支护工在煤炭开采工作中主要负责掘进工作面巷道帮部和顶部的支护。由于复杂的工作环境和单调重复的工作内容，支护工容易产生部分脑力疲劳和体力疲劳，疲劳会导致其工作效率降低，产生不安全心理和不安全行为。因此，及时检测支护工的疲劳是预防煤矿事故发生一种有效的手段。脑电信号作为检测疲劳的“金标准”，能有效识别疲劳的发生。然而，目前缺乏从实际工作环境和工作内容出发的支护工疲劳识别研究。鉴于此，本研究采用脑电方法开展了关于支护工工作疲劳产生情况识别研究。主要研究内容和结论如下：

首先，设计了支护工疲劳诱发实验。本研究通过对当前疲劳检测的国内外研究现状、支护工工作环境和工作内容的梳理，设计了采用长时间跑步运动负荷增加后的单调重复按键行为来模拟支护工体力疲劳和部分脑力疲劳的产生。在诱发过程中采用RPE量表和心率来实时记录疲劳诱发情况，并在诱发前后采用Oddball范式记录被试的行为数据指标，同时使用脑电设备同步记录被试脑电信号的变化。

其次，分析了诱发疲劳前后支护工的脑电信号和行为数据变化。通过SPSS25.0软件对诱发疲劳过程中RPE分数、心率以及诱发疲劳前后所采集到的行为能力指标进行处理。结果表明，在疲劳诱发过程中，被试的RPE分数和心率的变化趋势一致，且显著增加。在疲劳恢复过程中，被试的RPE分数仍大于疲劳诱发前，表明疲劳诱发成功。疲劳诱发后，被试重复按键行为的平均反应时、错误率和遗漏次数会增加。通过MATLAB软件对脑电数据进行处理，结果表明疲劳诱发后，δ波在F7、FT7、T7、TP7通道所在的颞区会显著减少，θ波在F7、P6通道所在的颞区会显著性增加，ɑ波在全脑会增加，但效果不显著，β波在FPZ、AF3、F1通道所在的大脑额区会显著减少。

最后，运用诱发疲劳过程中采集的数据完成支护工疲劳分级，建立支护工疲劳等级识别模型。结合所采集到的脑电信号运用专注度和放松度对疲劳进行分级，通过记录的RPE分数和行为能力指标运用K-means聚类法对疲劳进行分级，两次分级结果一致。将支护工疲劳分为三个等级：正常工作状态、中度疲劳状态、重度疲劳状态。采用随机森林方法建立支护工疲劳识别模型，选取疲劳后有显著性变化的指标作为模型的输入值，划分的三个疲劳等级作为输出值，并对该模型进行训练和测试，模型训练集的正确率为100%，测试集的正确率为81.82%。引入模型测试混淆矩阵对本研究疲劳识别模型进行评价，结果显示支护工疲劳识别模型对三类疲劳等级识别的整体性能较好。

综上，本研究通过对支护工疲劳诱发实验的设计，采集并分析了疲劳诱发前后被试的脑电信号和行为能力指标的变化。基于此，对支护工的疲劳等级进行了划分，建立的支护工疲劳识别模型有效检测了三个疲劳等级的产生，为支护工疲劳的识别和检测提供理论依据和参考。

The supporter is mainly responsible for the support of the side and top of the roadway in the coal mining work. Due to the complex working environment and monotonous and repetitive work content, support workers are prone to some mental fatigue and physical fatigue. Fatigue will lead to reduce work efficiency and unsafe psychology and unsafe behavior. Therefore, timely detection of fatigue of support workers is an effective means to prevent coal mine accidents. As the ' gold standard ' for detecting fatigue, EEG signals can effectively identify the occurrence of fatigue. However, there is lack of research on fatigue identification of support workers from the actual working environment and work content. In view of this, this study use EEG method to carry out research on the identification of support workers ' work fatigue. The main research contents and conclusions are as follows :

Firstly, the fatigue induction experiment of support workers is designed. In this study, through combing the current research status of fatigue detection at home and abroad, the working environment and work content of the support workers, a monotonous repetitive button behavior after increasing the load of long-term running exercise is designed to simulate the physical fatigue and partial mental fatigue of the support workers. During the induction process, the RPE scale and heart rate are used to record the fatigue induction in real time, and the Oddball paradigm is used to record the behavioral data of the subjects before and after the induction. At the same time, the changes of EEG signals of the subjects are recorded synchronously by EEG equipment.

