情绪在矿工对外界信息进行感知、接收、判断与决策过程中扮演着极其重要的角色，情绪状态会直接影响个人能力的发挥水平，及时检测人员的情绪状态是避免事故发生的一种有效手段。脑电信号（EEG）能够直接、客观地反映出人类的情绪状态，得益于测量设备与人工智能算法的快速发展，基于脑电信号的情绪识别研究应用越来越广泛，这为煤矿领域员工情绪分级和控制提供了新的思路和发展方向。目前，不同情绪状态下脑电特征规律还不明晰，多从情绪类型进行EEG信号情绪识别，对于情绪分级的研究较少。因此，本文针对以上不足展开研究，主要研究内容和成果如下：

首先，阐述了情绪及情绪脑电测量的研究现状，为后续的实验设计与数据分析奠定了理论基础，认为脑电信号的频域特征与人的情感状态有紧密关系，支持向量机在小样本识别中具有较高的准确率和鲁棒性。

其次，进行了基于脑电信号的情绪测量实验及数据分析研究。通过实验采集了高兴、中性、悲伤、恐惧、愤怒5种情绪下的脑电样本、自评量表数据和注意力实验数据。通过对注意力实验数据的分析发现负性情绪对矿工的注意力集中能力存在显著负面影响，注意力集中能力可以作为情绪分级的依据。使用SPSS和MATLAB软件对17个通道的不同情绪下α、β、θ频段的功率谱密度和频谱不对称指数进行分析，发现恐惧情绪下的功率谱密度最大，其次是悲伤情绪，愤怒情绪的功率谱密度较小，前额区、左额区的SASI指数的差异性最为明显，β频段的显著性较差，并筛选了10个具有显著差异的脑电特征的通道。

最后，结合支持向量机建立矿工情绪分级模型。对自评量表数据和注意力实验数据进行了系统聚类分析，根据聚类结果将情绪状态分为良好、一般、较差三个等级，以情绪状态分级结果作为模型的数据标签。根据脑电数据分析结果筛选了10个通道的θ和α频段功率谱密度以及前额区、左额区的频谱不对称指数作为特征值，利用支持向量机建立了矿工情绪分级模型，并比较了选择不同核函数时的模型准确率，最终选择径向基核函数建立情绪分级模型，准确率达到76%以上。

综上，本文通过基于脑电信号的情绪测量实验，得到了不同情绪状态下的脑电信号的频域特征规律性，证明提取的特征值和通道能够包含丰富的情绪特征，所提取的特征值可以作为一组稳定的特征值应用于矿工的情绪分级识别中。本研究能够有效减少在脑电信号采集过程中所需的通道数量，这可以为后续可穿戴设备的研发提供思路。基于支持向量机的情绪分级模型识别准确率较高，可以应用到煤矿企业进行矿工情绪状态的分级预测中。

Emotion plays an extremely important role in the process of miners' perception, reception, judgment and decision-making of external information. Emotional state will directly affect the level of personal ability. Timely detection of the emotional state of personnel is an effective means to avoid accidents. EEG signals can directly and objectively reflect the emotional state of human beings. Due to the rapid development of measuring instrument and artificial intelligence algorithms, the research and application of emotion recognition by EEG signals is more and more widespread. Control provides new ideas and development directions. At present, the laws of EEG characteristics in different emotional states are still unclear, and there are few studies on emotional grading. Therefore, this paper starts research based on the above research deficiencies, the main research content and results are as follows:

First, the research history and current situation of emotion and emotion recognition were described. The brain mechanism of emotion generation and the characteristics and analysis methods of EEG signals were sorted out, which laid a theoretical foundation for subsequent experimental design and data analysis. It is believed that the frequency domain characteristics of EEG signals are closely related to the emotional state of people, and emotion recognition by support vector machines is accurate and robust.

Secondly, the emotion measurement experiment and data analysis research based on EEG signals were carried out. EEG samples, subjective scale data and Attention experimental data were collected in happiness, neutrality, sadness, fear, and anger emotional states. Through the analysis of the attention experimental data, it is found that negative emotions have a significant negative impact on the concentration ability of miners, and the concentration ability can be used as the basis for emotion classification. Using SPSS and MATLAB software to analyze the power spectral density and spectral asymmetry of the α, β, θ frequency bands under different emotions of 17 channels, it is found that the power spectral density of the fear emotion is the largest, followed by the sad emotion, The power spectral density of the anger emotion is the smallest, the significance of the β band is poor. 10 channels with significant differences in EEG characteristics was selected.

