基于IWOA−SVM的煤矿突水预测模型

秋兴国; 李靖

doi:10.13272/j.issn.1671-251x.2021050043

基于IWOA−SVM的煤矿突水预测模型

doi: 10.13272/j.issn.1671-251x.2021050043

秋兴国,
李靖

西安科技大学计算机科学与技术学院, 陕西西安　710600

基金项目: 陕西省自然科学基础研究资助项目(2019JM-348)。

详细信息

作者简介:
秋兴国(1964－)，男，陕西乾县人，教授，研究方向为计算机监测与控制、智能信息处理、煤矿灾害监测预报，E-mail: sxxykj@126.com

中图分类号: TD745
计量
- 文章访问数: 153
- HTML全文浏览量: 86
- PDF下载量: 18
- 被引次数: 0
出版历程
- 收稿日期: 2021-05-18
- 修回日期: 2022-01-07
- 刊出日期: 2022-01-20

Prediction model of water inrush in coal mine based on IWOA-SVM

College of Computer Science and Technology, Xi'an University of Science and Technology, Xi'an 710600, China

摘要

摘要: 针对传统煤矿突水预测算法易陷入局部最优、预测结果准确率低及速度慢等问题，提出一种基于改进鲸鱼优化算法(IWOA)−支持向量机(SVM)的煤矿突水预测模型。IWOA从鲸鱼种群初始化、调节因子非线性化及随机差分进化(DE)3个方面入手对鲸鱼优化算法(WOA)进行改进：使用Tent映射初始化鲸鱼种群，提高鲸鱼种群寻找到最优猎物的可能性；通过调节因子非线性变化策略，提升算法在迭代前期的全局搜索能力及迭代后期的局部搜索能力，从而加快收敛速度；引入DE算法的变异、交叉、选择操作，以增强WOA的全局搜索能力。利用IWOA对SVM模型进行参数优化，将影响煤矿突水的水压、隔水层厚度、煤层倾角、断层落差、断层与工作面距离、采高共6个影响因素作为模型的输入特征向量，突水与安全2种突水结果作为模型的输出向量，以突水预测结果与实际结果间的误差最小化为目标建立目标函数，得到基于IWOA−SVM的煤矿突水预测模型。实验结果表明：与粒子群优化算法、DE算法、WOA相比，IWOA的预测准确率最高，标准误差最小，且收敛速度快，鲁棒性好；IWOA−SVM的突水预测准确率达到100%，与传统的突水系数法、SVM、WOA−SVM相比，IWOA−SVM表现出更高的准确率和稳定性。
- 煤矿突水预测 /
- 鲸鱼优化 /
- 支持向量机 /
- 差分进化 /
- 混沌映射
Abstract: The traditional prediction algorithm of water inrush in coal mine is easy to fall into local optimum, the prediction results accuracy is low and the speed is slow. In order to solve the above problems, a prediction model of water inrush in coal mine based on improved whale optimization algorithm (IWOA) and support vector machine (SVM) is proposed. IWOA improves the whale optimization algorithm (WOA) from three aspects, whale population initialization, nonlinear adjustment factor and random differential evolution (DE). Tent mapping is used to initialize the whale population to improve the possibility of the whale population finding the optimal prey. The non-linear change strategy of the adjustment factor is applied to improve the global search capability of the algorithm in the early stage of the iteration and the local search capability in the later stage of the iteration so as to speed up the convergence speed. The mutation, crossover and selection operations of DE algorithm are introduced to enhance the global search capability of WOA. The parameters of SVM model are optimized by IWOA. The six factors affecting water inrush in coal mine, including water pressure, thickness of aquiclude, dip angle of coal seam, fault drop, distance between fault and working face and mining height are taken as the input characteristic vectors of the model. The two water inrush results of water inrush and safety are taken as the output vectors. The objective function is established to minimize the error between the water inrush prediction results and the actual results, and the coal mine water inrush prediction model based on IWOA−SVM is obtained. The experimental results show that IWOA has the highest prediction accuracy, minimum standard error, fast convergence and good robustness compared with particle swarm optimization, DE algorithm and WOA. The accuracy of water inrush prediction of IWOA−SVM is 100%. Compared with the traditional water inrush coefficient method, SVM and WOA−SVM, IWOA−SVM shows higher accuracy and stability.
- prediction of water inrush in coal mine /
- whale optimization /
- support vector machine /
- differential evolution /
- chaotic mapping