Secondly, the changes of EEG signals and behavioral data of support workers before and after fatigue are analyzed. The RPE score, heart rate and behavioral ability indexes collect before and after fatigue are processed by SPSS25.0 software. The results show that during the fatigue induction process, the RPE score and heart rate of the subjects have the same trend and increase significantly. In the process of fatigue recovery, the RPE score of the subjects is still greater than that before fatigue induction, indicating that fatigue induction is successful. After fatigue induction, the average reaction time, error rate and omission times of repeat keystroke behavior increase. The EEG data are processed by MATLAB software. The results show that after fatigue induction, the δ wave in the temporal area of F7, FT7, T7 and TP7 channels decrease significantly, the θ wave in the temporal area of F7 and P6 channels increase significantly, the ɑ wave increase in the whole brain, but the effect is not significant, and the β wave in the frontal area of FPZ, AF3 and F1 channels decrease significantly.

Finally, the fatigue classification of support workers is completed by using the data collected during the induced fatigue process, and the fatigue grade identification model of support workers is established. Combine with the collected EEG signals, the fatigue is graded by concentration and relaxation, and the fatigue is graded by K-means clustering method through the recorded RPE scores and behavioral ability indicators. The two grading results are consistent. The fatigue of support workers is divided into three grades : normal working state, moderate fatigue state and severe fatigue state. The random forest method is used to establish the fatigue identification model of support workers. The indexes with significant changes after fatigue are selected as the input values of the model, and the three fatigue levels are divided as the output values. The model is trained and tested. The correct rate of the model training set is 100 %, and the correct rate of the test set is 81.82 %. The model test confusion matrix is introduced to evaluate the fatigue identification model of this study. The results show that the overall performance of the supporter fatigue identification model for the three types of fatigue level identification is better.

In summary, this study collect and analyze the changes of EEG signals and behavioral ability indexes of subjects before and after fatigue induction by designing the fatigue induction experiment of support workers. Based on this, the fatigue level of the supporter is divided, and the fatigue identification model of the supporter is established to effectively predict the generation of three fatigue levels, which provides a theoretical basis and reference for the identification and detection of the fatigue of the supporter.

[1]杨文清.我国煤炭需求组合预测模型及应用[J].西安科技大学学报,2014,34(05):553-557.

[2]Wang L L, Cao Q G, Zhou L J. Research on the Influencing Factors in Coal Mine Production Safety based on the Combination of DEMATEL and ISM[J]. Safety Science, 2018, 103, 51-61.

[3]田水承,王雪晨,苗彦平,等.基于扎根理论的矿工不安全状态影响因素研究[J].煤矿安全,2022,53(02):252-256.

[4]田水承,景国勋.安全管理学[M].北京:机械出版社,2022.

[5]张培森,李复兴,朱慧聪,等.2008—2020年煤矿事故统计分析及防范对策[J].矿业安全与环保,2022,49(01):128-134.

[6]任大伟,徐小华,梁亚敏,等.煤矿井下作业疲劳影响因素及对策研究[J].煤矿安全,2019,50(03):231-234.

[7]武静蕾.煤矿掘进过程中安全问题与工作[J].山西化工,2021,41(02):109-110+113.

[8]邢媛媛,王恩元,陈燕文.基于可变模糊识别理论的煤矿掘进工作面环境评价的研究[J].矿业安全与环保,2019,46(05):109-113.

[9]陈博.综掘工作面支护作业疲劳及工间休息研究[D].中国矿业大学,2020.

[10]陈晓琳,陈立,郭鑫禾,等.井下支护作业人员颈部肌肉疲劳特性研究[J].中国安全生产科学技术,2022,18(06):92-97.

