煤矿掘进安全是掘进工作面智能化建设的重要内容，煤矿作业人员面临着众多的健康威胁，如煤尘、噪音、震动、高温等，造成煤矿安全事故和诸多职业病隐患。传统的人体健康监测方法主要依赖于人工观察和测量，存在监测不便和评估困难的问题，目前煤矿作业人员健康监测与评估方面研究较少，亟待有效的监测手段和评估方法。针对以上问题，本文提出一种数字孪生（Digital Twin, DT）驱动的煤矿作业人员健康监测与威胁评估方法，利用因果图方法从5M1E六个维度对煤矿作业人员进行分析，结合计算机建模与仿真、智能传感器和机器学习等技术搭建煤矿作业人员健康监测数字孪生模型。虚拟表征煤矿作业人员生理状态和姿态信息，并利用MQTT通讯协议在unity3D虚拟环境中实现煤矿作业人员的虚实数据交互，实现人员生命体征、行为姿态和煤矿工况环境的实时监测。

针对煤矿作业人员健康监测模型构建难、监测数据不明确等问题，提出并构建煤矿作业人员健康监测数字孪生模型。通过分析煤矿作业人员物理对象与虚拟对象的统一映射关系，实现煤矿作业人员物理空间与虚拟空间的精准映射。采用人体测量和三维建模技术构建人体数字孪生拟态模型。基于计划行为理论和有限状态自动机（Finite State Machine，FSM）构建煤矿作业人员相应的行为状态模型，利用Media pipe-Blaze Pose进行人体姿态估计实现人体虚实同动，将煤矿作业人员健康状态具象到生理模型和行为模型的表达，为人员健康监测和状态评估奠定基础。

针对掘进工作面作业人员健康缺乏有效监测的问题，研究一种数字孪生驱动的煤矿作业人员健康监测方法。通过采集掘进工作面作业人员生命体征信息和姿态等实时数据驱动煤矿作业人员孪生模型，实现作业人员的健康可视化监测。同时提出了一种人员健康异常报警方法，包括异常报警机制设计、异常数据阈值设计以及报警实现方式，以便对掘进工作面煤矿作业人员的健康异常进行实时报警。

针对掘进工作面煤矿作业人员作业环境恶劣，难以进行有效威胁评估的问题，研究煤矿作业人员威胁评估模型。通过对煤矿作业人员健康的威胁要素进行分析，对采集到的生命体征数据和工况环境数据建立数据集，通过Stacking（堆叠）模型融合逻辑回归和随机森林两种基模型的方法构建煤矿作业人员威胁评估模型，并通过准确率、召回率、精确度和F1-score等指标对模型进行评估。通过参数搜索和交叉验证，找到模型最优的超参数选取最优评估模型，从而预测潜在的健康威胁和事故风险。

最后，通过搭建煤矿作业人员数字孪生健康监测系统，设计实验测试煤矿作业人员健康监测系统的功能和性能，对系统的数据交互功能、人体姿态估计与虚实同步、人体健康监测与异常报警等进行实验验证。该研究为煤矿作业人员的健康管理和安全生产提供了新的理论和技术支撑。

Safety in coal mine excavation is an important aspect of the intelligent construction of mining workface. Coal mine workers face numerous health threats such as coal dust, noise, vibration, and high temperatures, which lead to safety accidents and occupational hazards. In order to reduce safety accidents and occupational diseases, there is a pressing need for effective monitoring and evaluation of workers' health. Traditional methods of monitoring human health rely mainly on manual observation and measurement, which have problems with inconvenience and difficulty in evaluation.

To address these issues, this paper proposes a digital twin (DT)-driven method for monitoring the health of coal mine workers and evaluating threats. The 5M1E model is used to analyze coal mine workers from six dimensions, and combined with computer modeling and simulation, intelligent sensors, and machine learning techniques, a digital twin model for monitoring the health of coal mine workers is built. The virtual representation of the physiological status and posture information of coal mine workers is created, and the MQTT communication protocol is used to implement the exchange of virtual and real data in a unity3D virtual environment, achieving real-time monitoring of workers' vital signs, behavioral postures, and mining environment conditions.

To address the difficulty of constructing the health monitoring model for coal mine workers and the lack of clear monitoring data, a digital twin model for monitoring the health of coal mine workers is proposed and built. By analyzing the unified mapping relationship between the physical and virtual objects of coal mine workers, the accurate mapping between the physical and virtual spaces is achieved. The human digital twin simulation model is built using body measurement and 3D modeling technology. Based on the Planned Behavior Theory and the Finite State Machine (FSM), the corresponding behavioral state model of coal mine workers is constructed, and the Media pipe-Blaze Pose is used for human posture estimation to achieve virtual and real-time synchronization of human body. The expression of the health status of coal mine workers is concretized into physiological and behavioral models, laying a foundation for personnel health monitoring and status evaluation.

