基于GRU和XGBoost的矿压显现规律预测

柴敬; 刘义龙; 王安义; 屈世甲; 欧阳一博

doi:10.13272/j.issn.1671-251x.2021070062

基于GRU和XGBoost的矿压显现规律预测

doi: 10.13272/j.issn.1671-251x.2021070062

1.
西安科技大学能源学院，陕西西安　710054
2.
西安科技大学西部矿井开采及灾害防治教育部重点实验室，陕西西安　710054
3.
西安科技大学通信与信息工程学院，陕西西安　710054
4.
天地(常州)自动化股份有限公司，江苏常州　213015

基金项目: 国家重点研发计划项目(2018YFC0808301)；国家自然科学基金资助项目(51804244)。

详细信息

作者简介:
柴敬(1964—)，男，宁夏平罗人，教授，博士研究生导师，主要从事采矿工程、实验岩石力学及光纤传感方面的研究工作，E-mail: chaij@xust.edu.cn

中图分类号: TD323
计量
- 文章访问数: 112
- HTML全文浏览量: 36
- PDF下载量: 18
- 被引次数: 0
出版历程
- 收稿日期: 2021-07-20
- 修回日期: 2022-01-06
- 刊出日期: 2022-01-20

Prediction of strata behaviors law based on GRU and XGBoost

1.
College of Energy Engineering, Xi'an University of Science and Technology, Xi'an 710054, China
2.
Key Laboratory of Western Mine Exploitation and Hazard Prevention, Ministry of Education, Xi'an University of Science and Technology, Xi'an 710054, China
3.
College of Communication and Information Engineering, Xi'an University of Science and Technology, Xi'an 710054, China
4.
Tiandi(Changzhou) Automation Co., Ltd. , Changzhou 213015, China

摘要

摘要: 采用光纤传感器监测的光纤频移值对矿压显现规律进行表征的过程中，传感器采集的数据存在缺失现象，无法准确预测矿压显现规律。针对该问题，以千秋煤矿为工程背景，在假设光纤下半部分数据丢失的前提下，引入GRU(门控循环单元)和LSTM(长短期记忆网络)2种预测模型，对缺失的光纤频移值进行对比预测，得出GRU模型的收敛速度优于LSTM模型的收敛速度，说明基于GRU模型的缺失值处理方法较优。将原始完整的光纤频移值转换为可表征矿压显现位置的光纤平均频移变化度，引入XGBoost(极端梯度提升)模型和BP神经网络模型进行对比预测，XGBoost模型能准确预测出测试集中所有出现“尖峰”的位置，而BP神经网络模型只预测出2处“尖峰”位置，说明XGBoost模型的预测效果优于BP神经网络模型的预测效果。将预测出的光纤频移缺失值替换至缺失位置，形成“完整”光纤频移值数据，将该数据转换为光纤平均频移变化度后，采用XGBoost模型进行预测。验证结果表明：LSTM模型及GRU模型均可准确预测出光纤下半部分的数据，且GRU模型准确性较LSTM模型准确性高；使用XGBoost可准确预测出测试集中出现的周期来压；通过GRU模型预测出的缺失数据经整合至缺失位置后，使用XGBoost模型仍可进行有效的矿压预测。
- 矿压显现规律 /
- 极端梯度提升算法 /
- XGBoost /
- GRU /
- 光纤频移值 /
- 光纤平均频移变化度
Abstract: In the process of using optical fiber frequency shift value monitored by optical fiber sensor to characterize the strata behaviors law, the data collected by the sensor is missing, and the strata behaviors law can not be accurately predicted. In order to solve this problem, taking Qianqiu Coal Mine as the engineering background, under the premise of partial data loss of the lower half of the optical fiber, two prediction models, GRU (Gated Recurrent Unit) and LSTM (Long Short-Term Memory), are introduced to compare and predict the missing optical fiber frequency shift value. The convergence speed of the GRU model is better than that of the LSTM model, which shows that the missing value processing method based on the GRU model is better. The original and complete optical fiber frequency shift value is converted into the average optical fiber frequency shift change which can characterize the strata behaviors position, and the XGBoost (eXtreme Gradient Boosting) model and the BP neural network model are introduced for comparative prediction. The XGBoost model can predict all the 'peak' positions in the test set accurately. However, the BP neural network model can only predict two 'peak' positions, which shows that the prediction effect of the XGBoost model is better than that of the BP neural network model. The predicted optical fiber frequency shift missing value is replaced to the missing position to form 'complete' optical fiber frequency shift value data. The data is converted into the average optical fiber frequency shift change and then the XGBoost model is used for prediction. The results show that both the LSTM model and the GRU model can predict the data of the lower half of the optical fiber accurately, and the GRU model has higher accuracy than the LSTM model. The XGBoost model can predict the periodic pressure in the test set accurately. After the missing data predicted by the GRU model is integrated into the missing position, the XGBoost model can still predict the strata behaviors effectively.
- strata behaviors law /
- extreme gradient boosting algorithm /
- XGBoost /
- GRU /
- optical fiber frequency shift value /
- optical fiber average frequency shift change

