591788294@qq.com;gy.zhao@263.net

摘要:

为提高巷道围岩松动圈预测准确率,给围岩支护和地压管理提供更科学的指导,提出了一种新的预测方法。采用改进的Adaboost回归算法对3种机器学习算法进行集成优化,即在Adaboost回归算法中寻找误差率阈值的最优值,实现Adaboost全局最优的集成效果。应用网格搜索对BP、SVM和RF的超参数进行优化,建立BP-Adaboost、SVM-Adaboost和RF-Adaboost回归预测模型。结果表明:BP-Adaboost模型的预测性能最好,误差率为7.65%。结合矿山松动圈测试实例进行验证分析,平均相对误差为4.15%。因此,所提出的模型能够为围岩松动圈预测提供参考,可以满足工程应用的需求。

关键词: 围岩松动圈, 网格搜索, Adaboost算法, BP神经网络, 支持向量机, 随机森林

Abstract:

In order to improve the prediction accuracy of loose zone of excavation damaged zone around roadways and provide more scientific guidance for surrounding rock support and ground pressure management,a new prediction method was proposed.The improved Adaboost regression algorithm was used to integrate and optimize three machine learning algorithms,the optimal value of the error rate threshold was found to achieve the global optimal integration of Adaboost.The grid search was used to optimize the hyperparameters of BP,SVM and RF,and the regression prediction models of BP-Adaboost,SVM-Adaboost and RF-Adaboost were established.The results show that the prediction performance of BP-Adaboost is the best,it had the lowest error rate at 7.65 percent.The verification analysis was carried out based on the test example of excavation damaged zone around roadway,the results show that the mean relative error is 4.15%.Therefore,the model proposed in this paper can provide reference for the excavation damaged zone around roadway and meet the needs of engineering applications.

Key words: loosening ring of surrounding rock, grid search, Adaboost algorithm, Back Propagation Neural Network(BPNN), Support Vector Machine(SVM), Random Forest(RF)

中图分类号: 

  • TD322

图1

Adaboost算法优化的3种机器学习方法的围岩松动圈回归预测模型"

表1

样本数据"

样本编号巷道埋深/m巷道跨度/m掘进断面面积/m2单轴抗压强度/MPa节理发育程度厚度/m样本编号巷道埋深/m巷道跨度/m掘进断面面积/m2单轴抗压强度/MPa节理发育程度厚度/m
13622.66.862.420.6346893.07.615.141.8
26604.414.612.552.2354503.07.611.231.2
33843.511.58.531.2364103.611.713.341.4
41503.611.714.620.6373483.29.27.531.2
51782.66.423.831.2383573.28.510.531.1
65103.27.312.641.6392732.66.615.920.8
74203.610.314.331.2402802.87.112.720.8
84503.44.89.152.0413212.66.615.920.8
92363.07.514.331.2426654.414.610.941.7
104704.012.610.152.2433503.28.510.531.2
114673.48.211.231.0443212.66.69.231.2
124903.78.912.541.8453403.07.673.620.8
134503.610.813.341.6464703.611.29.152.1
142243.48.211.231.0472313.07.518.320.7
154603.29.7101.610.4481253.49.813.331.0
163732.56.314.620.9492963.47.822.441.4
173102.87.113.831.2504362.87.215.231.2
181252.87.113.320.7513433.29.632.220.7
193922.86.914.520.8525253.27.315.841.6
202493.48.216.831.0532643.29.211.231.1
211403.610.313.420.5542923.47.812.541.4
223453.07.665.020.7553622.66.858.020.8
233152.87.111.231.1561802.87.1110.210.3
245503.49.412.552.1573622.66.862.420.6
254103.27.213.331.1583403.29.632.220.7
264203.29.29.141.7594673.49.610.141.8
273403.29.219.831.3602683.07.512.031.4
283403.29.632.220.7612363.07.514.331.2
294203.78.99.141.4623212.66.613.331.1
303703.58.310.531.063973.28.811.241.2
314283.611.716.531.2643223.47.714.341.5
324654.012.69.541.6652933.58.311.931.1
334032.97.212.631.3664503.47.89.152.0

图2

样本散点与相关系数矩阵图"

图3

BPNN的网格搜索过程"

图4

SVM的网格搜索过程注:底面曲线图形为曲面等高线投影"

