当前，煤矿隐蔽致灾因素已成为导致煤矿重特大事故发生的主要诱因。由于隐蔽致灾因素的存在及其隐蔽性、时变性的特点，使得煤矿各类生产事故仍时有发生。在众多煤矿事故类型中，煤与瓦斯突出事故的危险性高、破坏性强，严重制约着矿井安全生产。鉴于此，本文运用扎根理论、灰色关联度分析法（GRA）、粒子群优化算法（PSO）与支持向量机（SVM）等方法，从隐蔽致灾因素角度开展煤与瓦斯突出预测研究。通过确定煤与瓦斯突出隐蔽致灾因素及预测指标，构建了基于PSO-SVM的煤与瓦斯突出预测模型，以期实现煤与瓦斯突出的准确预测，预防及避免煤与瓦斯突出事故发生。本文主要研究内容及结论如下：

完成了基于扎根理论分析的煤与瓦斯突出隐蔽致灾因素选取。运用网络爬虫技术实现了对官方网站的煤与瓦斯突出案例获取，基于扎根理论方法，经开放性译码、主轴译码、选择性译码及理论饱和性检验等步骤确定出应力因素、地质构造因素、瓦斯因素、煤体赋存状况因素4项煤与瓦斯突出隐蔽致灾因素。

（2）实现了基于GRA的煤与瓦斯突出预测指标降维优化筛选。基于煤与瓦斯突出隐蔽致灾因素并结合文献分析法进行预测指标初步选取，共得到9项初始预测指标。针对过多指标输入导致模型结构复杂、计算量大、维数灾难等问题，采用灰色关联度分析法（GRA）对初始预测指标进行特征提取，得到瓦斯涌出初速度、瓦斯含量、钻屑瓦斯解吸指标、最大钻屑量的关联度分别为0.854，0.843，0.821，0.802，根据关联规则实现了预测指标的降维优化筛选。

（3）构建了基于PSO优化SVM的煤与瓦斯突出预测模型。针对SVM模型训练过程中的预测性能受核心参数影响较大的问题，引入粒子群优化算法（PSO）从全局的角度实现支持向量机RBF核函数惩罚因子C和核参数g的参数优化，得到参数C和g的最优值分别为1354.2358与2.4721。基于GRA优化筛选的主控指标及PSO算法优化后的最佳参数赋值模型输入，构建了PSO-SVM煤与瓦斯突出预测模型。

（4）开展了PSO-SVM煤与瓦斯突出预测模型预测应用及对比验证分析。为验证PSO-SVM预测模型的预测性能，通过与SVM、GA-SVM、BP神经网络三种模型对比，得到四类预测模型的均方误差MSE分别为0.0774、0.2266、0.1273、0.1769，相关系数R²分别为0.9612、0.9241、0.8464、0.8571。结果表明：PSO-SVM模型具有较好的预测精度和泛化能力，预测结果的误差最低且相关系数最高，与实际情况最为接近。最后结合PSO-SVM模型的自检验分析过程，验证了本模型具有预测准确率高、稳定性好和泛化能力强的优势，具有较好的实用性和可操作性，对煤矿局部瓦斯防突治理工作具有重要参考价值。

At present, hidden disaster-causing factors in coal mines have become the main causative factors leading to the occurrence of serious and large-scale accidents in coal mines. Due to the existence of hidden disaster-causing factors and their hidden and time-varying characteristics, various types of production accidents in coal mines still occur from time to time. Among the many types of coal mine accidents, prominent coal and gas accidents are highly dangerous and destructive, seriously restricting safe production in mines.In view of this, this paper applies Zagan theory, grey correlation analysis (GRA), particle swarm optimization algorithm (PSO) and support vector machine (SVM) to carry out the research on coal and gas protrusion prediction from the perspective of hidden disaster-causing factors.By determining the hidden disaster factors and prediction indicators of coal and gas outburst, a PSO-SVM coal and gas outburst prediction model was constructed and analyzed for verification, in order to achieve accurate prediction of coal and gas outburst, prevent and reduce the occurrence of coal and gas outburst accidents. The main research content and conclusions of this article are as follows:

(1)Completed the selection of hidden disaster causing factors for coal and gas outburst based on grounded theory analysis. Web crawler technology is used to obtain the cases of coal and gas outbursts from the official website, and based on the rooted theory method, four hidden factors of coal and gas outbursts are identified through the steps of open decoding, principal axis decoding, selective decoding and theoretical saturation test, such as the stress factor, geological structure factor, gas factor, and the factor of the coal body's storage condition.