[11]董帆.基于OpencvOpenCV软件部队疲劳驾驶预警系统的设计[J].电子技术与软件工程,2016(02):61.

[12]He J, Zhang Y, Zhang C, et al. A Noninvasive Real-time Solution for Driving Fatigue Detection based on Left Prefrontal EEG and Eye Blink[C]//Brain Informatics and Health: International Conference, BIH 2016, Omaha, NE, USA, October 13-16, 2016 Proceedings. Springer International Publishing, 2016: 325-335.

[13]Zeng S, Li J, Jiang L, et al. A Driving Assistant Safety Method based on Human Eye Fatigue Detection[C]//2017 29th Chinese Control And Decision Conference (CCDC). IEEE, 2017: 6370-6377.

[14]杨渝书,姚振强,李增勇,等.心电图时频域指标在驾驶疲劳评价中的有效性研究[J].机械设计与制造,2002(05):94-95.

[15]Jamaluddin F N, Ibrahim F, Ahmad S A. A New Approach to Noninvasive-Prolonged Fatigue Identification based on Surface EMG Time-Frequency and Wavelet Features[J]. Journal of Healthcare Engineering, 2023, 2023: 1951165-1951165.

[16]Balandong P R, Ahmad F R, Saad M N M, et al. A Review on EEG-based Automatic Sleepiness Detection Systems for Driver[J]. IEEE Access, 2018, 6: 22908-22919.

[17]程如中,赵勇,戴勇,等.基于Adaboost方法的车载嵌入式疲劳驾驶预警系统[J].北京大学学报(自然科学版),2012,48(05):719-726.

[18]Patel M, Lal S, Kavanagh D, et al. Applying Neural Network Analysis on Heart Rate Variability Data to Assess Driver Fatigue[J]. Expert Systems with Applications, 2011, 38(6): 7235-7242.

[19]徐礼胜,张闻勖,庞宇轩,等.基于短时心电信号的疲劳驾驶检测算法[J].东北大学学报(自然科学版),2019,40(07):937-941.

[20]吴雪,王娆芬.基于迁移学习的深层卷积神经网络心电信号疲劳分类[J].中国医学物理学杂志,2021,38(10):1258-1263.

[21]Eglė B, Aleksėjus M, Liepa B. ECG Signal Features Classification for the Mental Fatigue Recognition[J]. Mathematics, 2022, 10(18): 3395-3395.

[22]宋海燕,张建国,王珺,等.基于表面肌电的人体背部负重行走肌肉疲劳特性研究[J].生物医学工程学杂志,2016,33(3):426-430.

[23]Fu R R, Wang H. Detection of Driving Fatigue by Using Noncontact EMG and ECG Signals Measurement System[J]. International Journal of Neural Systems, 2014, 24(3): 1450006.

[24]王越,张驎.矿工肌肉疲劳状况及事故预防研究[J].中国安全科学学报,2021,31(03):191-196.

[25]Kikuchi M, Fukushima K. Invariant Pattern Recognition with Eye Movement: A Neural Network Model[J]. Neurocomputing, 2001, 38: 1359-1365..

[26]Lyapunov S I, Shoshina I I, Lyapunov I S. Tremor Eye Movements as an Objective Marker of Driver’s Fatigue[J]. Human Physiology, 2022, 48(1): 71-77.

[27]王磊宇,武淑红,李海芳.基于眼部行为的驾驶疲劳评价指标研究[J].太原理工大学学报,2015,46(04):440-443.

[28]张志文,付永庆.基于眼球运动状态检测的疲劳预警系统研究[J].计算机与数字工程,2016,44(02):255-258.

[29]Zhang C, Yu X. Estimating Mental Fatigue based on Electroencephalogram and Heart Rate Variability[J]. Polish Journal of Medical Physics and Engineering, 2010, 16(2): 67-84.

[30]Borghini G, Vecchiato G, Toppi J, et al. Assessment of Mental Fatigue during Car Driving by Using High Resolution EEG Activity and Neurophysiologic Indices[C]//2012 Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE, 2012: 6442-6445.

