基于改进Smote-GBDT算法的岩爆预测模型

doi:10.16265/j.cnki.issn1003-3033.2023.09.0850

中国安全科学学报 ›› 2023, Vol. 33 ›› Issue (9): 25-32.doi: 10.16265/j.cnki.issn1003-3033.2023.09.0850

基于改进Smote-GBDT算法的岩爆预测模型

宋英华¹^,²(), 江晨², 李墨潇¹^,²^,^**(), 齐石²

¹ 武汉理工大学中国应急管理研究中心,湖北武汉 430070
² 武汉理工大学安全科学与应急管理学院,湖北武汉 430070

收稿日期:2023-03-12 修回日期:2023-06-14 出版日期:2023-09-28
通讯作者:
**李墨潇 (1987—),男,安徽安庆人,博士,副教授,硕士生导师,主要从事地下空间安全、矿山安全、应急管理等方面的研究。E-mail:lmx@whut.edu.cn。
作者简介:
宋英华 (1962—),男,湖北武汉人,博士,教授,博士生导师,主要从事科研管理、应急管理、项目管理等方面的研究。E-mail:song6688cccc@163.com。
基金资助:
国家自然科学基金资助(52209146)

Rockburst prediction model based on improved Smote-GBDT algorithm

SONG Yinghua¹^,²(), JIANG Chen², LI Moxiao¹^,²^,^**(), QI Shi²

¹ China Research Center for Emergency Management, Wuhan University of Technology, Wuhan Hubei 430070, China
² School of Safety Science and Emergency Management, Wuhan University of Technology, Wuhan Hubei 430070, China

Received:2023-03-12 Revised:2023-06-14 Published:2023-09-28

摘要/Abstract

摘要：

为准确预测岩爆等级,确保施工人员和设备安全,首先,从岩爆机制、数据和算法角度,分析埋深(D)、单轴抗压强度(UCS)、单轴抗拉强度(UTS)、岩石脆性指数(B₁、B₂)、围岩最大切向应力(MTS)、应力集中系数(SCF)和弹性变形能指数(W_et) 8个指标,建立岩爆预测指标体系;其次,针对岩爆样本存在的数据不均衡问题,引进托梅克联系(Tomek Link)对欠采样方法,改进合成少数类过采样(Smote)算法,对岩爆训练样本进行混合过采样;最后,构建SmoteTomek-梯度提升树(GBDT)岩爆预测模型,以38组数据验证模型的有效性,并与其他模型进行对比。结果表明:SmoteTomek-GBDT的准确率为92.1%,较未采样提升5.3%,Smote采样提升10.5%,优于随机过采样模型,并且避免跨等级的岩爆误判。

关键词: 岩爆预测, 梯度提升树(GBDT)算法, 合成少数类过采样(Smote)算法, 岩爆指标, 托梅克联系(Tomek Link)

Abstract:

This paper aims to accurately predict rockburst levels and ensure the safety of construction personnel and equipment. First, from the perspective of rock burst mechanism, eight indicators of burial depth (D), uniaxial compressive strength (UCS), uniaxial tensile strength (UTS), rock brittleness index (B₁、B₂), maximum tangential stress (MTS), stress concentration coefficient (SCF) and elastic deformation energy index (W_et) were analyzed, and a rock burst prediction index system was established. Secondly, to address the problem of data imbalance in rockburst samples, the Tomek Link of under sampling method was introduced to improve the (Smote) for mixed oversampling of rockburst training samples. Finally, the SmoteTomek-GBDT rockburst prediction model was constructed, and the validity of the model was verified with 38 sets of data and compared with other models. The results show that the accuracy of SmoteTomek-GBDT is 92.1%, and it is a 5.3% improvement over unsampled and 10.5% improvement over Smote sampled, which is better than the random oversampling model, and avoids cross-grade rockburst misclassification, which is of some significance for accurate rockburst prediction.

Key words: rockburst prediction, gradient boosting decision tree (GBDT), synthetic minority oversampling technique (Smote), rockburst index, Tomek Link

宋英华, 江晨, 李墨潇, 齐石. 基于改进Smote-GBDT算法的岩爆预测模型[J]. 中国安全科学学报, 2023, 33(9): 25-32.

SONG Yinghua, JIANG Chen, LI Moxiao, QI Shi. Rockburst prediction model based on improved Smote-GBDT algorithm[J]. China Safety Science Journal, 2023, 33(9): 25-32.