HTML全文

图 1 三维相似物理模型

Figure 1. 3D similar physical model

下载: 全尺寸图片幻灯片

图 2 光纤监测系统

Figure 2. Optical fiber monitoring system

下载: 全尺寸图片幻灯片

图 3 GRU网络结构

Figure 3. GRU network topology

下载: 全尺寸图片幻灯片

图 4 LSTM模型损失函数

Figure 4. LSTM model loss function

下载: 全尺寸图片幻灯片

图 5 GRU模型损失函数

Figure 5. GRU model loss function

下载: 全尺寸图片幻灯片

图 6 LSTM和GRU模型的预测结果

Figure 6. The prediction results of LSTM and GRU models

下载: 全尺寸图片幻灯片

图 7 光纤预测数据整合

Figure 7. The optical fiber prediction data integration

下载: 全尺寸图片幻灯片

图 8 光纤原始频移数据

Figure 8. The optical fiber raw frequency shift data

下载: 全尺寸图片幻灯片

图 9 Fv1光纤平均频移变化度曲线

Figure 9. The average frequency shift curve of Fv1 optical fiber

下载: 全尺寸图片幻灯片

图 10 XGBoost模型和BP神经网络模型的预测结果

Figure 10. Prediction results of XGBoost model and BP neural network model

下载: 全尺寸图片幻灯片

图 11 替换数据预测

Figure 11. Replacement data prediction

下载: 全尺寸图片幻灯片

表 1 覆岩结构及物理力学参数

Table 1. Rock structure and physical and mechanical parameters

序号	名称	岩层厚度/m	抗压强度/MPa	抗拉强度/MPa	弹性模量/ 10³ MPa	岩层容重/ （t̩·m⁻³）
15	黏土	15.0	15.0	1.5	5.0	1.6
14	泥灰岩	5.0	15.0	1.5	5.0	1.6
13	砾岩	65.0	35.0	5.5	32.0	1.8
12	细砂岩	85.0	30.0	4.0	28.0	1.6
11	砾岩	250.0	75.0	5.5	32.0	1.7
10	破碎带	1.0	40.0	4.0	28.0	1.7
9	砾岩	160.0	75.0	5.5	32.0	1.8
8	泥岩	50.0	50.0	1.2	5.0	1.9
7	细砂岩	40.0	40.0	9.0	35.0	1.6
6	粉砂岩	70.0	45.0	4.0	28.0	1.7
5	细砂岩	25.0	40.0	9.0	31.0	1.6
4	泥岩	25.0	50.0	1.2	5.0	1.9
3	2号煤	15.0	16.0	0.6	3.5	0.9
2	砾岩	8.0	65.0	5.5	32.0	1.8
1	泥岩	72.0	50.0	1.2	5.0	1.9

下载: 导出CSV

参考文献(17)