HTML全文

图 1 基于IWOA−SVM的煤矿突水预测流程

Figure 1. Water inrush prediction flow based on IWOA−SVM

下载: 全尺寸图片幻灯片

图 2 4种算法的收敛曲线

Figure 2. Convergence curves of four algorithms

下载: 全尺寸图片幻灯片

图 3 IWOA−SVM预测结果

Figure 3. IWOA−SVM prediction results

下载: 全尺寸图片幻灯片

表 1 归一化处理后的部分突水数据

Table 1. Partial water inrush data after normalization

样本编号	水压z₁	隔水层厚度z₂	煤层倾角z₃	断层落差z₄	断层与工作面距离z₅	采高z₆	突水状态
1	0.383 4	0.199 4	0.466 7	0.080 0	0.123 1	0	1
2	0.339 4	0.188 5	0.800 0	1.000 0	0.692 3	0.111 1	1
7	0.256 4	0.400 7	0.133 3	0.016 0	0.476 9	0.006 9	0
8	0.172 3	0.258 0	0.533 3	0.064 0	0.061 5	0.090 3	1
12	0	0.325 5	0.066 7	0.440 0	0.046 2	0.013 9	0
19	0.158 0	0.209 4	0.066 7	0.020 0	0.423 1	0.090 3	0
25	0.497 4	0.453 5	0.333 3	0.035 0	0.061 5	0.152 8	1
32	0.196 9	0.070 6	0.400 0	0.300 0	0.123 1	0.111 1	1
34	0.717 6	0.560 4	0.400 0	0.070 0	0.092 3	0.251 4	1
42	0.300 5	0.451 5	0.566 7	0.024 0	0.053 8	0.097 2	0
43	0.373 1	0.451 5	0.600 0	0.220 0	0.276 9	0.097 2	0

下载: 导出CSV

表 2 4种算法的寻优结果

Table 2. Optimizing results of four algorithms

算法	寻优结果		准确率/%	标准误差	收敛代数
算法	c	σ	准确率/%	标准误差	收敛代数
PSO	4.358	0.010	60.47	0.489	100
DE	7.208	0.325	83.72	0.386	29
WOA	29.136	0.066	81.39	0.408	43
IWOA	6.517	0.418	88.37	0.333	8

下载: 导出CSV

表 3 不同方法预测结果比较

Table 3. Comparison of prediction results of different methods

序号	实际结果	预测结果
序号	实际结果	突水系数法	SVM	WOA−SVM	IWOA−SVM
44	0	0	1	0	0
45	1	0	1	1	1
46	0	0	0	0	0
47	0	0	0	0	0
48	0	0	0	0	0
49	1	1	1	1	1
50	0	0	1	0	0
51	1	0	1	0	1
52	1	0	1	1	1
53	0	0	0	0	0
54	1	1	1	1	1
55	1	1	1	1	1

下载: 导出CSV

参考文献(19)

