煤矿井下瓦斯涌出量预测是煤矿安全领域中重要的研究内容，如何对瓦斯涌出量进行高效准确的预测，为井下瓦斯抽采提供数据基础，对我国煤矿的安全生产具有重要的研究意义。论文以采煤工作面瓦斯涌出量为研究对象，首先在分析瓦斯涌出量机理和影响因素的基础上，建立瓦斯涌出量多因素指标体系，然后采用主成分分析法（Principal Components Analysis, PCA）对影响因素进行降维处理，最后建立贝叶斯优化极端梯度提升预测模型，对采煤工作面的瓦斯涌出量进行预测。本文主要研究内容如下：

（1）综述瓦斯涌出量预测国内外研究现状和几种常用的瓦斯涌出量预测方法。分析煤矿瓦斯涌出量的相关影响因素，其中包括煤层厚度、煤层倾角、日进度、日产量等共计11种影响因素。在此基础上建立由多种影响因素组成的瓦斯涌出量预测指标体系。

（2）针对预测指标体系维数高和存在冗余信息的问题，采用主成分分析法对原始数据进行降维。提取了瓦斯涌出量数据的特征信息，实现数据的进一步简化，以减少原始数据中的冗余信息。

（3）针对神经网络精度欠佳的问题，选取极端梯度提升（eXtreme Gradient Boosting, XGBoost）算法作为瓦斯涌出量预测模型。针对XGBoost模型超参数难以确定的问题，分别用贝叶斯优化（Bayesian Optimization, BO）算法、网格搜索算法以及随机搜索算法对XGBoost的参数进行优化，以减少瓦斯涌出量预测模型的误差。经过实验分析对比得出：贝叶斯优化结果最佳。从而，建立基于贝叶斯优化XGBoost的预测模型。利用该模型对降维后的瓦斯量涌出数据进行预测，误差结果表明，平均绝对误差为0.0703，均方根误差为0.0957。

（4）为了验证论文所提出的PCA-BO-XGBoost预测模型的优良性，将该模型分别与PCA-XGBoost、BP（Back Propagation, BP）神经网络和支持向量机（Support Vector Machine, SVM）预测算法做对比。仿真结果表明：本论文提出的基于PCA-BO-XGBoost的瓦斯涌出量预测模型与其它三种对比模型的预测结果相比，均方根误差（RMSE）分别减少了0.92%，2.17%，8.88%，平均绝对误差（MAD）分别减少了1.29%，2.86%，6.27%。小于其它模型在相同样本下的预测误差，实验取得了良好的效果。

The prediction of gas emission is an important research content in the field of coal mine safety. How to predict gas emission efficiently and accurately provides data base for underground gas extraction, and has important research meaning for the safety production of coal mines in China. This paper takes the gas emission quantity of coal face as the research object. Firstly, based on the analysis of the mechanism and influencing factors of gas emission, the paper establishes a multi factor index system, then reduces the influencing factors by principal component analysis. Finally, Bayesian Optimization Extreme gradient elevation prediction model is established to predict the gas emission in coal mine. The main contents of this paper are as follows.

(1) This dissertation summarizes the research status of gas emission prediction at home and abroad. Several commonly used gas emission prediction methods are presented. Analysis of coal mine gas emission related factors are made, including coal seam thickness, coal seam dip angle, daily progress, daily production, a total of 11 factors. On this basis, the prediction index system of gas emission is established, which is composed of various influencing factors.

(2) Aiming at the problems of high dimension and redundant information in the prediction system, principal component analysis is employed to reduce the dimension of the original data. The feature information of gas emission data is extracted to simplify the data and reduce the redundant information in the original data.