[31]叶柠,孙宇舸.基于EEG小波包子带能量比的疲劳驾驶检测方法[J].东北大学学报(自然科学版),2012,33(08):1088-1092.

[32]金纯,曾伟.基于脑电的实时疲劳监测算法的研究[J].科学技术与工程,2015,15(34):231-234+241.

[33]Chai R, Naik G, Nguyun T N, et al. Driver Fatigue Classification with Independent Component by Entropy Rate Bound Minimization Analysis in an EEG-based System[J]. IEEE Journal of Biomedical and Health Informatics, 2016, 21(3): 715-724.

[34]Wang H T, Dragomir A, Abbasi N I, et al. A Novel Real-time Driving Fatigue Detection System based on Wireless Dry EEG[J]. Cognitive Neurodynamics, 2018, 12(4): 365-376.

[35]徐军莉,王平,穆振东.融合眼动和脑电特征的疲劳驾驶检测研究[J].重庆交通大学学报(自然科学版),2021,40(12):7-11.

[36]田水承,胥静,田方圆,等.基于EEG的夜班矿工疲劳检测[J].安全与环境学报,2022,22(04):2034-2040.

[37]周继红,杨傲,袁丹凤,等.疲劳驾驶及其评估方法进展[J].伤害医学(电子版),2021,10(03):45-50.

[38]唐美霞,何勇.基于双眼定位与状态判决疲劳检测算法[J].计算机工程与设计,2018,39(06):1750-1754+1787.

[39]洪子梦,周秀,邬小鲁.一种基于眼睛长宽比的船舵手疲劳检测方法[J].电子测试,2021(07):71-73.

[40]冯晓锋.基于融合眼嘴头部特征的疲劳驾驶检测方法[J].安全与环境学报,2022,22(01):263-270.

[41]Chu J W, Jin L S, Tong B L, et al. A Monitoring Method of Driver Mouth Behavior based on Machine Vision[A]. In IEEE Intelligent Vehicles Symposium[C]. Parma, Italy: IEEE Inc, 2004: 351-356.

[42]Choi I H, Kim Y G. Head Pose and Gaze Direction Tracking for Detecting a Drowsy Driver[C]. 2014 International Conference on Big Data and Smart Computing (BIGCOMP). Bangkok, Thailand, 2014: 241-244.

[43]Wu S L, Suo Z H, Chen H D, et al. Road Rage Detection Algorithm based on Fatigue Driving and Facial Feature Point Location[J]. Neural Computing and Applications, 2022, 34(15): 12361-12371.

[44]田水承,陈盈,邹元.手工搬运作业下矿工腰部肌肉疲劳分析[J].矿业安全与环保,2021,48(05):122-126.

[45]孙瑞山,王平.基于结构方程模型的飞行员疲劳影响因素研究[J].安全与环境学报,2022,22(06):3252-3258.

[46]谈宇琦,韩豫.建筑工人心理疲劳的形成与影响因素[J].土木工程与管理学报,2022,39(04):162-169.

[47]王雅荣,张宁.集成DEMATEL-ISM的SEM模型快递分拣人员作业疲劳影响因素测定[J].内蒙古科技大学学报,2021,40(04):326-332.

[48]郭伏,王天博,宁作江,等.单调作业中单调感对职业疲劳的影响研究[J].工业工程与管理,2017,22(1):134-140.

[49]田水承,乌力吉,寇猛,等.基于生理实验的矿工不安全行为与疲劳关系研究[J].西安科技大学学报,2016,36(03):324-330.

[50]田水承,丁洋,匡秘姈.疲劳中介效应下矿工心理因素对不安全行为的影响[J].西安科技大学学报,2023,43(01):47-54.

[51]贾宝柱,谢灯峰,林叶锦.海员疲劳及其影响因素[J].中国航海,2018,41(04):94-100.

[52]Zuraida R, Abbas S B. The Factors Influencing Fatigue Related to the Accident of Intercity Bus Drivers in Indonesia[J]. International Journal of Technology, 2020, 11(2), 342-352.