(2)Realized the optimization and screening of dimensionality reduction for coal and gas outburst prediction indicators based on GRA. Based on the hidden disaster factors of coal and gas outburst and combined with literature analysis method, a preliminary selection of prediction indicators was conducted, and a total of 9 initial prediction indicators were obtained. In response to the problems of complex model structure, high computational complexity, and disastrous dimensionality caused by excessive input of indicators, the Grey Relational Analysis (GRA) method was used to extract features from the initial prediction indicators. The correlation degrees of gas emission initial velocity, gas content, drilling gas analysis, and maximum drilling gas volume were 0.854, 0.843, 0.821, and 0.802, respectively. Based on the association rules,the dimensionality reduction and optimization screening of the prediction indicators were achieved.

(3)A coal and gas protrusion prediction model based on PSO optimised SVM was constructed. In response to the problem that the predictive performance of SVM models is greatly affected by core parameters during training, the particle swarm optimization algorithm (PSO) is introduced to globally optimize the RBF kernel function penalty factor C and kernel parameter g of SVM. The optimal values of parameters C and g are 1354.2358 and 2.4721, respectively. Based on the prediction indicators optimized by GRA and the input of the SVM core parameter assignment model optimized by PSO, a PSO-SVM coal and gas outburst prediction model was constructed.

(4)Carried out PSO-SVM coal and gas protrusion prediction model prediction application and comparative validation analysis. To verify the predictive performance of the PSO-SVM prediction model, by comparing it with three models: SVM, GA-SVM, and BP neural network, the mean square error (MSE) of the four prediction models were 0.0774, 0.2266, 0.1273,and 0.1769, respectively, and the correlation coefficients R² were 0.9612, 0.9241, 0.8464 ,and 0.8571, respectively.The results show that the PSO-SVM model has good prediction accuracy and generalization ability, with the lowest prediction error and the highest correlation coefficient, which is closest to the actual situation. Finally, combined with the self inspection analysis process of the PSO-SVM model, it was verified that this model has the advantages of high prediction accuracy, prediction stability, and strong generalization ability. It has good practicality and operability, and has important reference value for gas outburst prevention and control work in coal mine areas

[1]谢和平, 吴立新, 郑德志. 2025年中国能源消费及煤炭需求预测[J]. 煤炭学报, 2019, 44(07): 1949-1960.

[2]国家统计局. 中华人民共和国2022年国民经济和社会发展统计公报[J]. 中国统计, 2023(3): 12-29.

[3]谢和平, 王金华, 鞠杨, 等. 煤炭革命的战略与方向[M]. 北京: 科学出版社, 2018: 1-31.

[4]高新民. 煤矿隐蔽致灾因素的危害性及建议[C]//陕西省煤炭学会. 煤矿隐蔽致灾因素及探查技术研究. 煤炭工业出版社, 2015: 3.

[5]张培森, 张晓乐, 董宇航等. 2008-2021年我国煤矿事故规律分析及预测研究[J]. 矿业安全与环保, 2023, 50(02): 136-140+146.

[6]吴敏杰, 王相业, 张金贵,等. 神北矿区庙梁煤矿隐蔽致灾因素探查与分析[J]. 中国煤炭, 2023, 49(01): 35-43.

[7]申宝宏, 郑行周, 弯效杰. 煤矿隐蔽致灾因素普查技术指南[M]. 北京: 煤炭工业出版社, 2015: 10.

[8]谢斌红, 马非, 潘理虎, 等. 煤矿安全隐患信息自动分类方法[J]. 工矿自动化, 2018, 44(10): 10-14.

[9]范立民. 煤矿隐蔽致灾因素及探查技术[C]//陕西省煤炭学会. 煤矿隐蔽致灾因素及探查技术研究. 北京: 煤炭工业出版社, 2015: 15-24.