To address the lack of effective monitoring methods for the health of workers in the excavation workface, a DT-based method for monitoring the health of coal mine workers is proposed. By collecting real-time data on workers' vital signs and posture, data support is provided for subsequent health abnormal alarm. At the same time, a method for detecting health abnormalities is proposed, including the design of an abnormal alarm mechanism, the design of an abnormal data threshold, and the implementation of the alarm, in order to provide real-time alarms for health abnormalities in the excavation workface.

To address the problem of insufficient health data for underground coal mine workers, making it difficult to evaluate threats effectively, a coal mine worker hazard assessment model is proposed. By analyzing the threat factors to the health of underground coal mine workers and establishing a dataset of vital signs and working environment data, a hazard assessment model for coal mine workers is constructed by stacking two basic models, logistic regression and random forest. The model is evaluated by indicators such as accuracy, recall rate, precision, and F1-score. By parameter search and cross-validation, the optimal hyperparameters are selected to achieve the optimal evaluation model, predicting potential health threats and providing guidance for safety management in the coal mining industry.

Finally, a digital twin system is built, and experiments are designed to verify the functions and performance of the coal mine worker health monitoring system. The system's communication functions, human posture estimation and virtual-reality synchronization functions, human health abnormal alarm functions, and emergency one-key remote control device switching functions are experimentally verified. This study provides new theoretical and practical support for the health management and safe production of coal mine workers.

[1]中华人民共和国科学技术部．科技部关于印发《“十三五”国家社会发展科技创新规划》的通知[EB/OL]

[2]李博.基于人机工程学的割灌机作业人体疲劳评估及实验研究[D].哈尔滨:东北林业大学,2018.

[3]陈臻,付文俊.煤矿巷道掘进作业智能监测与安全预警系统设计[J].煤炭技术, 2018,37(2):214-216.

[4]赵红泽,原江涛,韦钊,等.煤矿班组隐患排查系统构建技术[J].煤矿安全,2017, 48(1): 228-230,234.

[5]韦钊,赵红泽,何桥.煤矿企业班组隐患排查体系研究及实践[J].煤炭工程,2017(1):135-137,141.

[6]陶飞,刘蔚然,刘检华,等.数字孪生及其应用探索[J].计算机集成制造系统，2018，24(1):1-18．

[7]李浩,陶飞,王昊琪,等.复杂产品设计制造一体化开发框架与关键技术[J].计算机集成制造系统,2019,25(06):1320-1336.

[8]葛世荣,王忠宾,王世博.互联网+采煤机智能化关键技术研究[J].煤炭科学技术, 2016,44(07):1-9.

[9]张旭辉,魏倩楠,王妙云,等.悬臂式掘进机远程虚拟操控系统研究[J].煤炭科学技术, 2020,48(11):44-51.

[10]张旭辉,王妙云,张雨萌,等.数据驱动下的工业设备虚拟仿真与远程操控技术研究[J].重型机械,2018(05):14-17.

[11]张旭辉,张雨萌,王妙云,杜昱阳,谢楠,鞠佳杉.基于混合现实的矿用设备维修指导系统[J].工矿自动化,2019,45(06):27-31.

[12]胡天亮,连宪辉,马德东,等.数字孪生诊疗系统的研究[J].生物医学工程研究, 2021, 40(01):1-7.

[13]邱一鸣.人体腰椎数字孪生健康监测系统及关键技术[D].大连:大连理工大学,2022.

[14]苗彦平,朱玉峰,张展鹏,等.4种风险评估方法在某煤矿企业粉尘职业健康风险评估中的应用及比较[J].工业卫生与职业病,2022,48(06):441-445.

[15]宋学官,来孝楠,何西旺,等.重大装备形性一体化数字孪生关键技术[J].机械工程学报, 2022,58(10):298-325.

[16]宋学官,何西旺,李昆鹏,等.人体骨骼数字孪生的构建方法及应用[J].机械工程学报, 2022,58(18):218-228.

[17]Okamoto K, Berntorp K, Di Cairano S. Driver intention-based vehicle threat assessment using random forests and particle filtering[J]. IFAC-PapersOnLine, 2017, 50(1): 13860-13865.

[18]Lu Y, Lei X, Zhou Z, et al. Approximate reasoning based on IFRS and DS theory with its application in threat assessment[J]. IEEE Access,2020,8:160558-160568.

[19]陶飞,刘蔚然,刘检华,等.数字孪生及其应用探索[J].计算机集成制造系统, 2018,24(01):1-18.