[53]Aicha F, Pierreval H, Gien D, et al. Taking into Account Worker' Fatigue in Production Tasks: a Combined Simulation Framework[M]. The 9th Eurosim Congress On Modelling and Simulation. 2016: 590-596.

[54]Akhtar J M, Utne B I. Human Fatigue's Effect on the Risk of Maritime Groundings-A Bayesian Network Modeling Approach[J]. Safety Science, 2014, 62: 427-440.

[55]Biggs C H, Biggs H, Dingsdag D, et al. Fatigue Factors Affecting Metropolitan Bus Drivers: A Qualitative Investigation[J]. Work, 2009, 32(1), 5-10.

[56]Knauth P. Extended Work Periods[J]. Industrial Health, 2007, 45(1): 125-136.

[57]Tourangeau A E, Cranley L A. Nurse Intention to Remain Employed: Understanding and Strengthening Determinants[J]. Journal of Advanced Nursing, 2006, 55(4): 497-509.

[58]王卫军.综采工作面人─机系统可靠性与作业环境的研究[J].中国安全科学学报,1998(04):16-19.

[59]景国勋,李创起,彭信山.掘进工作面复杂环境因素与人因伤亡事故的关联度分析[J].安全与环境学报,2012,12(01):238-240.

[60]韦善阳,杨涛.基于人-机-环境耦合效应的掘进工作面安全高效生产[J].华北科技学院学报,2014,11(07):10-13.

[61]徐明伟,金龙哲,田兴华,等.受限空间不同照度环境下VDT作业视觉疲劳[J].工程科学学报,2020,42(12):1605-1612.

[62]周克元.综合评价温度高度及湿度对人体舒适度的影响分析—以徐州市为例[J].保山学院学报,2019,38(02):19-22.

[63]刘厚金,俞佩镐,林嘉壁.矿山粉尘浓度标准和测尘技术发展动向[J].煤矿安全技术,1982(03):46-52.

[64]景国勋,王远声,周霏,等.短期噪声暴露对作业人员心率的影响[J].工业安全与环保,2019,45(08):70-73.

[65]韩帅,李安邦,李兴东,等.煤矿巷道掘进工序和时序优化研究[J].煤炭工程,2018,50(02):51-54+58.

[66]胡臻.机务维修人员疲劳风险指数研究[D].中国民航大学,2016.

[67]Abd-Elfattah M H, Abdelazeim H F, Elshennawy S. Physical and Cognitive Consequences of Fatigue: A Review[J]. Journal of Advanced Research, 2015, 6(3): 351-358.

[68]Chen M L, Lu S Y, Mao I F. Subjective Symptoms and Physiological Measures of Fatigue in Air Traffic Controllers[J]. International Journal of Industrial Ergonomics, 2019, 70: 1-8.

[69]Grandjean E. Fatigue in industry[J]. British Journal of Industrial Medicine, 1979, 36(3): 175-186.

[70]蔡启明.以动态心率为指标的体力疲劳的评价方法研究[J].人类工效学,1999(01):28-30+72.

[71]Tanaka M, Ishii A.Watanabe Y. Neural Effect of Mental Fatigue on Physical Fatigue: A magnetoencephalography study[J]. Brain Research, 2014. 4(2): 104-114.

[72]张浩,徐志云.生物起搏:电信号的产生与传导[J].国际心血管病杂志,2008(01):6-9.

[73]Zeidman L A, Stone J, Kondziella D. New Revelations about Hans Berger, Father of the Electroencephalogram (EEG), and His Ties to the Third Reich[J]. Journal of Child Neurology, 2014, 29(7): 1002-1010.

[74]张淞杰.基于脑电信号识别的驾驶疲劳检测方法研究[D].杭州电子科技大学,2018.

[75]彭军强,吴平东,殷罡.疲劳驾驶的脑电特性探索[J].北京理工大学学报,2007(07):585-589.

[76]王黎,于涛,闻邦椿.基于脑电α波的非线性参数人体疲劳状态判定[J].东北大学学报,2005,(12):1174-1177.