[10]齐庆杰, 刘文岗, 李首滨, 等. 煤矿事故隐患消除科技支撑对策研究[J]. 煤炭科学技术, 2021, 49(04): 20-27.

[11]LM Gochioco, JR Gochioco, Fruev. Coalgeophysics expands with growing global demands for mine safety and productivity. LeadingEdge, 2012, 31(03): 308-314.

[12]宝隆, 王向前, 李慧宗, 等. 煤矿事故案例本体知识库的构建及推理研究[J]. 工矿自动化, 2018, 44(03): 35-41.

[13]李兵. 张双楼煤矿冲击地压隐蔽致灾因素分析及治理技术研究[J]. 中国煤炭, 2022, 48(S2): 76-85.

[14]徐德金, 邵德盛. 地方小煤矿水害隐蔽致灾因素识别方法[J]. 辽宁工程技术大学学报(自然科学版), 2014, 33(12): 1599-1602.

[15]许多康, 李雷波. 织金县煤矿隐蔽致灾因素研究与防控实践[J]. 中国煤炭, 2022, 48(S2): 135-142.

[16]刘建江. 胡底煤业隐蔽致灾因素普查与相应治理措施研究[J]. 中国煤炭,2022, 48(S2): 166-170.

[17]霍雨佳. 河北省煤矿主要隐蔽致灾因素分布规律及成因机制分析[D]. 邯郸: 河北工程大学, 2019.

[18]赵惊涛, 彭苏萍, 陈宗南, 等. 煤矿隐蔽致灾地质体地震绕射波探测方法[J]. 矿业科学学报, 2022, 7(01): 1-8.

[19]牟义. 神府矿区隐蔽采空区相关致灾因素分析及勘查技术[J]. 地球物理学进展, 2020, 35(03): 1017-1024.

[20]秦永军. 煤矿隐蔽致灾因素探查的技术路线[C]//陕西省煤炭学会. 煤矿隐蔽致灾因素及探查技术研究. 北京: 煤炭工业出版社, 2015: 3.

[21]李强, 吴红华. GeoEast解释系统在煤田隐蔽致灾因素排查中的应用[C]//中国地球物理学会,中国地震学会,全国岩石学与地球动力学研讨会组委会,中国地质学会构造地质学与地球动力学专业委员会,中国地质学会区域地质与成矿专业委员会.2017中国地球科学联合学术年会论文集(三十)专题58: 油气田与煤田地球物理勘探. 中国和平音像电子出版社, 2017: 3.

[22]柳顺彬, 李文祥, 冉军林, 等. 半航空瞬变电磁在煤矿隐蔽致灾领域的应用[J]. 新疆地质, 2023, 41(01): 113-117.

[23]方俊. 煤矿井下隐蔽致灾因素定向钻孔探查技术研究[D].西安: 西安科技大学, 2019.

[24]Tian Shuicheng, Mao Junrui, Li Hongxia. Disaster-Causing Mechanism of Hidden Disaster-Causing Factors of Major and Extraordinarily Serious Gas Explosion Accidents in Coal Mine Goafs[J]. Sustainability, 2022, 14(19), 12018.

[25]Zhao Yifan, Tian Shuicheng. Identification of hidden disaster causing factors in coal mine based on Naive Bayes algorithm[J]. Journal of Intelligent & Fuzzy Systems, 2021, 41(02), 2823-2831.

[26]田水承, 景国勋. 安全管理学[M]. 北京: 机械工业出版社, 2022.

[27]刘晓伟. 钻屑量差值法在贺西矿防突工作中的应用[J]. 山东煤炭科技, 2021, 39(05): 112-114+117 +120.

[28]李宗保, 苏龙, 胥鹏. 基于钻屑指标法的掘进工作面区防突技术研究[J]. 科技创新导报, 2014, 11(27): 29.

[29]孙亚杰. 灰色关联分析在煤与瓦斯突出预测中的应用[J]. 中州煤炭, 2013, No.208(04): 82-84.

[30]Aitao Zhou, Kai Wang, Lei Li, et al. A roadway driving technique for preventing coal and gas outbursts in deep coal mines[J]. Environmental Earth Sciences, 2017, 76(6):1-10.

