img

QQ群聊

img

官方微信

高级检索

黄金科学技术 ›› 2021, Vol. 29 ›› Issue (6): 826-833.doi: 10.11872/j.issn.1005-2518.2021.06.089

• 采选技术与矿山管理 • 上一篇    下一篇

基于改进迁移学习算法的岩体质量评价模型

胡建华(),郭萌萌(),周坦,张涛   

  1. 中南大学资源与安全工程学院,湖南 长沙 410083
  • 收稿日期:2021-07-07 修回日期:2021-09-21 出版日期:2021-12-31 发布日期:2022-03-07
  • 通讯作者: 郭萌萌 E-mail:hujh21@126.com;gmm0118@163.com
  • 作者简介:胡建华(1975-),男,湖南衡南人,教授,从事高效安全采矿技术与工程稳定性研究工作。hujh21@126.com
  • 基金资助:
    国家自然科学基金项目“深部采动下地质结构体跨尺度时变力学行为试验及机理”(41672298)

Rock Mass Quality Evaluation Model Based on Improved Transfer Learning Algorithm

Jianhua HU(),Mengmeng GUO(),Tan ZHOU,Tao ZHANG   

  1. School of Resources and Safety Engineering,Central South University,Changsha 410083,Hunan,China
  • Received:2021-07-07 Revised:2021-09-21 Online:2021-12-31 Published:2022-03-07
  • Contact: Mengmeng GUO E-mail:hujh21@126.com;gmm0118@163.com

摘要:

岩体质量分级是进行工程设计和施工的基础。通过搜集不同地区55组实测样本和17组插值样本建立案例库,考虑岩体的复杂不确定性和异地岩体的差异性,在案例库基础上提出了一种改进两阶段回归迁移学习(Two-stage TrAdaBoost.R2)—孤立森林(Isolated Forest)多因素岩体质量等级预测模型。将广州抽水蓄能电站第1期地下工程的12个样本用于模型测试,结果表明:(1)迁移学习可以通过权重调整选出与目标区域岩体相似的样本,解决了传统机器学习方法中同区域训练样本数量不足的问题。(2)孤立森林算法与迁移学习相结合可以排除异常数据的影响,增加模型的稳定性。(3)利用训练好的模型对12个测试样本进行多次判定,结果与实际情况基本相符,验证了模型的有效性。

关键词: 岩石力学, 岩体质量评价, 机器学习, 迁移学习, 孤立森林, TrAdaBoost算法

Abstract:

Rock mass quality classification is an important foundation for engineering design and construction, and it is also an important research topic at present. Taking into account the complexity and uncertainty of rock masses and the differences of rock masses in different regions, machine learning methods are widely used in rock mass quality evaluation. A case database was established by collecting 55 sets of measured samples and 17 sets of interpolated samples from different regions. RQD, uniaxial saturated compressive strength (Rw), rock mass integrity coefficient (Kv), structural plane strength coefficient (Kf), groundwater seepage volume (ω) are determined as the input conditions of the model, and the rock mass quality grade is the output condition. Based on the case library, an improved two-stage regression migration learning (Two-stage TrAdaBoost.R2)-Isolated Forest multi-factor rock mass quality grade prediction model is proposed. The advantages of this model are of follows: (1) The idea of migration learning is introduced into the rock mass quality classification. Taking into account the differences of rock masses in different regions, using the idea of weight adjustment, a sample similar to the target rock mass is selected from the known samples to assist in the training of the model. Solved the problem of insufficient training samples, and achieve high-precision prediction of the model when there are fewer learning samples in the target field. (2) When using the migration algorithm to classify the quality of the rock mass, the classification problem is transformed into a regression problem. The regression algorithm is used to predict the quality of the rock mass. Only one model can be used to judge the multiple levels of the sample, which overcomes the limitation of the classification algorithm in solving the multi-classification problem. (3) The sample weight is adjusted in two stages, which solves the problem of the source domain weight falling too fast in the TrAdaBoost algorithm. (4) Combined the Two-stage TrAdaBoost.R2 algorithm with the Isolated Forest anomaly detection algorithm,the influence of abnormal data on the model is eliminated, and the stability of the model is increased. The trained model was used to make multiple judgments on 12 samples of the first phase underground project of Guangzhou Pumped Storage Power Station, and the prediction accuracy of the model was evaluated by the mean square error. The average mean square error of the test sample is 0.067, and the prediction accuracy is high. It proves that the model has good performance in the application of rock mass quality grade prediction.