图5

误差率参数寻优过程"

表2

不同预测模型评价指标对比(测试集)"

模型误差MSER2MAE/%
BPNN-0.015270.9239.83
BP_Ada0.500. 013760.9318.05
BP_Ada0.600.012730.9367.65
SVM-0.014120.92810.81
SVM_Ada0.500.013440.93210.29
SVM_Ada0.590.013180.9339.58
RF-0.023250.88215.36
RF_Ada0.330.024730.87715.13
RF_Ada0.500.024730.87715.13

图6

真实值与不同模型的预测值(测试集)"

表3

松动圈厚度实测值与计算值的对比"

序号矿山名称H/mB/mS/m2R/MPaF实际值/m预测值/m相对误差/%
实例一三山岛金矿6003.814.271.2631.101.165.45
实例二

大顶山矿区

大顶山矿区

4203.67.814.3031.101.121.82
实例三1413.27.837.9041.351.425.18
Chao Z, Kim H,2020. Brain image segmentation based on the hybrid of back propagation neural network and Adaboost system[J].Journal of Signal Processing Systems for Signal Image and Video Technology,92(3):289-298.
Chen Chong, Li Xibing, Feng Fan,2018.Numerical study on damage zones of the induced roadway surrounding rock[J].Gold Science and Technology,26(6):771-779.
Dong Fangting,2001.The Supporting Theory Based on Broken Rock Zone and Its Application Technology[M]. Beijing:China Coal Industry Publishing Home.
Gao Wei, Zheng Yingren,2002.Evolutionary neural network model on prediction of loosen zone around roadway[J]. Chinese Journal of Rock Mechanics and Engineering,(5):658-661.
Jing Feng,2008.Research on the Distribution Rule of the Shallow Crustal Geostress Field in the China Mainland and Engineering Disturbance Characteristics[D].Beijing:Institute of Rock and Soil Mechanics,Chinese Academy of Sciences.
Jing Yue, Wang Shaofeng, Lu Jintao,2021.Thickness prediction of the excavation damage zone and non-explosive mechanized mining criterion[J].Gold Science and Technology,29(4):525-534.
Li Guosheng, Zhang Hui, Jiang Shuaiqi,2018.Technology for enhancing supporting roadway surrounding rock influenced by frequent mining and its application[J].China Sa-fety Science Journal,28(7):142-147.
Momeni M, Peyghami M R, Tarzanagh D A,2020. A new stochastic limited memory BFGS algorithm[J].Journal of Ma-thematical Extension,14(3):65-83.
Qian Zhenyu, Ren Fenhua, Miao Shengjun,et al,2013.Optimization design of roadway support based on field measurement of surrounding rock broken zone[J].Metal Mine,42(10):16-20.
Shen Jinsheng, Tang Xionghou,2012.Study on influence factors of broken rock zone around mining roadway[J].Mineral Engineering Research,27(2):10-14.
Wang Wei,2014.Study on Dading Mountain Mine RoadwayRock Loose Circle the Mining and Its Influence[D].Mianyang:Southwest University of Science and Technology.