[31]牛俊豪. 响水煤矿煤与瓦斯突出区域和局部预测敏感指标及临界值研究[D]. 徐州: 中国矿业大学, 2021.

[32]张浪. 突出煤体变形破坏过程声发射演化特征综合分析[J]. 煤炭学报, 2018, 43(S1): 130-139.

[33]王雨虹, 刘璐璐, 付华, 等. 基于声发射多参数时间序列的瓦斯突出预测[J]. 中国安全科学学报, 2018, 28(05): 129-134.

[34]Hui Li, Zengchao Feng, Dong Zhao, et al. Simulation Experiment and Acoustic Emission Study on Coal and Gas Outburst[J]. Rock Mechanics and Rock Engineering, 2017, 50(8): 2193-2205.

[35]Li Bing, Wang Enyuan, Shang Zheng, et al. Deep learning approach to coal and gas outburst recognition employing modified AE and EMR signal from empirical mode decomposition and time-frequency analysis[J]. Journal of Natural Gas Science and Engineering, 2021, 90:103942.

[36]何学秋, 王安虎, 窦林名, 等. 突出危险煤层微震区域动态监测技术[J]. 煤炭学报, 2018, 43(11): 3122-3129.

[37]Liming Qiu,Zhonghui Li,Enyuan Wang, et al. Characteristics and precursor information of electromagnetic signals of mining-induced coal and gas outburst[J]. Journal of Loss Prevention in the Process Industries, 2018, 54(04): 206-215.

[38]王恩元, 李忠辉, 何学秋, 等. 煤与瓦斯突出电磁辐射预警技术及应用[J]. 煤炭科学技术, 2014, 42(06): 53-57+91.

[39]陈亮, 范志浩, 付江伟, 等. 基于电磁辐射峰谷比值法的煤与瓦斯突出预警[J].中国安全生产科学技术, 2019, 15(06): 94-98.

[40]李楠, 王恩元, GE Mao-chen. 微震监测技术及其在煤矿的应用现状与展望[J]. 煤炭学报, 2017, 42(S1): 83-96.

[41]朱权洁, 李青松, 李绍泉, 等. 煤与瓦斯突出试验的微震动态响应与特征分析[J]. 岩石力学与工程学报, 2015, 34(S2): 3813-3821.

[42]刘超. 采动煤岩瓦斯动力灾害致灾机理及微震预警方法研究[D]. 大连: 大连理工大学, 2011.

[43]Shu Longyong, Liu Zhengshuai, Wang Kai,et al. Characteristics and Classification of Microseismic Signals in Heading Face of Coal Mine: Implication for Coal and Gas Outburst Warning[J]. Rock Mechanics and Rock Engineering, 2022, 55(11): 6905-6919.

[44]周西华, 孙家正. 基于主因子分析的改进BP神经网络瓦斯涌出量预测[J]. 矿业安全与环保, 2018, 45(06): 43-47+52.

[45]侯福营. 基于改进PCA-ELM的煤与瓦斯突出软测量研究[D]. 阜新: 辽宁工程技术大学, 2014.

[46]李心杰. 基于SA-GA-FCM的煤与瓦斯突出预测研究[D]. 阜新: 辽宁工程技术大学, 2015.

[47]Zhigang Yan, Kan Yao, Yuanxuan. Yang. A novel adaptive differential evolution SVM model for predicting coal and gas outbursts[J]. Journal of Difference Equations and Applications, 2016, 23(1-2): 238-248.

[48]Yueqiang Liang, Deyong Guo, Zhanfeng Huang, et al. Prediction model for coal-gas outburst using the genetic projection pursuit method[J]. Int. J. of Oil, Gas and Coal Technology, 2017, 16(3):271-282.

[49]林海峰. 煤与瓦斯突出预测敏感指标敏感性分析与研究[J]. 能源技术与管理, 2018, 43(02): 20-21+33.

[50]Junhui Mou, Huihui Liu, Yinhui Zou, et al. A new method to determine the sensitivity of coal and gas outburst prediction index[J]. Arabian Journal of Geosciences, 2020, 13(12): 253-266.

[51]刘硕. K1值预测突出敏感性的实验研究[D]. 徐州: 中国矿业大学, 2016.