Key words: rock mechanics, rock mass quality evaluation, machine learning, transfer learning, Isolated Forest, TrAdaBoost algorithm

中图分类号: 

  • TU457

图1

TrAdaBoost算法权重更新机制"

图2

多分类方法"

图3

孤立森林算法原理"

图4

两阶段迁移学习算法流程"

表1

岩体质量分级标准"

类别RQD/%Rw/MPaKvKfω/[L·(min·10m)-1
90~100200~1201.00~0.751.0~0.80~5
75~90120~600.75~0.450.8~0.65~10
50~7560~300.45~0.300.6~0.410~25
25~5030~150.30~0.200.4~0.225~125
0~2515~00.20~0.000.2~0.0125~300

表2

训练样本"

序号RQD/%Rw/MPaKvKfω/[L·(min·10m)-1类别
1*100.0200.01.001.000.0
2*97.5180.00.940.951.3
3*95.0160.00.880.902.5
492.5140.00.810.853.8
586.3105.00.680.756.3
682.590.00.600.707.5
778.875.00.530.658.8
868.852.50.410.5513.8
962.545.00.380.5017.5
1056.337.50.340.4521.3
1143.826.30.280.3550.0
1237.522.50.250.3075.0
1331.318.80.230.25100.0
14*18.811.30.150.15168.8
15*12.57.50.100.10212.5
16*6.33.80.050.05256.3
17*0.00.00.000.00300.0
1882.095.00.700.3520.0
1968.090.00.570.3520.0
2040.025.00.220.3520.0
2187.095.00.700.5010.0
2276.090.00.570.5010.0
2376.095.00.700.5010.0
2472.090.00.570.5010.0
2551.040.00.380.5010.0
2652.025.00.220.5010.0
2768.090.00.380.3020.0
2828.040.00.320.3020.0
2951.025.00.150.3020.0
3075.095.00.700.500.0
3177.590.00.570.4510.0
3275.590.00.450.528.0
3385.594.00.650.550.0
3485.093.00.600.500.0
3578.592.00.550.506.0
3680.095.00.500.450.0
3785.092.00.700.5010.0
3878.080.00.750.500.0
3976.590.00.550.5010.0
4085.095.00.650.500.0
4175.090.00.550.507.0
4275.090.00.550.5010.0
4387.095.00.500.450.0
4482.096.00.750.350.0
4550.070.00.500.355.0
4650.626.00.260.3520.0
4750.040.20.500.5010.0
4852.025.00.200.505.0
4971.090.00.350.305.0
5050.934.00.320.3521.0
5150.090.00.500.255.0
5230.270.00.400.2010.0
5350.045.00.120.305.0
5451.035.00.320.3515.0
5550.934.00.320.3520.0
5650.045.00.150.355.0
5726.036.00.220.355.0
5831.520.00.230.2546.0
5935.070.50.350.3010.0
6031.520.00.230.2550.0

表3

测试样本"

序号RQD/%Rw/MPaKvKfω/[L·(min·10m)-1实测等级
171.890.10.570.450
276.095.00.700.5512.0
387.095.00.700.509.8
482.095.00.700.350
576.090.00.570.5011.0
668.090.00.570.3518.5
751.040.20.380.5510.5
850.035.00.320.3520.0
968.090.00.380.3821.0
1051.045.00.150.305.0
1152.025.00.220.5212.0
1228400.320.3018.5

表4

测试结果"