图/表 9

图1

图2

图3

表1

图4

表2

表3

图5

表4

参考文献 16

[1]	胡建华, 黄鹏莅, 周坦, 等. 岩爆倾向性的改进有限云评价模型与工程应用[J]. 中国安全科学学报, 2022, 32(2):90-98. doi: 10.16265/j.cnki.issn1003-3033.2022.02.013
	HU Jianhua, HUANG Pengli, ZHOU Tan, et al. Evaluation and engineering application of an improved finite cloud model for rockburst proneness[J]. China Safety Science Journal, 2022, 32(2):90-98. doi: 10.16265/j.cnki.issn1003-3033.2022.02.013
[2]	张凯, 张科, 李昆. 主元分析-神经网络岩爆等级预测模型[J]. 中国安全科学学报, 2021, 31(3):96-104. doi: 10.16265/j.cnki.issn1003-3033.2021.03.014
	ZHANG Kai, ZHANG Ke, LI Kun. Prediction model of rockburst grade based on PCA-neural network[J]. China Safety Science Journal, 2021, 31(3):96-104. doi: 10.16265/j.cnki.issn1003-3033.2021.03.014
[3]	吴顺川, 张晨曦, 成子桥. 基于PCA-PNN原理的岩爆烈度分级预测方法[J]. 煤炭学报, 2019, 44(9):2767-2776.
	WU Shunchuan, ZHANG Chenxi, CHENG Ziqiao. Prediction of intensity classification of rockburst based on PCA-PNN principle[J]. Journal of China Coal Society, 2019, 44(9):2767-2776.
[4]	许瑞, 侯奎奎, 王玺, 等. 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.
	XU Rui, HOU Kuikui, WANG Xi, et al. Combined prediction model of rockburst intensity based on kernel principal component analysis and SVM[J]. Gold Science and Technology, 2020, 28(4):575-584.
[5]	郭延华, 赵帅. 基于KPCA-WOA-KELM的岩爆烈度预测[J]. 河北工程大学学报:自然科学版, 2021, 38(2):1-7.
	GUO Yanhua, ZHAO Shuai. Classified prediction model of rockburst using KPCA-WOA-KELM[J]. Journal of Hebei University of Engineering: Natural Science Edition, 2021, 38(2):1-7.
[6]	谭文侃, 叶义成, 胡南燕, 等. LOF与改进SMOTE算法组合的强烈岩爆预测[J]. 岩石力学与工程学报, 2021, 40(6):1186-1194.
	TAN Wenkan, YE Yicheng, HU Nanyan, et al. Severe rockburst prediction based on combination of LOF and improved SMOTE algorithm[J]. Chinese Journal of Rock Mechanics and Engineering, 2021, 40(6):1186-1194.
[7]	汤志立, 王雪, 徐千军. 基于过采样和客观赋权法的岩爆预测[J]. 清华大学学报:自然科学版, 2021, 61(6):543-555.
	TANG Zhili, WANG Xue, XU Qianjun. Rockburst prediction based on oversampling and objective weighting method[J]. Journal of Tsinghua University:Natural Science Edition, 2021, 61(6):543-555.
[8]	CHAWLA V, KEVIN W, LAWRENCE O, et al. SMOTE: synthetic minority over-sampling technique[J]. Journal of Artificial Intelligence Research, 2002, 16:321-357. doi: 10.1613/jair.953
[9]	LIU Jingjing, LIU Jianchao. An intelligent approach for reservoir quality evaluation in tight sandstone reservoir using gradient boosting decision tree algorithm: a case study of the Yanchang Formation, mid-eastern Ordos Basin, China[J]. Marine and Petroleum Geology, 2021,126: DOI: 10.1016/j.marpetgeo.2021.104939.
[10]	ZHANG Chuanqing, ZHOU Hui, FENG Xiating. An index for estimating the stability of brittle surrounding rock mass: FAI and its engineering application[J]. Rock Mechanics and Rock Engineering, 2011, 44(4):401-414. doi: 10.1007/s00603-011-0150-9
[11]	冯夏庭, 肖亚勋, 丰光亮, 等. 岩爆孕育过程研究[J]. 岩石力学与工程学报, 2019, 38(4):649-673.
	FENG Xiating, XIAO Yaxun, FENG Guangliang, et al. Research on rockburst inoculation process[J]. Chinese Journal of Rock Mechanics and Engineering, 2019, 38(4):649-673.
[12]	ZHANG Chuanqing, ZHOU Hui, FENG Xiating. An index for estimating the stability of brittle surrounding rock mass: FAI and its engineering application[J]. Rock Mechanics and Rock Engineering, 2011, 44(4):401-414. doi: 10.1007/s00603-011-0150-9
[13]	杨金林, 李夕兵, 周子龙, 等. 基于粗糙集理论的岩爆预测模糊综合评价[J]. 金属矿山, 2010(6):26-29.
	YANG Jinlin, LI Xibing, ZHOU Zilong, et al. A fuzzy assessment method of rockburst prediction based on rough set theory[J]. Metal Mine, 2010 (6):26-29.
[14]	刘焕新, 谭卓英, 王玺, 等. 新城金矿深竖井开挖岩爆危险性预测[J]. 中国矿业大学学报, 2020, 49(2):296-304.
	LIU Huanxin, TAN Zhuoying, WANG Xi. et al. Prediction of rockburst risk in deep shaft excavation of Xincheng gold mine[J]. Journal of China University of Mining and Technology, 2020, 49(2):296-304.
[15]	刘冉. 岩爆分级预测的多维正态云模型[D]. 武汉: 武汉科技大学, 2020.
	LIU Ran. Multi-dimensional normal cloud model for rockburst classification prediction[D]. Wuhan: Wuhan University of Science and Technology, 2020.
[16]	田睿, 孟海东, 陈世江, 等. RF-AHP-云模型下岩爆烈度分级预测模型[J]. 中国安全科学学报, 2020, 30(7):166-172. doi: 10.16265/j.cnki.issn1003-3033.2020.07.025
	TIAN Rui, MENG Haidong, CHEN Shijiang, et al. Prediction model of rockburst intensity classification based on RF-AHP-Cloud model[J]. China Safety Science Journal, 2020, 30(7):166-172. doi: 10.16265/j.cnki.issn1003-3033.2020.07.025