(3) In view of the poor accuracy of neural network, the extreme gradient boosting (XGBoost) algorithm is selected as the prediction model. Aiming at the problem that it is difficult to determine the super parameters of XGBoost model, Bayesian Optimization (BO) algorithm, grid search algorithm and random search algorithm are used to optimize the parameters of XGBoost respectively to reduce the error of gas emission prediction model, respectively. According to the experimental analysis and comparison, it is concluded that the Bayesian optimization result is the best. Thus, the prediction model based on Bayesian Optimization XGBoost is established. The model is used to predict the gas emission data after dimensionality reduction. The error results show that the average absolute error is 0.0703 and the root mean square error is 0.0957.

(4) In order to verify the performance of the proposed PCA-BO-XGBoost prediction model, the model is compared with PCA-XGBoost, PCA-BP Neural Network and PCA-SVM prediction algorithm, respectively. The simulation results show that the root mean square error (RMSE) of the gas emission prediction model based on PCA-BO-XGBoost is reduced by 0.92%, 2.17%, 8.88% , and the mean absolute error (MAD) is reduced by 1.29%, 2.86%, 6.27% based on the proposed model. It is less than the prediction error of other models with the same sample, and the experiment has achieved good results.

[1]王新平, 于淮钰, 雷景婷. 煤炭企业高质量发展评价体系研究[J]. 中国矿业, 2021, 30(03): 18-24.

[2]秦容军. 我国煤炭开采现状及政策研究[J]. 煤炭经济研究, 2019, 39(01): 57-61.

[3]孙喜民. 煤炭工业高质量发展方略研究与实践[J]. 煤炭工程, 2019, 51(01): 152-156.

[4]Tang J, Jiang C, Chen Y, et al. Line prediction technology for forecasting coal and gas outbursts during coal roadway tunneling[J]. Journal of Natural Gas Science & Engineering, 2016, 34: 412-418.

[5]刘峰, 曹文君, 张建明, 等. 我国煤炭工业科技创新进展及“十四五”发展方向[J]. 煤炭学报, 2021, 46(01): 1-15.

[6]郭隆鑫, 李希建, 刘柱, 等. 基于融合权集对云的煤矿安全评价及应用[J]. 中国安全生产科学技术, 2021, 17(02): 65-70.

[7]蒋星星, 李春香. 2013—2017年全国煤矿事故统计分析及对策[J]. 煤炭工程, 2019, 51(01): 101-105.

[8]石必明, 牛宜辉, 张雷林, 等. 角联管网瓦斯爆炸超压演化及火焰传播特性研究[J]. 煤炭科学技术, 2021, 49(01): 257-263..

[9]任智旗. 采煤工作面的瓦斯涌出预测系统研究[D]. 西安: 西安科技大学, 2020.

[10]袁亮. 我国深部煤与瓦斯共采战略思考[J]. 煤炭学报, 2016, 41(01): 1-6.

[11]齐翠玉. 基于改进万有引力算法-KELM的瓦斯涌出量预测方法研究[D]. 阜新: 辽宁工程技术大学, 2019.

[12]肖鹏, 谢行俊, 双海清, 等. 小波-极限学习机在瓦斯涌出量时变序列预测中的应用[J]. 西安科技大学学报, 2020, 40(05): 839-845.

[13]王晓蕾, 姬治岗, 谢怡婷, 等. 采煤工作面瓦斯涌出量预测技术现状及发展趋势[J]. 科学技术与工程, 2019, 19(33): 1-9.

[14]Creedy David P. Geological controls on the formation and distribution of gas in British coal measure strata[J]. Elsevier, 1988, 10(01): 1-31.

[15]何宇峰. 基于改进PCA-MEA-BP神经网络的瓦斯涌出量预测研究[D]. 西安: 西安科技大学, 2020.

[16]Harpalani S, Mcpherson M J. The effect of gas evacuation on coal permeability test specimens[J]. International Journal of Rock Mechanics & Mining Sciences & Geomechanics Abstracts, 1984, 21(03): 161-164.

[17]Harpalani S. Gas flow through stressed coal[D]. Berkeley: University of California, 1985.