[1]	陈恋, 袁梅, 向维, 等. PCA−Fisher判别模型在煤层底板突水预测中的应用[J]. 数学的实践与认识,2021,51(6):103-111. CHEN Lian, YUAN Mei, XIANG Wei, et al. Application of PCA-Fisher discriminant model in prediction of water inrush from coal seam floor[J]. Mathematics in Practice and Theory,2021,51(6):103-111.
[2]	张立新, 李长洪, 赵宇. 矿井突水预测研究现状及发展趋势[J]. 中国矿业,2009,18(1):88-90. doi: 10.3969/j.issn.1004-4051.2009.01.025 ZHANG Lixin, LI Changhong, ZHAO Yu. State of research on prediction and forecast of groundwater inrush in mine and its development trend[J]. China Mining Magazine,2009,18(1):88-90. doi: 10.3969/j.issn.1004-4051.2009.01.025
[3]	国家煤矿安全监察局. 煤矿防治水细则[M]. 北京: 煤炭工业出版社, 2018. State Administration of Coal Mine Safty. Rules for coal mine water prevention and control[M]. Beijing: China Coal Industry Publishing House, 2018.
[4]	潘晖, 王继尧. 基于粒子群优化神经网络的煤层底板突水预测[J]. 山西焦煤科技,2009(1):34-36. doi: 10.3969/j.issn.1672-0652.2009.01.013 PAN Hui, WANG Jiyao. Forecast for water-inrush from coal floor based on neural networks trained by particle swarm optimization[J]. Shanxi Coking Coal Science & Technology,2009(1):34-36. doi: 10.3969/j.issn.1672-0652.2009.01.013
[5]	宋国娟. 基于极限学习机的煤矿突水预测及避险路线优化研究[D]. 徐州: 中国矿业大学, 2016. SONG Guojuan. Research on mine water inrush prediction based on extreme learning machine and route optimization[D]. Xuzhou: China University of Mining and Technology, 2016.
[6]	WANG Ge, WEI Junjie, YAO Banghua. A coal mine water inrush prediction model based on artificial intelligence[J]. International Journal of Safety and Security Engineering,2020,10(4):501-508. doi: 10.18280/ijsse.100409
[7]	师煜, 朱希安, 王占刚, 等. 基于GAPSO−RFR的矿井底板突水预测模型与应用[J]. 中国矿业,2020,29(8):152-157. SHI Yu, ZHU Xi'an, WANG Zhangang, et al. Forecast model of mine floor water inrush based on genetic particle swarm optimization and random forest regression and its application[J]. China Mining Magazine,2020,29(8):152-157.
[8]	李颖. 基于支持向量机的煤层底板突水预测方法研究[D]. 北京: 煤炭科学研究总院, 2007. LI Ying. Research on prediction method of coal seam floor water inrush based on support vector machine [D]. Beijing: China Coal Research Institute, 2007.
[9]	张风达, 申宝宏. 深部煤层底板突水危险性预测的PSO_SVM模型[J]. 煤炭科学技术,2018,46(7):61-67. ZHANG Fengda, SHEN Baohong. PSO_SVM prediction model for evaluating water inrush risk from deep coal seam floor[J]. Coal Science and Technology,2018,46(7):61-67.
[10]	李腾, 朱希安, 王占刚. 矿井突水水源判别的FOA−LSSVM模型[J]. 北京信息科技大学学报(自然科学版),2020,35(3):41-45. LI Teng, ZHU Xi'an, WANG Zhangang. FOA-LSSVM model for source identification of mine water inrush[J]. Journal of Beijing Information Science & Technology University(Natural Science Edition),2020,35(3):41-45.
[11]	MIRJALILI S, LEWIS A. The whale optimization algorithm[J]. Advances in Engineering Software,2016,95:51-67. doi: 10.1016/j.advengsoft.2016.01.008
[12]	宰慧. 基于PSO−SVM的煤层底板突水危险性预测研究[D]. 青岛: 山东科技大学, 2017. ZAI Hui. The coal floor water-inrush risk prediction research based on the PSO-SVM[D]. Qingdao: Shandong University of Science and Technology, 2017.
[13]	赵欣. 不同一维混沌映射的优化性能比较研究[J]. 计算机应用研究,2012,29(3):913-915. doi: 10.3969/j.issn.1001-3695.2012.03.031 ZHAO Xin. Research on optimization performance comparison of different one-dimensional chaotic maps[J]. Application Research of Computers,2012,29(3):913-915. doi: 10.3969/j.issn.1001-3695.2012.03.031
[14]	李雪岩, 李雪梅, 李学伟, 等. 基于混沌映射的元胞遗传算法[J]. 模式识别与人工智能,2015,28(1):42-49. LI Xueyan, LI Xuemei, LI Xuewei, et al. Cellular genetic algorithm based on chaotic map[J]. Pattern Recognition and Artificial Intelligence,2015,28(1):42-49.
[15]	赵宇, 彭珍瑞. 混沌鲸鱼优化算法及其在有限元模型修正中的应用[J]. 兰州交通大学学报,2021,40(1):39-45. ZHAO Yu, PENG Zhenrui. Whale optimization algorithm with chaotic maps and its application in finite element model updating[J]. Journal of Lanzhou Jiaotong University,2021,40(1):39-45.
[16]	刘伟韬. 矿井水害与防治[M]. 北京: 煤炭工业出版社, 2016. LIU Weitao. Mine water disaster and prevention[M]. Beijing: China Coal Industry Publishing House, 2016.
[17]	石秀伟, 胡耀青, 张和生. 基于GIS的煤层底板突水预测理论模型[J]. 太原理工大学学报,2008,38(增刊2):244-246. SHI Xiuwei, HU Yaoqing, ZHANG Hesheng. GIS-based forecasting model of floor water bursting in coal mines[J]. Journal of Taiyuan University of Technology,2008,38(S2):244-246.
[18]	NIU Huigong, WEI Jiuchuan, YIN Huiyong. An improved model to predict the water-inrush risk from an Ordovician limestone aquifer under coal seams: a case study of the Longgu Coal Mine in China[J]. Carbonates and Evaporites,2020,35(3):1-16.
[19]	SHI Longqing, GAO Weifu, HAN Jin, et al. A nonlinear risk evaluation method for water inrush through the seam floor[J]. Mine Water and the Environment,2017,36:597-605. doi: 10.1007/s10230-017-0449-1