Wang Xinfeng, He Yi, Lu Mingyuan,et al,2021.Study on deformation and failure characteristics of deep roadway surrounding rock under excavation unloading disturbance [J]. China Safety Science Journal,31(8):83-90.
Wu Tao, Dai Jun, Du Meili,et al,2015.Surrounding rock loosing circle test based on acoustic test technology[J].Safety in Coal Mines,46(1):169-172.
Wu Yongping, Zhai Jin, Xie Panshi,et al,2013.Measurement of loosing circle in surrounding rock of gateway based on technology of geological radar detection[J].Coal Science and Technology,41(3):32-34,38.
Xiong B, Li R, Ren D,et al,2021.Prediction of flooding in the downstream of the Three Gorges Reservoir based on a back propagation neural network optimized using the Adaboost algorithm[J].Natural Hazards,107(2):1559-1575.
Xue Xinhua,2006.Application of algorithm neural network method in the prediction of loosen zone around roadway[J]. Geotechnical Engineering Technique,(5):237-239.
Yu Qinglei, Pu Jiangyong, Le Zhihua,et al,2021.Study on the broken rock zone of roadway in skarn copper-iron mine based on geological radar[J]. Metal Mine,50(3):46-53.
Zhao Guoyan, Wu Hao,2013.Support vector machine model for predicting the thickness of excavation damaged zone[J]. Journal of Guangxi University(Natural Science Edition),38(2):444-450.
Zhao Linlin, Wen Guofeng, Shao Liangshan,2018.PCA-Adaboost model for predicting coal spontaneous combustion in caving zone with imbalanced data[J].China Safety Science Journal,28(3):74-78.
Zhou J, Li E, Yang S,et al,2019. Slope stability prediction for circular mode failure using gradient boosting machine approach based on an updated database of case histories[J]. Safety Science,118:505-518.
Zhou J, Li X B,2011. Evaluating the thickness of broken rock zone for deep roadways using nonlinear SVMs and multiple linear regression model[J]. Procedia Engineering,26:972-981.
Zhu Chuanqu,1999.Stability classification and prediction of lo-osening circle size in the surrounding rock of back mining roadway[J].Gold Science and Technology,7(Supp.1):64-67.
Zhu Zhijie, Zhang Hongwei, Chen Ying,2014. Prediction model of loosening zones around roadway based on MPSO-SVM[J].Computer Engineering and Applications,50(12):1-5.
陈冲,李夕兵,冯帆,2018.诱导巷道的围岩松动破坏区数值研究[J].黄金科学技术,26(6):771-779.
董方庭,2001. 巷道围岩松动圈支护理论与应用技术[M]. 北京:煤炭工业出版社.
高玮,郑颖人,2002.巷道围岩松动圈预测的进化神经网络法[J].岩石力学与工程学报,(5):658-661.
景锋,2008. 中国大陆浅层地壳地应力场分布规律及工程扰动特征研究[D].北京:中国科学院武汉岩土力学研究所.
景岳,王少锋,鲁金涛,2021.矿岩开挖松动区厚度预测及非爆机械化开采判据[J].黄金科学技术,29(4):525-534.
李国盛,张辉,蒋帅旗,2018. 频繁采动影响巷道围岩强化支护技术及其应用[J].中国安全科学学报,28(7):142-147.
钱振宇,任奋华,苗胜军,等,2013.基于围岩松动圈现场测量的巷道支护优化[J].金属矿山,42(10):16-20.