[18]St. George J D, Barakat M A. The change in effective stress associated with shrinkage from gas desorption in coal[J]. International Journal of Coal Geology, 2001, 45(02): 105-113.

[19]Saghafi A. A new computer simulation of in seam gas flow and its application togas emission prediction and gas drainage[C]//22nd international conference of safety in mines research institutes. 1987, 147-158.

[20]Nakayama. In-situ Measurement and Simulation by CFD of Methane Gas Distribution at a Heading Faces[J]. Shigen to Sozai, 1998, 114(11): 769-775.

[21]Valliappan S, Wohua Z. Numerical modelling of methane gas migration in dry coal seams[J]. International Journal for Numerical & Analytical Methods in Geomechanics, 2015, 20(08): 571-593.

[22]Han F, Busch A, van Wageningen N, et al. Experimental study of gas and water transport processes in the inter-cleat (matrix) system of coal: anthracite from Qinshui Basin, China[J]. International Journal of Coal Geology, 2010, 81(02): 128-138.

[23]陈俊明. 煤矿掘进工作面瓦斯涌出规律研究[J]. 能源与节能, 2017, (07): 42-43.

[24]郝全明, 张晓旭. 瓦斯涌出量预测与埋深的关系研究[J]. 内蒙古煤炭经济, 2017, (02): 89-91.

[25]徐玉胜, 李春元, 张勇, 等. 大采高采场瓦斯治理模型分析及通风系统优化研究[J]. 煤炭科学技术, 2019, 47(03): 142-149..

[26]夏海斌, 郭林生, 林建成, 等. 小庄矿40309工作面瓦斯涌出规律及影响因素分析[J]. 中国矿业, 2020, 29(10): 134-138.

[27]聂晓飞, 王文, 王恒. 高瓦斯长距离掘进工作面瓦斯涌出规律研究[J].能源与环保, 2019, 41(08): 31-34+39.

[28]蔺亚兵, 秦勇, 王兴, 等. 彬长低阶煤高瓦斯矿区瓦斯地质及其涌出特征[J]. 煤炭学报, 2019, 44(07): 2151-2158.

[29]施峰, 王宏图, 舒才. 基于固气耦合模型的煤巷掘进煤壁瓦斯动态涌出规律[J]. 煤炭学报, 2018, 43(04): 1024-1030.

[30]杨茂林, 薛友欣, 姜耀东, 等. 离柳矿区综采工作面瓦斯涌出规律研究[J]. 煤炭学报, 2009, 34(10): 1349-1353.

[31]秦玉金, 苏伟伟, 姜文忠, 等. 我国矿井瓦斯涌出量预测技术研究进展及发展方向[J]. 煤矿安全, 2020, 51(10): 52-59.

[32]Wang H, Wang E, Li Z. Study on dynamic prediction model of gas emission in tunneling working face[J]. Combustion Science and Technology, 2020, 1-17.

[33]Frodsham K, Gayer R A. The impact of tectonic deformation upon coal seams in the South Wales coalfield, UK[J]. International journal of coal geology, 1999, 38(3-4): 297-332.

[34]Shepherd J, Rixon L K, Griffiths L. Outbursts and geological structures in coal mines: A review[J]. International Journal of Rock Mechanics and Mining Science & Geomechanics Abstracts, 1981, 18(04): 267-283.

[35]Lunarzewski L W. Gas emission prediction and recovery in underground coal mines[J]. International Journal of Coal Geology, 1998, 35(1-4): 117-145.

[36]Karacan C O. Modeling and prediction of ventilation methane emissions of U.S. longwall mines using supervised artificial neural networks[J]. International Journal of Coal Geology, 2008, 73(3-4): 371-387.

[37]Krause E. Short-Term Predictions Of Methane Emissions During Longwall Mining[J]. Archives of Mining Sciences, 2015, 60(02): 581-594.