[77]Fouladi D B, Mohammadi A, Nematpour L. Mental Fatigue Assessment Using Recording Brain Signals: Electroencephalography[J]. Journal of Ergonomics, 2019, 7(2): 45-53.

[78]姚娟娟,路堃,马兴宇,等.基于脑电和眼电的疲劳检测方法的研究[J].电子设计工程,2020,28(06):115-120.

[79]钟文潇,安兴伟,狄洋,等.基于脑电信号的身份特征提取方法研究综述[J]. 生物医学工程学杂志,2021,38(06):1203-1210.

[80]杨刚,杨小青,钟欣.基于短时傅里叶变换和时域包络的AEB系统预警声信号时间提取方法[J].汽车技术,2021,No.554(11):42-48.

[81]陈彦龙,毛万丽,刘鸿优.主观疲劳量表(RPE)评估足球运动员训练负荷的实证研究[J].体育科研,2019,40(01):57-63.

[82]景国勋,吕鹏飞,周霏,等.基于E-Prime的煤矿噪声对矿工反应时与出错数的影响效应研究[J].安全与环境学报,2018,18(06):2285-2289.

[83]Eston R G, Faulkner J A, Mason E A, et al. The Validity of Predicting Maximal Oxygen Uptake from Perceptually Regulated Graded Exercise Tests of Different Durations[J]. European Journal of Applied Physiology, 2006, 97(5): 535-541.

[84]杨圣韬,尹晓峰,高炳宏.应用主观疲劳量表量化运动负荷的研究进展[J].体育科研2019,40(05):85-94.

[85]孙盛燕,杨兵.递增负荷运动时RPB、心率和血乳酸的关系[J].湖州师范学院学报,2015,37(02):63-68.

[86]Akerstedt T, Gillberg M. Subjective and Objective Sleepiness in the Active Individual[J]. International Journal of Neuroscience, 1990, 52(1-2): 29-37.

[87]Belyavin A, Wright N A. Changes in Electrical Activity of the Brain with Vigilance[J]. Electroencephalography and Clinical Neurophysiology, 1987, 66(2): 137-144.

[88]王福旺,王宏.长途客车驾驶员疲劳状态脑电特征分析[J].仪器仪表学报,2013,34(5):1146-1152.

[89]Roelands B, Meeusen R. The Relation between Brain Neurotransmitter Concentrations and Supraspinal Fatigue[J]. Plos One, 2013, 8(10): 53631.

[90]董建梁,李鹏,马晨光.基于脑电检测的煤矿矿工井下疲劳作业监测研究[J].中国科技信息,2014(16):82-83.

[91]He J, Liu D, Wan Z, et al. A Noninvasive Real-time Driving Fatigue Detection Technology based on Left Prefrontal Attention and Meditation EEG[C]// International Conference on Multisensor Fusion and Information Integration for Intelligent Systems. IEEE, 2014: 1-6.

[92]廖红强,邱勇,杨侠,等.对应用层次分析法确定权重系数的探讨[J].机械工程师, 2012(06):22-25.

[93]王莲芬,许树柏.层次分析法引论[M].北京:中国人民大学出版社,1990.

[94]田俊峰.关于煤矿安全风险的评价及预警研究[J].中国包装工业,2015(14):105-106+109.

[95]李松峰,冯旭杰,沙茜,等.城市轨道交通运营安全事故分级方法[J].交通运输研究,2020,6(05):66-74.

[96]Anil K J. Data Clustering: 50 years beyond K-Means[J]. Pattern Recognition Letters, 2010, 31(8): 651-666.

[97]Al-kababchee S G M, Algamal Z Y, Qasim O S. Enhancement of K-means Clustering in Big Data based on Equilibrium Optimizer Algorithm[J]. Journal of Intelligent Systems, 2023, 32(1): 20220230.

[98]Breiman. Random Forests [J]. Machine Learning, 2001, 45(1): 5-32.

[99]陈伽洛,陈龙然.决策树与随机森林[J].信息与电脑(理论版),2019,31(17):43-45.