Evaluation model of rockburst intensity of diversion tunnel based on WOA-SVM

doi:10.16265/j.cnki.issn1003-3033.2023.09.1143

Abstract

Abstract:

This study aims to reduce the occurrence of rockburst accidents in the construction of diversion tunnels. Before construction, rockburst intensity assessment and prevention measures were put forward. Four evaluation indexes were selected as rock burst evaluation indexes in the study, namely, rock uniaxial saturated compressive strength R_c, rock uniaxial tensile strength R_t, maximum tangential stress σ_θ of surrounding rock and rock elastic energy index W_et. According to the research results, 120 groups of rockburst instances were selected as the machine learning sample data, and the WOA-SVM assessment model was proposed. Taking a section of a deep-buried and long diversion tunnel in Xianglushan of the Central Yunnan Water Diversion Project as an example, the intensity assessment results of rockburst intensity were verified. The results show that machine learning can better avoid human factors, and it is completely data-driven with an assessment accuracy of 97.22% for WOA-SVM. By comparison, its assessment accuracy and generalization are better than those of PSO-SVM, GA-SVM and WOA-BP neural network models. The results show that the WOA-SVM model can better capture the link between rockburst levels and indicators for rockburst problems.

Key words: whale optimization algorithm (WOA), support vector machine (SVM), water intake tunnel, rockburst intensity, machine learning

JIN Chunling, JI Zhaotai, GONG Li, AN Xiang, ZHOU Yi. Evaluation model of rockburst intensity of diversion tunnel based on WOA-SVM[J]. China Safety Science Journal, 2023, 33(9): 41-48.

Figures/Tables 15

Tab.1

Tab.2

Fig.1

Fig.2

Fig.3

Fig.4

Fig.5

Tab.3

Fig.6

Fig.7

Tab.4

Fig.8

Fig.9

Tab.5

Tab.6

References 16

[1]	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. Study on the development process of rockbursts[J]. Chinese Journal of Rock Mechanics and Engineering, 2019, 38(4):649-673.
[2]	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
[3]	ZHOU Jian, LI Xibing, MITRI H S. Evaluation method of rockburst: State-of-the-art literature review[J]. Tunnelling and Underground Space Technology, 2018, 81:632-659. doi: 10.1016/j.tust.2018.08.029
[4]	PU Yuanyuan, APEL D B, LIU V, et al. Machine learning methods for rockburst prediction-state-of-the-art review[J]. International Journal of Mining Science and Technology, 2019, 29(4):565-570. doi: 10.1016/j.ijmst.2019.06.009
[5]	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: Science and Technology, 2021, 61(6):543-555.
[6]	靳春玲, 党丹丹, 贡力, 等. IPP-PNN模型在川藏铁路深埋长大隧道岩爆预测中的应用[J]. 铁道科学与工程学报, 2023, 20(3):986-995.
	JIN Chunling, DANG Dandan, GONG Li. Application of IPP-PNN model in rockburst prediction occurring deep-buried long tunnel of Sichuan-Tibet Railway[J]. Journal of Railway Science and Engineering, 2023, 20(3):986-995.
[7]	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
[8]	李宁, 王李管, 贾明涛. 基于粗糙集理论和支持向量机的岩爆预测[J]. 中南大学学报:自然科学版, 2017, 48(5):1268-1275.
	LI Ning, WANG Liguan, JIA Mingtao. Rockburst prediction based on rough set theory and support vector machine[J]. Journal of Central South University: Science and Technology, 2017, 48(5):1268-1275.
[9]	PU Yuanyuan, APEL D B, XU Huawei. Rockburst prediction in kimberlite with unsupervised learning method and support vector classifier[J]. Tunnelling and Underground Space Technology, 2019, 90:12-18. doi: 10.1016/j.tust.2019.04.019
[10]	SEYEDALI M, ANDREW L. The whale optimization algorithm[J]. Advances in Engineering Software, 2016, 95:51-67. doi: 10.1016/j.advengsoft.2016.01.008
[11]	周煦桐. 基于神经网络算法的岩爆预测方法研究[D]. 湘潭: 湘潭大学, 2020.
	ZHOU Xutong. Prediction of rockbursts based on BP neural network[D]. Xiangtan: Xiangtan University, 2020.
[12]	冯磊磊. 基于组合赋权和蝴蝶突变模型的岩爆等级评判[D]. 邯郸: 河北工程大学, 2021.
	FENG Leilei. Evaluation of rock burst grade based on combinatorial weighting and butterfly mutation model[D]. Handan: Hebei University of Engineering, 2021.
[13]	VAPNIK V N. An overview of statistical learning theory[J]. IEEE Transactions on Neural Networks, 1999, 10(5):988-999. doi: 10.1109/72.788640 pmid: 18252602
[14]	丁世飞, 齐丙娟, 谭红艳. 支持向量机理论与算法研究综述[J]. 电子科技大学学报, 2011, 40(1):2-10.
	DING Shifei, QI Bingjuan, TAN Hongyan. An overview on theory and algorithm of support vector machines[J]. Journal of University of Electronic Science and Technology of China, 2011, 40(1):2-10.
[15]	张龙, 彭小明, 熊国良, 等. 基于MSE与PSO-SVM的机车轮对轴承智能诊断方法[J]. 铁道科学与工程学报, 2021, 18(9):2408-2417.
	ZHANG Long, PENG Xiaoming, XIONG Guoliang, et al. Intelligent diagnosis of locomotive wheelset bearings using MSE and PSO-SVM[J]. Journal of Railway Science and Engineering, 2021, 18(9):2408-2417.
[16]	LI Yufang, CHEN Mingnuo, LU Xiaoding, et al. Research on optimized GA-SVM vehicle speed prediction model based on driver-vehicle-road-traffic system[J]. Science China Technological Sciences, 2018, 61(5):782-790. doi: 10.1007/s11431-017-9213-0