[38]Farahbakhsh E, Pourjafari M J, Faramarzi L, et al. Investigating the effect of fractures on unusual gas emission in coal mines; case study of Parvadeh coal mine, Iran[J]. International Journal of Mining and Geo-Engineering, 2016, 50(02): 163-168.

[39]Zapletal P, Koudelková J, Zubíček V, et al. New method of gas drainage as a solution of danger phenomena in underground coal mines[J]. Rudarsko-geološko-naftni zbornik, 2018, 33(01): 7-13.

[40]Meshkov Sergey. Determination of the Parameters of Technological Schemes for the Development of Coal Seams Prone to Spontaneous Combustion using Computer Simulation[J]. E3S Web of Conferences, 2019, 105.

[41]王晓蕾, 姬治岗, 谢怡婷, 等. 采煤工作面瓦斯涌出量预测技术现状及发展趋势[J]. 科学技术与工程, 2019, 19(33): 1-9.

[42]张占国, 侯志华. 基于分源预测法的厚煤层分层开采瓦斯治理技术研究[J]. 矿业安全与环保, 2017, 44(01): 78-82.

[43]李兴, 杜少华, 李洪生, 等. 高瓦斯综采工作面瓦斯涌出预测与立体防治技术研究[J]. 河北工程大学学报(自然科学版), 2019, 36(01): 98-102.

[44]张建威. 基于分源预测法的瓦斯抽采关键参数研究[J]. 神华科技, 2018, 16(08): 40-42.

[45]程波, 颜文学, 杨亮, 等. 煤矿瓦斯涌出量预测方法的研究[J]. 中国煤炭, 2019, 45(11): 63-67.

[46]马龙跃, 和毅, 乔善成. 某矿西部采区瓦斯涌出预测及分析[J]. 中国高新技术企业, 2015(15): 163-164.

[47]刘阳, 张树海. 阳泉矿区内石灰岩对瓦斯涌出量预测影响研究[J]. 测试技术学报, 2017, 31(02): 170-174.

[48]刘军锋. 矿井瓦斯涌出量预测及应用研究[J]. 山西煤炭, 2014, 34(03): 29-31+34.

[49]王艳双. 矿井瓦斯涌出量预测研究中的数学模型[J]. 煤炭技术, 2015, 34(11): 185-187.

[50]覃木广, 赵旭生, 张庆华, 等. 基于瓦斯地质特征的突出预警模型研究[J]. 煤炭科学技术, 2018, 46(12): 138-144.

[51]刘黎. 瓦斯灾害多元信息叠加预测法研究及应用[J]. 煤炭工程, 2019, 51(02): 54-57.

[52]Yuan B. Study on gas emission prediction of working face based on GM (1, 1) model[C]//Journal of Physics: Conference Series. IOP Publishing, 2020, 1549(4): 042031.

[53]Guo X, Ren Z, Wang Q, et al. Prediction of Gas Emission by BP Neural Network Based On Wavelet Analysis[C]//IOP Conference Series: Earth and Environmental Science. IOP Publishing, 2019, 252(05): 052046.

[54]Xiong Y, Cheng J T, Duan Z M. Gas Emission Prediction Model of Coal Mine Based on CSBP Algorithm[C]//ITM Web of Conferences. EDP Sciences, 2016, 7: 09006.

[55]Sirui Z, Botao W, Xueen L, et al. Research and Application of Improved Gas Concentration Prediction Model Based on Grey Theory and BP Neural Network in Digital Mine[J]. Procedia CIRP, 2016, 56: 471-475.

[56]Nie Z, Ma H, Zhang Y. Research on Gaussian Plume Model of Gas Diffusion in Coal Mine Roadway Based on BP Neural Network Optimized by Genetic Algorithm[C]//IOP Conference Series: Earth and Environmental Science. IOP Publishing, 2020, 526(01): 012158.

[57]Wu Y, Gao R, Yang J. Prediction of coal and gas outburst: A method based on the BP neural network optimized by GASA[J]. Process Safety and Environmental Protection, 2020, 133: 64-72.