[52]Zhonghui Li, et al. Hazard evaluation of coal and gas outbursts in a coal-mine roadway based on logistic regression model[J]. International Journal of Rock Mechanics and Mining Sciences, 2015, 80: 185-195.

[53]梁冰, 秦冰, 孙维吉, 等. 智能加权灰靶决策模型在煤与瓦斯突出危险评价中的应用[J]. 煤炭学报, 2013, 38(09): 1611-1615.

[54]赵云涛, 王佳, 周小平. 基于支持向量机的公共建筑火灾风险评估[J]. 建筑科学, 2015, 31(03): 98- 101+109.

[55]赵娜. 基于支持向量机建筑结构损伤识别方法研究[J]. 黑龙江科技信息, 2017(13): 125.

[56]钟欣. 支持向量机的建筑施工安全预警模型仿真[J]. 计算机仿真, 2019, 36(08): 459-462.

[57]张笑, 张明. 基于SVM的偏斜车牌识别[J]. 现代计算机(专业版), 2016(23): 64-68.

[58]张传伟, 崔万豪. 基于Gabor特征提取和SVM交通标志识别方法研究[J]. 现代电子技术, 2018, 41(17): 136-140.

[59]杨萌宇, 张雷, 曾悦. 改进粒子群算法优化的SVM在恶性肿瘤诊断中的应用[J]. 现代电子技术, 2020, 43(15): 110-114+118.

[60]李欢, 贾佳, 杨秀宇, 等. 煤矿综采工作面瓦斯浓度预测模型[J]. 工矿自动化, 2018, 44(12): 48-53.

[61]张帝, 孟磊, 董飞, 等. 矿井突水水源识别的GA-SVM方法研究[J]. 煤炭技术, 2018, 37(04):144-147.

[62]李烨, 贾进章. 基于改进GS-SVM的煤矿冲击地压预测研究[J]. 世界科技研究与发展, 2016, 38(04): 758-762.

[63]杨力, 王蕾. 基于熵值法和支持向量机的煤矿应急救援能力评价模型[J]. 中国安全生产科学技术, 2015, 11(04): 113-119.

[64]赵华天. 基于支持向量机的矿井瓦斯浓度预测应用研究[J]. 煤炭科技, 2018(04): 41-44.

[65]曾维顺, 韩欣澎, 刘金霖. 基于粒子群优化支持向量机的瓦斯突出层研究[J]. 中国煤炭地质, 2018, 30(11): 44-46.

[66]张文娟. 基于支持向量机和免疫遗传BP的瓦斯浓度预测研究[D]. 西安: 西安科技大学, 2017.

[67]Liao Jianxing, Lv Junwei, Shao Yuhan, et al. Application of BP Neural Network Ensemble Model Based on Bagging Algorithm[J]. International Journal of Machine Learning and Computing, 2019, 9(2): 121-128.

[68]梁跃强. 基于地质数据挖掘和信息融合的煤与瓦斯突出预测方法[D]. 北京: 中国矿业大学(北京), 2018.

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

[70]范立民, 煤矿隐蔽致灾因素与探查[M]. 北京: 煤炭工业出版社, 2014: 5.

[71]赵春侠. 我国煤矿生产安全事故统计分析及预测[D]. 绵阳: 西南科技大学, 2017.

[72]国家安全生产监督管理局. 国家煤矿安监局关于深刻吸取事故教训进一步防范和遏制煤矿重特大事故的通知[J]. 煤矿开采, 2016, 21(2): 119-120.

[73]张超林, 王恩元, 王奕博,等. 近20年我国煤与瓦斯突出事故时空分布及防控建议[J]. 煤田地质与勘探, 2021, 49(4): 134-141.

[74]Strauss A L. Qualitative analysis for social scienti-sts[M]. Cambridge university press, 1987: 5.

[75]孙庆兰, 田水承. 基于扎根理论的煤矿运输险兆事件影响因素研究[J]. 安全与环境学报, 2017, 17(3): 1031-1037.

[76]杜二玲, 范毅君, 李海军. 统计学习理论基础研究新进展[J]. 现代工业经济和信息化, 2016, 6(18): 27-28.