Finally, combining the support vector machine to establish the miner emotional grading model. A hierarchical clustering analysis was carried out on the emotional self-evaluation scale data and the attention experimental data. According to the clustering results, the emotional state was divided into three levels: good, normal, and poor, and the emotional state grading result was used as the data label of the model. According to the results of EEG data analysis, the power spectral density of the θ and α bands of 10 channels and the spectral asymmetry index of the forehead and left frontal area were selected as characteristic values. A support vector machine was used to establish a grading model of miners’ emotions. The model accuracy rate under different kernel functions were compared, and finally the radial basis function kernel is selected to establish the emotional grading model, and the accuracy rate reaches more than 76%.

In summary, through experimental measurement of miners’ negative emotion based on EEG, this article find the Characteristic regular pattern of EEG signals in different emotional states, and proves that the extracted feature values ​​and channels can contain rich emotional features, and the extracted feature values ​​can be used in the emotion classification and recognition of miners as a set of stable feature values. This research can effectively reduce the number of channels required in the process of EEG signal acquisition, which can provide ideas for the subsequent research and development of wearable devices. The emotion classification model based on support vector machines has a high recognition accuracy and can be applied to coal mining enterprises for emotional grading.

[1] 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.

[2] 宁成浩.基于马尔科夫链的2030年中国煤炭生产格局预测[J].中国煤炭,2017,43(01):11-15.

[3] LEGAULT G, CLEMENT A L, KENNY G P, et al. Cognitive consequences of sleep deprivation,shiftwork,and heat exposure for underground miners[J]. Applied Ergonomics, 2017:58,144-150.

[4] 田水承,景国勋.安全管理学[M].北京:机械工业出版社,2009.

[5] 王国彤,何刚.安全心理学在安全生产监督中的应用[J].化工管理,2019(14):71-72.

[6] 杨雪,冯念青,张瀚元,杨娟.情感事件视角矿工不安全行为影响因素SD仿真[J].煤矿安全,2020,51(03):252-256.

[7] 邢宝君,唐水清,李乃文,等.基于SC-IAT的矿工内隐安全态度研究[J].中国安全科学学报,2018,28(05):18-23.

[8] Wsaton D, Clark LA. Development and validation of brief measures of positive and negative affect:the PANAS scales[J]. Journal of Personality and Social Psychology, 2013,54(6):1063-1073.

[9] Radenhansen R A, Anker J M. Efects of depressed mood induction on reasoning performance[J]. Perceptual and motor skill,1988,66(3):855-860.

[10] 陶爱华,戚迪.负性情绪对驾驶员风险决策的影响[J].交通医学,2016,30(1):39-42.

[11] 王春雪,吕淑然.情绪对建筑工人故意违章行为影响研究[J].工业安全与环保, 2017,43(7):58-61.

[12] 高静,王成军,李发本,等.基于安全氛围视角的建筑工人消极情绪与组织安全绩效的机理研究[J].施工技术,2018,47(17):127-132.

[13] 李乃文,李倩文,李玉涵.工作不安全感、情绪耗竭与安全注意力的关系[J].科技促进发展,2019,15(08):872-877.

[14] 惠璐.不安全情绪对矿工风险决策影响的实验研究[D].西安:西安科技大学,2018.

[15] 李乃文,郭利霞.基于Multi-agent的矿工情绪稳定性模型构建[J].中国安全生产科学技术,2014,10(12):172-177.

[16] Kim K W, Lim H C, Park J H, et al. Developing a Basic Scale for Workers' Psychological Burden from the Perspective of Occupational Safety and Health[J]. Safety and Health at Work,2018,9(2):224-231.

[17] 覃文波.地铁施工不安全行为的情绪作用机理与实证研究[D].武汉:华中科技大学,2019.

[18] 庄锦英.决策心理学[M].上海:上海教育出版.2006,2-3.

[19] Robinson M D, Clore G L. Episodic and semantic knowledge in emotional self-report: Evidence for two judgment processes[J]. Journal of Personality & Social Psychology, 2002,83(1):198-215.

[20] 孟昭兰.情绪心理学[M].北京大学出版社,2005,(03):256-257.