[58]Sun Z, Li D. Coal Mine Gas Safety Evaluation Based on Adaptive Weighted Least Squares Support Vector Machine and Improved Dempster–Shafer Evidence Theory[J]. Discrete Dynamics in Nature and Society, 2020, 2020(03): 1-12.

[59]Wang Z Y. Research on coal mine gas prediction algorithm based on improved SVM[J]. Agro Food Industry Hi Tech, 2017, 28(01): 1729-1733.

[60]Liu H, Dong Y, Wang F. Gas Outburst Prediction Model Using Improved Entropy Weight Grey Correlation Analysis and IPSO-LSSVM[J]. Mathematical Problems in Engineering, 2020, 2020(07): 1-10.

[61]解明垒, 马尚权. 基于多元线性回归理论的煤矿瓦斯涌出量预测[J]. 陕西煤炭, 2021, 40(01): 26-29+52.

[62]张宝, 何健. 基于BP神经网络的小断层构造区域瓦斯涌出预测方法研究[J]. 煤炭工程, 2020, 52(09): 106-110.

[63]Qiu J, Jiang B, Tang M, et al. Visual Simulation and Case Inversion of Gas Explosion in Underground Mine[J]. Advances in Civil Engineering, 2021, 2021(08):1-15.

[64]徐刚, 王磊, 金洪伟, 等. 因子分析法与BP神经网络耦合模型对回采工作面瓦斯涌出量预测[J]. 西安科技大学学报, 2019, 39(06): 965-971.

[65]刘鹏, 魏卉子, 景江波, 等. 基于增强CART回归算法的煤矿瓦斯涌出量预测技术[J]. 煤炭科学技术, 2019, 47(11): 116-122.

[66]李鑫灵, 袁梅, 敖选俊, 等. PCA-SVM模型在煤层瓦斯涌出量预测中的应用[J]. 工业安全与环保, 2019, 45(10): 35-39.

[67]陈景凯, 朱振祥, 祁新波. 影响煤层瓦斯赋存规律的地质因素研究[J]. 技术与市场, 2019, 26(07): 223-225.

[68]郭晨, 夏玉成, 孙学阳, 等. 高瓦斯矿井采煤工作面瓦斯地质分级评价方法与实践[J]. 煤炭学报, 2019, 44(08): 2409-2418..

[69]张旭东. 影响矿井瓦斯涌出的因素及治理措施研究[J]. 矿业装备, 2018(06): 112-113.

[70]王小朋. 厚煤层分层开采综放工作面瓦斯涌出量预测[J]. 工矿自动化, 2020, 46(06): 72-75.

[71]柏树森. 矿井瓦斯涌出量影响因素分析[J]. 企业技术开发, 2014, 33(14): 179-180.

[72]原德胜, 陈跟马, 李子文, 等. 大佛寺煤矿40108综采工作面瓦斯涌出规律及影响因素分析[J]. 煤矿安全, 2014, 45(05): 155-158.

[73]候三中. 工作面掘进过程中瓦斯涌出量与掘进速度的关系探讨[J]. 煤矿现代化, 2016(02): 121-124.

[74]赵玉娟. 数据降维的常用方法分析[J]. 科技创新导报, 2019, 16(32): 118-119.

[75]Chen T, Guestrin C. Xgboost: A scalable tree boosting system[C]//Proceedings of the 22nd acm sigkdd international conference on knowledge discovery and data mining. 2016: 785-794.

[76]Jiang Y, Tong G, Yin H, et al. A pedestrian detection method based on genetic algorithm for optimize XGBoost training parameters[J]. IEEE Access, 2019, 7: 118310-118321.

[77]邓新国, 游纬豪, 徐海威. 贝叶斯极限梯度提升机结合粒子群算法的电阻点焊参数预测[J]. 电子与信息学报, 2021, 43(04): 1042-1049.