[20]罗明坤,邓三鹏,薛强,等.基于数字孪生的工业机器人状态监测技术研究[J].机器人技术与应用,2022(06):15-17.

[21]徐非.基于数字孪生技术的自动化微电子生产线智能控制研究[J].自动化与仪器仪表,2022(10):93-97.

[22]董培林,寇向宇,吴玓文,等.数字孪生技术在矿山物联网管控系统的应用[J].矿业研究与开发,2022,42(10):163-168.

[23]鲍劲松,张荣,李婕,等.面向人-机-环境共融的数字孪生协同技术[J].机械工程学报, 2022,58(18):103-115.

[24]黄彬彬,张映锋,黄博,等.数字孪生驱动的复杂产品智能运维服务体系与核心技术[J].机械工程学报,2022,58(12):250-260.

[25]葛世荣,张帆,王世博,等.数字孪生智采工作面技术架构研究[J].煤炭学报, 2020,45(06):1925-1936.

[26]丁恩杰,俞啸,夏冰,等.矿山信息化发展及以数字孪生为核心的智慧矿山关键技术[J].煤炭学报,2022,47(01):564-578.

[27]张旭辉,张雨萌,王岩,等.融合数字孪生与混合现实技术的机电设备辅助维修方法[J].计算机集成制造系统,2021,27(08):2187-2195.

[28]陶飞,张贺,戚庆林等.数字孪生模型构建理论及应用[J].计算机集成制造系统, 2021,27(01):1-15.

[29]董雷霆,周轩,赵福斌等.飞机结构数字孪生关键建模仿真技术[J].航空学报, 2021,42(03):113-141.

[30]王运达,张钢,于泓,等.基于数字孪生的城轨供电系统高保真建模方法[J].高电压技术, 2021,47(05):1576-1583.

[31]邹琦,侯志霞,王明阳,等.机加零件的数字孪生模型构建方法[J].航空制造技术, 2020,63(03):67-75.

[32]肖通,江海凡,丁国富,等.五轴磨床数字孪生建模与监控研究[J].系统仿真学报, 2021,33(12):2880-2890.

[33]谢嘉成,王学文,杨兆建,等.基于数字孪生的综采工作面生产系统设计与运行模式[J].计算机集成制造系统,2019,25(06):1381-1391.

[34]丁华,杨亮亮,杨兆建,等.数字孪生与深度学习融合驱动的采煤机健康状态预测[J].中国机械工程,2020,31(07):815-823.

[35]赵浩然,刘检华,熊辉,等.面向数字孪生车间的三维可视化实时监控方法[J].计算机集成制造系统,2019,25(06):1432-1443.

[36]李浩,刘根,文笑雨,等.面向人机交互的数字孪生系统工业安全控制体系与关键技术[J].计算机集成制造系统,2021,27(02):374-389.

[37]阮锴燚,寇子明,王彦栋,等.矿井提升系统数字孪生快速建模方法研究[J/OL].煤炭科学技术:1-12[2023-04-20].

[38]张帆,王世雄,田睿芳,等.基于数字孪生的智能安全煤场管控系统设计[J].煤矿机械, 2022,43(07):1-4.

[39]赵沁平,李帅,宋震,等.虚拟生理人体建模与仿真关键技术研究进展[J].中国科学基金, 2022,36(02):187-197.

[40]王璋奇,黄增浩,葛永庆,等.线路作业人员可穿戴健康与安全监测装置研究[J].广东电力,2015,28(09):66-71.

[41]庄向,李津,陈建龙,等.2019年福建省重点职业病监测苯作业人员职业健康检查结果分析[J].慢性病学杂志,2021,22(06):845-849.

[42]叶锦娇,温良,王红尧,等.下井人员生命体征传感器的设计[J].工矿自动化, 2013,39(01):52-54.

[43]甘建哲.广州市放射工作人员健康监测系统的研究[J].科技创新导报,2017,14(33):130-132.

[44]王鹏,温良,武先利,等.煤矿救护队员生命体征监测技术[J].煤矿安全,2013,44(11):98-99+103.

[45]张捷,钱虹,周宏远,等.数字孪生技术在社区老年人安全健康监测领域的应用探究[J].中国医疗器械杂志,2019,43(06):410-413+421.

[46]Halldin C N, Wolfe A L, Beeckman-Wagner L A F, et al. Expanding A National Program Of Respiratory Health Surveillance For Coal Miners-Update On The Implementation Of New Requirements For Coal Miner Health Surveillance In The United States[M]//C56. OCCUPATIONAL MEDICINE AND LUNG FUNCTION. American Thoracic Society, 2016: A5449-A5449..