[21] 陈婉仪,郝凯灵,周守珍.情绪体验测量方法研究综述[C].荆楚学术2017年第7期(总第十五期):海归智库(武汉)战略投资管理有限公司,2017:289-293.

[22] 葛燕,陈亚楠,刘艳芳,等.电生理测量在用户体验中的应用[J].心理科学进展,2014,22(06):959-967.

[23] Caponecchia Carlo, Sheils Ian. Perceptions of personal vulnerability to workplace hazards in the Australian construction industry[J]. Journal of Safety Research,2011, 42(4):253-258.

[24] 蒋小梅,张俊然,陈富琴,等.基于J48决策树分类器的情绪识别与结果分析[J].计算机工程与设计,2017,38(03):761-767.

[25] Jauniaux J, Tessier M H, Regueiro S, et al. Emotion regulation of others' positive and negative emotions is related to distinct patterns of heart rate variability and situational empathy[J]. PLOS ONE,2020,15(12):e0244427.

[26] 白学军,刘颖,章鹏,等.非情绪与情绪背景下行为抑制时外侧前额叶的功能分化:一项fNIRS研究[J].心理与行为研究,2015,13(05):606-613+620.

[27] Hu Xin, Zhuang Chu, Wang Fei, et al. FNIRS Evidence for Recognizably Different Positive Emotions[J]. Frontiers in human neuroscience,2019,13:120.

[28] Duan L, Dam N, Ai H, et al. Intrinsic organization of cortical networks predicts state anxiety: an functional near-infrared spectroscopy (fNIRS) study[J]. Translational Psychiatry,2020,10:402.

[29] Duan L, Feng Q, Xu P. Using Functional Near-Infrared Spectroscopy to Assess Brain Activation Evoked by Guilt and Shame[J]. Frontiers in Human Neuroscience, 2020,14:197.

[30] Yan Wenhua,Zhang Meng,Liu Yuting. Regulatory effect of drawing on negative emotion: A functional near-infrared spectroscopy study[J]. The Arts in Psychotherapy, 2021,74.

[31] 蔡阿燕,杨洁敏,许爽,袁加锦.表达抑制调节负性情绪的男性优势——来自事件相关电位的证据[J].心理学报,2016,48(05):482-494.

[32] 戴逸翔,王雪,戴鹏,等.面向可穿戴多模传感网络的栈式自编码器优化情绪识别[J].计算机学报,2017,40(08):1750-1763.

[33] 陈蕾.恐惧情绪下面部识别注意偏向的性别差异研究[D].南京:南京师范大学,2016.

[34] 张波.连续对话语音愤怒情绪检测算法研究[D].呼和浩特:内蒙古大学，2018.

[35] Bublatzky Florian, Kavcıoğlu Fatih, Guerra Pedro,et al. Contextual information resolves uncertainty about ambiguous facial emotions: Behavioral and magnetoencephalographic correlates[J]. NeuroImage,2020,215.

[36] 王丽岩,陈梦飞,王洪彪.面部表情还是肢体语言:如何判断网球运动员比赛情境中的情绪变化?[J].中国体育科技,2021,57(01):72-80.

[37] 冯晓婷,丁月恒,顾锦.基于EEG的情感识别[J].科技视界,2017(03):114+72.

[38] Hernandez J, Mc Duff D, Benavides X, et al. Auto Emotive: Bringing empathy to the driving experience to manage stress[C]. Proceedings of the 2014 Companion Publication on Designing Interactive Systems,pp53-56.

[39] Harmon-Jones E, Gable P A, Peterson C K. The role of asymmetric frontal cortical activity in emotion-related phenomena: A review and update[J]. Biological Psychology, 2010, 84(3):451-462.

[40] ZHENG W, ZHU J, LU B. Identifying stable patterns over time for emotion recognition from EEG[J]. IEEE transactions on affective computing,2019,10(3):417–429.

[41] ZHENG W, LIU W, LU Y, et al. Emotionmeter: A mul timodal framework for recognizing human emotions[J]. IEEE transactions on cybernetics,2018,49(3):1110–1122.