[77]边肇祺. 模式识别第2版[M]. 北京: 清华大学出版社, 2005: 5

[78]Vapnik V. The nature of statistical learning theory [M]. Springer science & business media, 2013:12.

[79]胡千庭, 周世宁, 周心权. 煤与瓦斯突出过程的力学作用机理[J]. 煤炭学报, 2008, 33(12): 1368-1372.

[80]李希建, 林柏泉. 煤与瓦斯突出机理研究现状及分析[J]. 煤田地质与勘探, 2010, 38(01): 7-13.

[81]李涛, 陈刚, 付乙财, 等. 地应力分布规律对煤与瓦斯突出的影响[J]. 价值工程, 2015, 34(12): 170-172.

[82]刘业娇, 袁亮, 薛俊华, 等. 煤与瓦斯突出机理和模拟试验研究现状及发展趋势[J].工矿自动化, 2018, 44(02): 43-50.

[83]舒龙勇. 煤与瓦斯突出的关键结构体致灾机理[D]. 北京: 中国矿业大学(北京), 2019.

[84]Zhou Jiangjie, Wang Shengfeng. Application of Python Web Crawler Technology in Infodemiology[J]. Chinese Journal of Epidemiology, 2020, 41(6): 952-956.

[85]张晓铭, 张爱绒, 郭勇义. 基于PSO算法优化的自组织竞争神经网络在煤与瓦斯突出预测中的应用研究[J]. 中国煤炭, 2013, 39(01): 106-109.

[86]俞启香. 矿井瓦斯防治[M], 徐州: 中国矿业大学出版社, 1992: 8.

[87]李文娟. 改进的极限学习机在煤与瓦斯突出预测中的应用[D]. 阜新: 辽宁工程技术大学, 2015.

[88]Liang Wang, Yuan-ping Cheng, Lei Wang, et al. Safety line method for the prediction of deep coal-seam gas pressure and its application in coal mines[J]. Safety Science, 2011, 50(3), 523-529.

[89]Junhui Mou, Huihui Liu, Yinhui Zou, et al. A new method to determine the sensitivity of coal and gas outburst prediction index[J]. Arabian Journal of Geosciences, 2020, 13(12), 253-266.

[90]韩颖, 余伟凡, 杨志龙, 等. 基于灰关联方法的突出预测敏感指标研究[J]. 河南理工大学学报(自然科学版), 2012, 31(01): 16-18.

[91]程远平, 周红星. 煤与瓦斯突出预测敏感指标及其临界值研究进展[J]. 煤炭科学技术, 2021, 49(01): 146-154.

[92]杜泽生, 罗海珠, 孙波, 等. 基于四率法的煤与瓦斯突出预测敏感指标的确定[J]. 煤炭科学技术, 2010, 38(07): 44-47.

[93]仇海生, 都峰, 王丰. 钻孔瓦斯涌出初速度指标的影响因素分析[J]. 煤矿安全, 2010, 04: 99-101.

[94]邓聚龙. 灰色系统综述[J]. 世界科学, 1983(07): 1-5.

[95]田水承, 马云龙, 寇猛, 等. 基于灰色关联分析的煤矿险兆事件致因分析[J]. 煤炭技术, 2015, 34(03): 334-336.

[96]王雨虹, 孙福成, 付华, 等. 基于优化的量子门节点神经网络的煤与瓦斯突出预测[J]. 信息与控制, 2020, 49(02): 249-256.

[97]黄文秀. 粒子群优化算法的发展研究[J]. 软件, 2014, 35(04): 73-77.

[98]杨慧斌. 滚动轴承故障诊断中的特征提取与选择方法[D]. 株洲:湖南工业大学, 2011: 68-77.

[99]杨韬. 粒子群算法的参数分析[J]. 科技信息(学术研究), 2008(14): 295+297.

[100]Chang C C, Lin C J. LIBSVM: A library for support vector machines[J]. ACM Transactions on Intelligent Systems and Technology (TIST), 2011, 2(3, article 27): 1-27.

[101]汤小燕, 齐佳新, 崔宇寒,等. 神经网络在煤层瓦斯含量预测中的应用[J]. 地球物理学进展, 2023, 38(1): 494-501.