[47]张晨光,李丽,范耀春,等.公共场所空气甲醛对从业人员健康风险评估[J].环境卫生学杂志,2022,12(08):606-610.

[48]赵大勇,李素敏,王伟,等.某高锰酸钾生产企业遗留重金属污染调查分析与健康风险评估[J].广东化工,2022,49(14):104-107.

[49]朱亮亮,吴琨,史伟伟,等.汽车制造企业焊装车间噪声健康风险评估[J].职业卫生与应急救援,2022,40(03):311-314.

[50]任倩,王博.护理专业实习生心理健康风险评估研究[J].职业卫生与应急救援, 2022,40(01):32-38.

[51]孙正,曹亚非,刘峰,等.离子型稀土矿原地浸矿场地分类及人体健康风险预测[J/OL].土壤通报:1-10[2023-03-20].

[52]Zhang S, Sun X, Liu N, et al. Human Health Risk Prediction Method of Regional Atmospheric Environmental Pollution Sources Based on PMF and PCA Analysis under Artificial Intelligence Cloud Model[J]. International Journal of Analytical Chemistry, 2022, 2022.

[53]李国为,田水承.煤矿岗前人员不安全状态智能检测系统研究——以红柳林煤矿为例[J].中国安全生产科学技术,2023,19(02):106-113.

[54]洪韵,马勇,陈榕,等.基于机器学习算法的冠心病风险预测及其特征分析模型[J].广东医科大学学报,2023,41(01):53-57.

[55]Li K W, Chu J C, Chen C C. Strength decrease, perceived physical exertion and endurance time for backpacking tasks[J]. International Journal of Environmental Research and Public Health, 2019, 16(7): 1296.

[56]Yi C, Tang F, Peng L, et al. Modeling of maximum endurance time for static pulling tasks[J]. Work, 2018, 60(3): 455-463.

[57]Liu B, Ma L, Chen C, et al. Experimental validation of a subject-specific maximum endurance time model[J]. Ergonomics, 2018, 61(6): 806-817.

[58]Riemann B L, Davis S E, Huet K, et al. Intersession reliability of upper extremity isokinetic push-pull testing[J]. International Journal of Sports Physical Therapy, 2016, 11(1): 85.

[59]Calé-Benzoor M, Dickstein R, Arnon M, et al. Dynamic push–pull characteristics at three hand-reach envelopes: Applications for the workplace[J]. Applied ergonomics, 2016, 52: 216-221.

[60]李静,王庆,王胜利,等.改进FSM步态检测的行人航位推算方法[J].大地测量与地球动力学,2023,43(02):164-167+202.

[61]许颖.视觉显示终端（VDT）作业疲劳影响因素研究[D].西南交通大学,2020.

[62]Liu T, Xia M, Hong Q et al. The modelling and application of cross-scale human behavior in realizing the shop-floor digital twin. Digital Twin 2021, 1:11

[63]张博超,杨朝,郭立泉,等.基于机器学习的慢性阻塞性肺疾病急性加重预测模型的研究[J].中国康复理论与实践,2022,28(06):678-683.

[64]段书麟.基于5M1E分析法的RQ国有企业内部审计质量控制问题研究[D].杭州；浙江理工大学,2021.

[65]陈意.分级诊疗模式下患者就医行为与机构选择研究[D].合肥；合肥工业大学,2019.

[66]齐兵,秦宇,李敏虹,等.基于MQTT协议扩展的IoT设备完整性监控[J].计算机系统应用,2022,31(11):68-78.

[67]李静,王庆,王胜利,等.改进FSM步态检测的行人航位推算方法[J].大地测量与地球动力学,2023,43(02):164-167+202.

[68]夏江流云.基于Stacking融合算法的企业贷款违约风险预警模型设计与应用[D].南昌；江西财经大学,2021.

[69]辜勇,杨泽昭.基于随机森林的集装箱多式联运货运量预测[J].武汉理工大学学报,2023,45(01):35-44.

[70]袁科,黄雅冰,杜展飞,等.基于混合梯度提升决策树和逻辑回归模型的分组密码算法识别方案[J].工程科学与技术,2022,54(04):218-227.

[71]Xinxue L, Chang L. A Stacking-Based Ensemble Model for Prediction of Metropolitan Bike Sharing Demand[J]. American Journal of Information Science and Technology,2023,7(2).

[72]Chen S, Liao Y, Zhao J, et al. PACVP: Prediction of Anti-Coronavirus Peptides Using A Stacking Learning Strategy with Effective Feature Representation.[J]. IEEE/ACM transactions on computational biology and bioinformatics,2023,PP.

汪云云,陈松灿.基于AUC的分类器评价和设计综述[J].模式识别与人工智能,2011,24(01):64-71.