[42] YAN X, ZHENG W, LIU W, et al. Investigating Gender differences of brain areas in emotion recognition using LSTM neural network[C]//Poceedings of the Internation al Conference on Neural Information Processing.Guang zhou,China,2017:820−829

[43] Nie D, Wang X W, Shi L C, et al. EEG-based emotion recognition during watching movies[C]. //Neural Engineering (NER),2011 5th International IEEE/EMBS Conference on.IEEE,2011:667-670.

[44] Reza K, Michel H, Abdul W, et al. The dynamic emotion recognition system based on functional connectivity of brain regions[C]. 2010 IEEE Intelligent Vehicles Symposium, 2010:377-381.

[45] Lokannavar S, Lahane P, Gangurde A, et al. Emotion recognition using EEG signals[J]. Emotion,2015,4(5).

[46] Arunkumar N, Kumar K R, Venkataraman V. Entropy Features for Focal EEG and Non Focal EEG[J]. Journal of Computational Science,2018,27(JUL.):440-444.

[47] LU H M, WAN M, SANGAIAH A K. Human Emotion Recognition Using an EEG Cloud Computing Platform[J]. Mobile Networks & Applications,2020,25(3):1023-1032.

[48] WANG M, ZHANG S Z, LV Y J, et al. Anxiety Level Detection Using BCI of Miner's Smart Helmet[J]. Mobile Networks & Applications,2018,23(2):336-343.

[49] Li WJ, Li YJ, Cao D. The effectiveness of emotion cognitive reappraisal as measured by self-reported response and its link to EEG alpha asymmetry[J]. BEHAVIOURAL BRAIN RESEARCH,2021,400:113042.

[50] Gao YY, Wang XK, Potter T, et al. Single-trial EEG emotion recognition using Granger Causality/Transfer Entropy analysis[J]. JOURNAL OF NEUROSCIENCE METHODS, 2021,346:108904.

[51] Gupta V, Priya T, Yadav A K, et al. Automated detection of focal EEG signals using features extracted from flexible analytic wavelet transform[J]. Pattern Recognition Letters, 2017,94(jul.15):180-188.

[52] PETRANTONAKIS P C, HADJILEONTIADIS L J. Emotion Recognition From EEG Using Higher Order Crossings[J]. IEEE Transactions on Information Technology in Biomedicine A Publication of the IEEE Engineering in Medicine & Biology Society, 2010,14(2):186-197.

[53] 李幼军,黄佳进,王海渊,钟宁.基于SAE和LSTM RNN的多模态生理信号融合和情感识别研究[J].通信学报,2017,38(12):109-120.

[54] 李幼军,钟宁,黄佳进,等.基于高斯核函数支持向量机的脑电信号时频特征情感多类识别[J].北京工业大学学报,2018,44(02):234-243.

[55] Hou H R, Zhang X N, Meng Q H. Odor-induced emotion recognition based on average frequency band division of EEG signals[J]. JOURNAL OF NEUROSCIENCE METHODS, 2020,334:108599 .‎

[56] Khateeb M, Anwar S M, Alnowami M. Multi-Domain Feature Fusion for Emotion Classification Using DEAP Dataset[J]. IEEE Access,2021,9:12134-12142.

[57] 张韩,杨济民.基于深度学习的脑电信号特征识别[J].电脑知识与技术, 2018,14(05):206-208.

[58] Keelawat Panayu, Thammasan Nattapong, Numao Masayuki, et al. A Comparative Study of Window Size and Channel Arrangement on EEG-Emotion Recognition Using Deep CNN[J]. Sensors,2021,21(5).

[59] 阚威.基于脑电的情绪识别研究与系统开发[D].南京:南京邮电大学,2019.

[60] Yin YQ, Zheng XW, Hu B, et al. EEG emotion recognition using fusion model of graph convolutional neural networks and LSTM[J]. Applied Soft Computing,2021,100:106954.

[61] James W. What is an emotion?[J]. Mind,1884(34):188-205.

[62] Robert Plutchik. The Nature of Emotions[J]. American Scientist,2001,89(4):344.

[63] Ekman Paul. An argument for basic emotions[J]. Cognition & Emotion,1992.6(3-4): 169-200.

[64] Izard C E. Basic emotions, relations among emotions, and emotion-cognition relations[J]. Psychological Review,1992,99(3):561-5.

[65] Lang P J, Bradley M M, Cuthbert B N. International affective picture system (IAPS): Technical manual and affective ratings[J]. NIMH Center for the Study of Emotion and Attention, 1997: 39-58.