沈金生,唐雄厚,2012. 回采巷道围岩松动圈影响因素分析[J].矿业工程研究,27(2):10-14.
王伟,2014. 大顶山矿区回采巷道围岩松动圈及影响研究[D]. 绵阳:西南科技大学.
王新丰,何毅,陆明远,等,2021.开挖卸荷扰动深部巷道围岩变形破坏特征研究[J].中国安全科学学报,31(8):83-90.
吴涛,戴俊,杜美利,等,2015.基于声波法测试技术的巷道围岩松动圈测定[J].煤矿安全,46(1):169-172.
伍永平,翟锦,解盘石,等,2013.基于地质雷达探测技术的巷道围岩松动圈测定[J].煤炭科学技术,41(3):32-34,38.
薛新华,2006.遗传神经网络法在巷道围岩松动圈预测中的应用[J].岩土工程技术,(5):237-239.
于庆磊,蒲江涌,勒治华,等,2021.基于地质雷达的矽卡岩型铜铁矿巷道松动圈研究[J].金属矿山,50(3):46-53.
赵国彦,吴浩,2013.松动圈厚度预测的支持向量机模型[J].广西大学学报(自然科学版),38(2):444-450.
赵琳琳,温国锋,邵良杉,2018.不均衡数据下的采空区煤自燃PCA-Adaboost预测模型[J].中国安全科学学报,28(3):74-78.
朱川曲,1999.回采巷道围岩稳定性分类及松动圈尺寸预测[J].黄金科学技术,7(增1):64-67.
朱志洁,张宏伟,陈蓥,2014.基于MPSO-SVM巷道围岩松动圈预测研究[J].计算机工程与应用,50(12):1-5.
[1] 温廷新,苏焕博. 基于MICE_RF的组合赋权—极限随机树岩爆预测模型[J]. 黄金科学技术, 2022, 30(3): 392-403.
[2] 谢饶青, 陈建宏, 肖文丰. 基于NPCA-GA-BP神经网络的采场稳定性预测方法[J]. 黄金科学技术, 2022, 30(2): 272-281.
[3] 胡建华,郭萌萌,周坦,张涛. 基于改进迁移学习算法的岩体质量评价模型[J]. 黄金科学技术, 2021, 29(6): 826-833.
[4] 骆正山,黄仁惠,申国臣. 基于KPCA-IPSO-LSSVM的充填管道磨损风险预测[J]. 黄金科学技术, 2021, 29(2): 245-255.
[5] 谭吉玉,刘高常. 基于小波支持向量机模型的矿区生态安全评价方法研究[J]. 黄金科学技术, 2020, 28(6): 902-909.
[6] 田睿,孟海东,陈世江,王创业,孙德宁,石磊. 基于机器学习的3种岩爆烈度分级预测模型对比研究[J]. 黄金科学技术, 2020, 28(6): 920-929.
[7] 许瑞, 侯奎奎, 王玺, 刘兴全, 李夕兵. 基于核主成分分析与SVM的岩爆烈度组合预测模型[J]. 黄金科学技术, 2020, 28(4): 575-584.
[8] 卜斤革,陈建宏. 基于粒子群算法优化BP神经网络的溶浸开采浸出率预测[J]. 黄金科学技术, 2020, 28(1): 82-89.
[9] 韩超群,陈建宏,周智勇,杨珊. 基于主成分分析—支持向量机模型的矿岩可爆性等级预测研究[J]. 黄金科学技术, 2019, 27(6): 879-887.
[10] 肖文丰,陈建宏,陈毅,王喜梅. 基于神经网络与遗传算法的多目标充填料浆配比优化[J]. 黄金科学技术, 2019, 27(4): 581-588.
[11] 随晓丹,罗周全,秦亚光,王玉乐,彭东. 基于小波分解的尾矿坝浸润线预测方法研究[J]. 黄金科学技术, 2019, 27(1): 137-143.
[12] 刘强,李夕兵,梁伟章. 岩体质量分类的PCA-RF模型及应用[J]. 黄金科学技术, 2018, 26(1): 49-55.
[13] 邓高,杨珊. 基于组合预测与变精度粗糙模糊集的采空区稳定性评价[J]. 黄金科学技术, 2017, 25(3): 98-107.
[14] 张钦礼,王兢*,王新民. 基于核主成分分析与PSO-SVM的充填管道失效风险性分级评价模型[J]. 黄金科学技术, 2017, 25(3): 70-76.
[15] 史秀志,范玉乾,尚雪义. 基于PCA-BP神经网络模型的充填体强度预测[J]. 黄金科学技术, 2016, 24(3): 64-69.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
[1] 刘金鹏, 刘万强, 李定坤, 李天栋. 金亭岭金矿竖井提升系统的合理化改造[J]. J4, 2010, 18(4): 80 -81 .
[2] 王宝胜. 氰化作业中降低氰化钠消耗的实践[J]. J4, 2007, 15(6): 43 -46 .
[3] 吴姗姗, 赵莎, 牛树银, 王宝德, 孙爱群, 李富, 杨宝军. 河北崇礼县中山沟金矿深部成矿分析[J]. J4, 2009, 17(1): 6 -9 .
[4] 孙刚,曲声,高臣. 四山林场铜矿地质特征及找矿标志[J]. J4, 2008, 16(3): 40 -42 .
[5] 黄广平, 杨增武, 董传统. 黑龙江金场沟区铜钼矿成矿地质条件分析[J]. J4, 2007, 15(3): 24 -27 .
[6] 张印飞,李晓波. 陕西双王金矿床边界品位和最低工业品位的计算及意义[J]. J4, 2004, 12(6): 23 -26 .
[7] 陈海明, 刘智明. 东安金矿床控矿地质条件的研究[J]. J4, 2005, 13(03): 17 -20 .
[8] 赵国荣, 张新勇, 张志臣. 河北王林口金矿地质特征与控矿机制分析[J]. J4, 2006, 14(2): 36 -38 .
[9] 侯林, 汪雄武, 彭惠娟. 四川丹巴燕子沟金矿床地质特征及控矿因素初探[J]. J4, 2009, 17(6): 23 -28 .
[10] 雷军. 甘肃黄金工业现状及其发展思索[J]. J4, 1998, 6(5-6): 50 -56 .