序号期望输出实际输出
12345678910
均方误差0.0830.1670.167000.083000.0830.083
123*3*2222222
22222222222
32222222222
42222222222
52222222222
63*3*3*223*223*3*
73333333333
83333333333
93333333333
103333333333
113333333333
124444444444
null Barton N,2002.Some new Q-value correlations to assist in site characterization and tunnel design[J].International Journal of Rock Mechanics and Mining Sciences,39(2):185-216.
null Barton N, L?set F, Lien R, al et,1981.Application of Q-System in design decisions concerning dimensions and appropriate support for underground installations[J].Subsurface Space,(12):553-561.
null Bieniawski ZT,1978.Determining rock mass deformability:Experience from case histories[J].International Journal of Rock Mechanics and Mining Sciences & Geomechanics Abstracts,15(5):237-347.
null Cai Guangkui,2001.Research on BP Network Model of Surrounding Rock Stability Classification[D].Nanjing:Hohai University.
null Chen Xing,2018. Application of fuzzy pattern recognition and LVQ neural network in the rock mass classification[J].Pearl River,39(10):1-6.
null Dai W Y, Yang Q, Xue G R, al et,2007.Boosting for transfer learning[C]//Proceedings of the 24th International Conference on Machine Learning.Corvalis:Oregon State University: 193-200.DOI:https://doi.org/10.1145/1273496.1273521
null Gong Fengqiang, Li Xibing,2007.Application of distance discriminant analysis method to classification of engineering quality of rock masses[J].Chinese Journal of Rock Mechanics and Engineering,26(1):190-194.
null He Yunsong, Xue Qiuchi, Zhao Qihua,2017.Improved support vector machine-based study on rock mass quality classification[J].Water Resources and Hydropower Engineering,48(1):133-138.
null Hu Jianhua, Shang Junlong, Lei Tao,2012.Rock mass quality evaluation of underground engineering based on RS-TOPSIS method[J].Journal of Central South University(Science and Technology),43(11):4412-4419.
null Liu F T, Ting K M, Zhou Z H,2012.Isolation-based anomaly detection[J].ACM Transactions on Knowledge Discovery from Data,6(1):1-39.
null Liu Wanjun, Li Tianhui, Qu Haicheng,2018.Hyperspectral similar sample classification algorithm based on Fisher criterion and TrAadboost[J].Remote Sensing for Land and Resources,30(4):41-48.
null Pardoe D, Stone P,2010. Boosting for regression transfer[C]//Proceedings of the 27th International Conference on Machine Learning.Haifa:International Machine Learning Society: 863-870.
null Qin Y G, Luo Z Q, Wang J, al et,2019.Evaluation of goaf stability based on transfer learning theory of artificial intelligence[J].IEEE Access,7:18862440.DOI:10.1109/ACCESS.2019.2929533
null Tu W F, Li L P, Li S C, al et,2019.Research on the application of dynamic weighting on the rock mass quality rating[J].Arabian Journal of Geosciences,12.DOI:https://doi.org/10.1007/s12517-019-4264-9
null Water Resources of the People’s Republic of China,2014. Standard for engineering classification of rock masses: [S].Beijing:China Planning Press.
null Wen Changping,2008.Classification of rock-mass stability based on attributive mathematical theory[J].Journal of Hydroelectric Engineering,27(3):75-80.
null Xu Guizhi, Lin Fang, Gong Minghong, al et,2019.A TrAdaBoost-based method for detecting multiple subjects’ P300 potentials[J].Journal of Biomedical Engineering,36(4):531-540.
null Yang Zhaohui, Liu Haowu,1999.Artificial neural network model for the stability classification of adjoining rock of underground construction[J].Journal of Sichuan Union University(Engineering Science Edition),(4):66-72.