[66] Mehrabian A, Russell J A. An approach to environmental psychology[M]. the MIT Press, 1974.

[67] Bradley M M, Lang P J. Measuring emotion: the self-assessment manikin and the semantic differential[J]. Journal of behavior therapy and experimental psychiatry,1994, 25(1):49-59.

[68] 任通.基于视觉刺激的脑电信号情绪识别研究[D].杭州:杭州电子科技大学,2017.

[69] Tajadura-Jiménez A, Larsson P, Väljamäe A, et al. When room size matters: acoustic influences on emotional responses to sounds[J]. Emotion,2010,10(3):416.

[70] 刘涛生,罗跃嘉,马慧,等.本土化情绪声音库的编制和评定[J].心理科学,2006, 29(2):406-408.

[71] 段若男.基于脑电信号的视频诱发情绪识别[D].上海:上海交通大学, 2014.

[72] Epple G, Herz R S. Ambient odors associated to failure influence cognitive performance in children[J]. Developmental Psychobiology,1999,35(2):103-7.

[73] 李志学.基于多模生理信号的精神疲劳检测系统的设计与研究[D].兰州:兰州大学,2018.

[74] Loukas M, Pennell C, Groat C, et al. Korbinian Brodmann (1868-1918) and His Contributions to Mapping the Cerebral Cortex[J]. Neurosurgery,2011,68(1):6-11.

[75] Chanel G, Kronegg J, Grandjean D, et al. Emotion assessment: Arousal evaluation using EEG’s and peripheral physiological signals[C]//International Workshop on Multimedia Content Representation,Classification and Security. Springer Berlin Heidelberg,2006: 530-537.

[76] Martini N, Menicucci D, Sebastiani L, et al. The dynamics of EEG gamma responses to unpleasant visual stimuli: From local activity to functional connectivity[J]. Neuro Image, 2012,60(2):922-932.

[77] 李焕.情绪与矿工不安全行为关系实验研究[D].西安:西安科技大学,2015.

[78] 王璐.基于脑电实验的情绪对矿工冒险行为影响研究[D].西安:西安科技大学,2016.

[79] Han KeTsung. Effects of Three Levels of Green Exercise, Physical and Social Environments, Personality Traits, Physical Activity, and Engagement with Nature on Emotions and Attention[J]. Sustainability,2021,13(5).

[80] K.R. Anne, S. Kuchibhotla, H.D. Vankayalapati. Emotion Recognition Using Spectral Features[J]. Springer International Publishing,2015.

[81] Aftanas L I, Reva N V, Savotina L N, et al. Neurophysiological Correlates of Induced Discrete Emotions in Humans: An Individually Oriented Analysis[J]. Neuroscience & Behavioral Physiology,2004,90(12):1457-71.

[82] Aftanas L I, Pavlov S V. Trait anxiety impact on posterior activation asymmetries at rest and during evoked negative emotions:EEG investigation[J]. International Journal of Psychophysiology,2005,55(1):85-94.

[83] Rusalova M N, Kostyunina M B, Kulikov M A. Spatial distribution of coefficients of asymmetry of brain bioelectrical activity during the experiencing of negative emotions[J]. Neuroscience & Behavioral Physiology,2003,33(7):703-6.

[84] DUAN R N, WANG X W, LU B L. EEG-based emotion recognition in listening music by using support vector machine and linear dynamic system[M]. Neural Information Processing Springer Berlin Heidelberg,2012.

[85] Duan R N, Zhu J Y, Lu B L. Differential entropy feature for EEG-based emotion classification[C]. 6th International IEEE/EMBS Conference on Neural Engineering (NER) SanDiego,California,USA:IEEE,2013.

[86] Bekkedal M Y, Rd R J, Panksepp J. Human brain EEG indices of emotions: delineating responses to affective vocalizations by measuring frontal theta event-related synchronization[J].Neuroscience & Biobehavioral,2011,35(9):1959-1970.

[87] 焦凯强,王湖斐,郭茂田.脑电信号中的频谱不对称指数特征与情绪识别[J].科学技术与工程,2018,18(17):145-149.

[88] JS H. On the integration of discontinuous functions[J]. Proceedings of the London Mathematical Society,1874,1(6):140–153.

[89] 李立.基于脑电信号样本熵的情感识别[Ｄ].太原:太原理工大学,2014.