[1]	何满潮, 马新根, 王炯, 等. 中厚煤层复合顶板切顶卸压自动成巷工作面矿显现特征分析[J]. 岩石力学与工程学报,2018,37(11):2425-2434. HE Manchao, MA Xingen, WANG Jiong, et al. Feature analysis of working face strata pressure with roof cutting pressure releasing in medium-thick seam and compound roof condition[J]. Chinese Journal of Rock Mechanics and Engineering,2018,37(11):2425-2434.
[2]	高瑞.远场坚硬岩层破断失稳的矿压作用机理及地面压裂控制研究[D].徐州:中国矿业大学,2018. GAO Rui.The mechanism of ground pressure induced by the breakage of far-field hard strata and the control technology of ground fracturing[D].Xuzhou:China University of Mining and Technology,2018.
[3]	张朔滕. 基于光纤传感的矿压在线监测系统在阳煤五矿的应用研究[D]. 徐州: 中国矿业大学, 2019. ZHANG Shuoteng. Application of on-line monitoring system of mine pressure based on optical fiber sensing in Yangquan No. 5 Coal Mine[D]. Xuzhou: China University of Mining and Technology, 2019.
[4]	袁应吉, 刘洪洋, 吴文飞. 光纤光栅传感技术在煤矿安全监测中的应用[J]. 科学技术创新,2019(36):159-160. YUAN Yingji, LIU Hongyang, WU Wenfei. Application of fiber grating sensing technology in coal mine safety monitoring[J]. Scientific and Technological Innovation,2019(36):159-160.
[5]	杜文刚. 基于光纤感测的采动覆岩变形演化特征试验研究[D]. 西安: 西安科技大学, 2020. DU Wengang. Experimental study on deformation evolution characteristics of mining rock based on optical fiber sensing [D]. Xi'an: Xi'an University of Science and Technology, 2020.
[6]	王润沛. 基于机器学习的分布式光纤监测覆岩变形矿压预测研究[D]. 西安: 西安科技大学, 2020. WANG Runpei. Study on the prediction of deformation and rock pressure of overburden monitored by distributed optical fiber based on machine learning[D]. Xi'an: Xi'an University of Science and Technology, 2020.
[7]	巩师鑫, 任怀伟, 杜毅博, 等. 基于MRDA−FLPEM集成算法的综采工作面矿压迁移预测[J]. 煤炭学报,2021,46(增刊1):529-538. GONG Shixin, REN Huaiwei, DU Yibo, et al. Transfer prediction of underground pressure for fully mechanized mining face based on MRDA-FLPEM integrated algorithm[J]. Journal of China Coal Society,2021,46(S1):529-538.
[8]	赵毅鑫, 杨志良, 马斌杰, 等. 基于深度学习的大采高工作面矿压预测分析及模型泛化[J]. 煤炭学报,2020,45(1):54-65. ZHAO Yixin, YANG Zhiliang, MA Binjie, et al. Deep learning prediction and model generalization of ground pressure for deep longwall face with large mining height[J]. Journal of China Coal Society,2020,45(1):54-65.
[9]	常峰. 基于GA−BP神经网络的工作面顶板矿压预测模型应用研究[D]. 徐州: 中国矿业大学, 2019. CHANG Feng. Application research of the prediction model for the coal working face roof pressure based on GA-BP neural networks[D]. Xuzhou: China University of Mining and Technology, 2019.
[10]	李泽萌. 基于LSTM的采煤工作面矿压预测方法研究[D]. 西安: 西安科技大学, 2020. LI Zemeng. Research on prediction method of mining pressure in coal face based on LSTM[D]. Xi'an: Xi'an University of Science and Technology, 2020.
[11]	赵铭生, 刘守强, 纪润清, 等. 基于遗传算法优化BP神经网络的华北型煤田矿压破坏带深度预测[J]. 矿业研究与开发,2020,40(6):89-93. ZHAO Mingsheng, LIU Shouqiang, JI Runqing, et al. Depth prediction of mining pressure failure zone in North China coalfield based on BP neural network optimized by genetic algorithm[J]. Mining Research and Development,2020,40(6):89-93.
[12]	LUO Junling, ZHANG Zhongliang, FU Yao, et al. Time series prediction of COVID-19 transmission in America using LSTM and XGBoost algorithms[J]. Results in Physics,2021,27(3):104462-104471.
[13]	袁强. 采动覆岩变形的分布式光纤检测与表征模拟试验研究[D]. 西安: 西安科技大学, 2017. YUAN Qiang. Experimental study on distributed and representation of mining-induced overburden deformation with distributed optical fiber sensing[D]. Xi'an: Xi'an University of Science and Technology, 2017.
[14]	柴敬, 魏世明. 相似材料中光纤传感检测特性分析[J]. 中国矿业大学学报,2007,36(4):458-462. doi: 10.3321/j.issn:1000-1964.2007.04.008 CHAI Jing, WEI Shiming. Transmission character analysis of fiber optical sensing in similar material of simulation experiments[J]. Journal of China University of Mining & Technology,2007,36(4):458-462. doi: 10.3321/j.issn:1000-1964.2007.04.008
[15]	柴敬, 霍晓斌, 钱云云, 等. 采场覆岩变形和来压判别的分布式光纤监测模型试验[J]. 煤炭学报,2018,43(增刊1):36-43. CHAI Jing, HUO Xiaobin, QIAN Yunyun, et al. Model test for evaluating deformation and weighting of overlying strata by distributed optical fiber sensing[J]. Journal of China Coal Society,2018,43(S1):36-43.
[16]	朱磊. 基于光纤频移变化度的采动覆岩变形表征试验研究[D]. 西安: 西安科技大学, 2018. ZHU Lei. Experimental research on overburden deformation characterized by fiber frequency shift variation degree[D]. Xi'an: Xi'an University of Science and Technology, 2018.
[17]	TAKENS F. On the numerical determination of the dimension of an attractor[J]. Dynamical Systems and Bifrucations,1985,898:99-106.