序号	工程名称	D/m	B₁	MTS/MPa	SCF	W_et	岩爆等级
1^*	安禄隧道 S3	373	64.31	16.70	0.20	6.53	Ⅲ
2	安禄隧道 S4	373	66.75	17.35	0.20	3.22	Ⅲ
3^*	马路坪矿 K2(600m)	600	6.67	4.60	0.23	1.39	Ⅰ
4	马路坪矿 K2(700m)	700	6.67	3.80	0.19	1.39	Ⅰ
5	马路坪矿 K2(750m)	750	6.67	2.60	0.13	1.39	Ⅰ
6	江边水电站引 4+768	827	17.54	66.77	0.45	5.08	Ⅱ
7^*	江边水电站引 7+366	1 140	47.97	18.32	0.19	1.87	Ⅰ
8^*	安禄隧道 S1	370	78.69	17.39	0.17	6.58	Ⅲ
︙	︙	︙	︙	︙	︙	︙	︙
187^*	金川二矿区 K6	1 000	13.13	60.00	0.44	2.12	Ⅱ
188	二滩水电站 2 号支洞	194	29.73	90.00	0.41	7.30	Ⅱ

统计量	D	B₁	MTS	SCF	W_et
极大值	2 372	80.0	297.8	3.87	15.6
极小值	119.0	4.48	2.60	0.11	0.70
均值	751.8	22.9	51.7	0.60	4.44
方差	339.8	16.1	37.8	0.65	2.78

GBDT模型参数	参数范围	优化步长	参数值
学习率	[0,1]	0.05	0.49
决策树数量	[80,300]	5	213
决策树分枝最小样本数	[0,1]	0.02	0.08
叶子节点的最小样本数	[0,50]	2	13
决策树深度	[1,20]	1	7
最大叶子节点数	[2,50]	2	19

岩爆预测模型	准确率	等级标签	精确率	召回率	F₁值	岩爆预测模型	准确率	等级标签	精确率	召回率	F₁值
GBDT	0.868	Ⅰ	1.00	1.00	1.00	SmoteTomek-SVM	0.816	Ⅰ	0.82	1.00	0.90
		Ⅱ	0.80	0.57	0.67			Ⅱ	0.75	0.86	0.80
		Ⅲ	0.76	0.93	0.84			Ⅲ	0.92	0.79	0.85
		Ⅳ	1.00	0.88	0.93			Ⅳ	0.86	0.76	0.8
Smote-GBDT	0.816	Ⅰ	0.89	0.89	0.89	SmoteTomek-RF	0.842	Ⅰ	0.89	0.89	0.89
		Ⅱ	0.62	0.71	0.67			Ⅱ	0.67	0.86	0.75
		Ⅲ	0.85	0.79	0.81			Ⅲ	0.92	0.79	0.85
		Ⅳ	0.88	0.88	0.88			Ⅳ	0.88	0.88	0.88
SmoteTomek-GBDT	0.921	Ⅰ	1.00	0.89	0.94	ROS-GBDT	0.895	Ⅰ	1.00	0.78	0.88
		Ⅱ	0.78	1.00	0.88			Ⅱ	0.78	1.00	0.88
		Ⅲ	0.93	0.93	0.93			Ⅲ	0.87	0.93	0.90
		Ⅳ	1.00	0.88	0.93			Ⅳ	1.00	0.88	0.93

基于改进Smote-GBDT算法的岩爆预测模型

Rockburst prediction model based on improved Smote-GBDT algorithm

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 9

参考文献 16

相关文章 3

编辑推荐

Metrics

本文评价

[1]	周英豪, 王文杰, 卢西洲, 王康伟. 岩爆灾害博弈论组合赋权预测模型及应用[J]. 中国安全科学学报, 2022, 32(7): 105-112.
[2]	张凯, 张科, 李昆. 主元分析-神经网络岩爆等级预测模型[J]. 中国安全科学学报, 2021, 31(3): 96-104.
[3]	黄建, 夏元友, 吝曼卿. 基于改进组合赋权的岩爆多维云模型预测研究[J]. 中国安全科学学报, 2019, 29(7): 26-32.