岩爆等级	类别	分级依据
岩爆等级	类别	破坏方式	破坏过程	破坏程度
Ⅰ级	无岩爆	—	—	—
II级	轻微岩爆	劈裂成板,剪断脱离母体, 产生射落	洞壁表面局部轻微破坏,噼啪声响	不损坏机械设备, 威胁不大
III级	中级岩爆	劈裂—剪断—弹射向洞壁内发展,形成V形坑。洞壁大范围破坏	类似子弹射击声响,个别情况下损坏设备	生产中断
IV级	强烈岩爆	劈裂-剪断-弹射急速发生,并急剧向洞壁深处扩展,持续时间长	有似闷雷巨响,几乎全断面破坏	人材损失严重, 生产停顿

序号	R_c/MPa	R_t/MPa	σ_θ/MPa	W_et	等级
1	118.46	3.509 9	26.061 2	2.89	Ⅱ
2	52	3.7	7.5	1.3	Ⅰ
3	306.58	13.9	105	6.38	Ⅳ
4	175	7.25	62.5	5	Ⅲ
5	120.05	7.41	90	7.3	Ⅱ
6	180	8.3	48.75	5	Ⅲ
︙	︙	︙	︙	︙	︙
115	225.6	17.2	91.3	7.3	Ⅳ
116	180	8.3	48.75	5	Ⅲ
117	196	8.3	89	5	Ⅲ
118	54.2	12.1	34.15	3.17	Ⅰ
119	52	3.7	7.05	1.3	Ⅰ
120	69.7	4.8	9.57	3.8	Ⅰ

算法	基本参数		待优化参数	参数范围
WOA	最大迭代次数	100	惩罚因子c 核函数参数g	(0.001,100) (0.001,100)
	初始化种群数	10
	训练目标最小误差	0.000 01
PSO	最大迭代次数	100	惩罚因子c	(0.001,100)
	种群数量	20	惩罚因子c	(0.001,100)
	局部学习因子c₁	2	核函数参数g	(0.001,100)
	全局学习因子c₂	2	核函数参数g	(0.001,100)
GA	最大迭代次数	100	惩罚因子c	(0.001,100)
	种群数量	20	惩罚因子c	(0.001,100)
	交叉概率	0.5	核函数参数g	(0.001,100)
	变异概率	0.01	核函数参数g	(0.001,100)

算法	c_b	g_b	最佳适应度/%	准确率/%	训练时长/s
WOA-SVM	0.163 8	0.607 4	99.225 7	97.22	32.762
PSO-SVM	0.241 6	2.136 8	98.347 1	88.88	47.165
GA-SVM	1.263 5	5.384	97.819 8	91.67	63.468

序号	R_c/MPa	R_t/MPa	σ_θ/MPa	W_et	等级
1	82.3	5.6	38.5	2.6	Ⅱ
2	119.472	7.8	28.4	6.71	Ⅲ
3	63.4	2.21	17.24	1.6	Ⅰ
4	90.5	9.7	117.52	4.9	Ⅲ
5	90.16	5.4	58.4	3.4	Ⅱ
6	156.4	8.7	106.4	5.12	Ⅲ