null Yao Yinpei, Li Xibing, Gong Fengqiang, al et,2010.Application of weighted Mahalanobis distance discriminant analysis method to classification of rock mass quality[J].Chinese Journal of Rock Mechanics and Engineering,29(Supp.2):4119-4123.
null Yuan Guohong, Chen Jianping, Ma Lin,2005.Application of extenics in evaluating of engineering quality of rock masses[J]. Chinese Journal of Rock Mechanics and Engineering,(9):1539-1544.
null Zhang Biao, Dai Xingguo,2017.A classification model of rock mass based on finite interval cloud model and distance discrimination weighting[J].Hydrogeology & Engineering Geology,44(5):150-157.
null Zheng S, Jiang A N, Yang X R, al et,2020.A new reliability rock mass classification method based on least squares support vector machine pptimized by bacterial foraging optimization algorithm[J].Advances in Civil Engineering,(1):1-13.
null Zhou Shuda, Pei Qitao, Ding Xiuli,2016.Application of grey evaluation model based on classification degree and weight of classification of index to rock mass quality evaluation of underground engineering[J].Chinese Journal of Rock Mechanics and Engineering,35(Supp.2):3671-3679.
null Zhou Tan, Hu Jianhua, Kuang Ye,2019.Rock mass quality evaluation method and application based on fuzzy RES-multidimensional cloud model[J].The Chinese Journal of Nonferrous Metals,29(8):1771-1780.
null 蔡广奎,2001.围岩稳定性分类的BP网络模型研究[D].南京:河海大学.
null 陈星,2018.模糊模式识别法及LVQ神经网络在岩体质量分级中的应用研究[J].人民珠江,39(10):1-6.
null 宫凤强,李夕兵,2007.距离判别分析法在岩体质量等级分类中的应用[J].岩石力学与工程学报,26(1):190-194.
null 何云松,薛秋池,赵其华,2017.基于改进向量机的岩体质量分级研究[J].水利水电技术,48(1):133-138.
null 胡建华,尚俊龙,雷涛,2012.基于RS-TOPSIS法的地下工程岩体质量评价[J].中南大学学报(自然科学版),43(11):4412-4419.
null 刘万军,李天慧,曲海成,2018.基于Fisher准则和TrAdaboost的高光谱相似样本分类算法[J].国土资源遥感,30(4):41-48.
null 文畅平,2008.基于属性数学理论的岩体质量分级方法[J].水力发电学报,27(3):75-80.
null 徐桂芝,林放,宫铭鸿,等,2019.基于TrAdaBoost的跨脑辨识P300电位研究[J].生物医学工程学杂志,36(4):531-540.
null 杨朝晖,刘浩吾,1999.地下工程围岩稳定性分类的人工神经网络模型[J].四川联合大学学报(工程科学版),(4):66-72.
null 姚银佩,李夕兵,宫凤强,等,2010.加权距离判别分析法在岩体质量等级分类中的应用[J].岩石力学与工程学报,29(增2):4119-4123.
null 原国红,陈剑平,马琳,2005.可拓评判方法在岩体质量分类中的应用[J].岩石力学与工程学报,(9):1539-1544.
null 张彪,戴兴国,2017.基于有限区间云模型和距离判别赋权的岩体质量分类模型[J].水文地质工程地质,44(5):150-157.
null 中华人民共和国水利部,2014. 工程岩体分级标准: [S].北京:中国计划出版社.
null 周述达,裴启涛,丁秀丽,2016.改进分类区分度及权重的岩体质量评价灰评估模型及应用[J].岩石力学与工程学报,35(增2):3671-3679.
null 周坦,胡建华,匡也,2019.基于模糊RES-多维云模型的岩体质量评判方法与应用[J].中国有色金属学报,29(8):1771-1780.
[1] 李地元, 杨博, 刘子达, 刘永平, 赵君杰. 基于集成树算法的岩石黏聚力和内摩擦角预测方法[J]. 黄金科学技术, 2024, 32(5): 847-859.
[2] 李振阳, 张宝岗, 熊信, 杨承业, 白玉奇. 基于PSO-XGBoost的露天矿山PPV预测模型研究[J]. 黄金科学技术, 2024, 32(4): 620-630.
[3] 姜志宏, 陈澳. 融合全监督学习的半监督矿石粒度预测算法[J]. 黄金科学技术, 2024, 32(3): 539-547.
[4] 赵国彦, 胡凯译, 李洋, 刘雷磊, 王猛. 基于BWO-RF模型的岩体质量评价方法[J]. 黄金科学技术, 2024, 32(2): 270-279.
[5] 凡兴禹, 王雪林. 基于改进XGBoost算法的深部巷道松动圈智能预测研究[J]. 黄金科学技术, 2024, 32(1): 109-122.
[6] 凡奥奇, 王万禄, 李树建, 张斌, 刘映辉, 吴浩. 基于优化组合赋权的可拓学磷矿山岩体质量评价[J]. 黄金科学技术, 2024, 32(1): 132-143.
[7] 许方颖, 邹艳红, 易卓炜, 杨福强, 毛先成. 基于非均衡数据的ADASYN-CatBoost测井岩性智能识别——以胶西北招贤金矿床为例[J]. 黄金科学技术, 2023, 31(5): 721-735.
[8] 赵亚楠, 赵一航, 蒋中明, 赵红敏. 基于离散元法的高放核废料储罐静动力稳定性初步研究[J]. 黄金科学技术, 2023, 31(4): 592-604.
[9] 徐先锋,邢鹏飞,王岁红,汪泳. 基于博弈论G1-EW-TOPSIS法的岩体质量评价和应用[J]. 黄金科学技术, 2022, 30(5): 704-712.
[10] 曾强,黄小荣,王晓军,陈青林,刘健,龚囱. 不同埋深灰岩岩爆倾向性及声发射特征试验研究[J]. 黄金科学技术, 2021, 29(6): 863-873.
[11] 赵国彦,党成凯,刘焕新,刘洋,肖屈日,李洋,陈立强,毛文杰. 基于距离判别分析的矿山岩爆倾向性评价[J]. 黄金科学技术, 2021, 29(5): 690-697.
[12] 王玺,马春德,刘兴全,姜明伟,范玉赟. 滨海矿区地应力与岩石力学参数随埋深的变化规律及其相互关系[J]. 黄金科学技术, 2021, 29(4): 535-544.
[13] 胡建华,董喆喆,马少维,秦亚光,徐晓,代转. 应力—渗流耦合作用下损伤岩石渗流特性[J]. 黄金科学技术, 2021, 29(3): 355-363.
[14] 贾敬锎,黄滚,汪龙,成墙,甄利兵. 单轴压缩试验中减弱端部效应新型方法研究[J]. 黄金科学技术, 2021, 29(3): 382-391.
[15] 黄进,刘科伟,靳绍虎. 高强弹体侵彻白麻花岗岩靶体的数值模拟研究[J]. 黄金科学技术, 2021, 29(3): 411-420.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!