~煤矿生产活动中“人”的因素依然是制约煤矿安全的首要因素之一，95%以上的煤矿事故均由人的不安全心理和行为引起。矿工作为高风险行业的工作群体，在进行生产工作时，除了需熟练掌握技术操作，严格遵守相关规章制度外，更需具备良好的行为能力与工作状态。为了正确判断并有效识别矿工岗前的状态，达到预防和控制矿工不安全行为的目的。本研究选取不安全状态中对矿工负面影响较为显著的不安全情绪状态，将面部表情作为情绪检验指标，借助现代计算机技术，构建了快速辨识矿工岗前状态的系统，以期有效预防煤矿人因事故的发生。论文的主要研究内容及结果如下：
开展了不安全状态与情绪理论研究。通过对不安全行为、事故致因理论等相关分析界定了矿工的不安全状态的内涵，并选取不安全状态中较为重要的不安全情绪状态，阐述了情绪与面部表情识别的相关理论基础。
开展了矿工不安全状态的实验及分级研究。设计并开展不同情绪下矿工的注意力集中能力实验，从情绪视角来进行矿工不安全状态的分级。通过对主观问卷和行为数据分析，得到情绪对矿工的注意力能力有显著影响，可以将注意力能力作为不安全状态的分类依据。运用K-means聚类方法，对20个被试在不同情绪下的脱靶次数和在靶时间两个注意力指标进行聚类分析，从情绪类型和强度两个维度将矿工的岗前状态分为“安全状态”、“较不安全状态”以及“不安全状态”三个类别。“安全状态”包括“中性、高兴 （中）”两类情绪；“较不安全状态”包括“厌恶（中）、愤怒（中）、恐惧（中）、愤怒（中）”；“不安全状态”包括“高兴（强）、厌恶（强）、愤怒（强）、恐惧（强）、愤怒（强）”。
构建了矿工岗前状态的辨识系统。首先，基于PyQt5设计用户界面，用以完成人与系统的交互。其次，结合CNN与SVM算法构建了矿工不安全状态识别模型，通过RAF-DB数据集训练CNN模型，根据实验过程中采集到的11种情绪的面部表情制作了数据集Emotion，以Emotion数据集中的样本为输入数据，以矿工的岗前状态分类结果为数据标签，通过CNN-SVM算法实现了矿工情绪的识别，将情绪识别结果输出至状态分类模块，得到了矿工的岗前状态的等级，输出至用户界面并储存。最后，通过实例验证，系统整体的准确率为66%，平均识别速度为3.8ms，具有良好的识别效果。
综上，本文实现了矿工不安全状态的分级，以状态分类结果为依据，构建了基于面部表情识别的矿工不安全状态辨识系统，该系统准确率高，鲁棒性好，可应用于煤矿企业对矿工岗前状态的监测中。

~The "human" factor in coal mine production is still one of the primary factors restricting coal mine safety. More than 95% of coal mine accidents are caused by people's unsafe psychology and behavior. As a working group in high-risk industries, miners should not only master technical operation skillfully and strictly abide by relevant rules and regulations, but also have good behavior ability and working status when carrying out production work.In order to correctly judge and identify the miners pre-service status effectively and achieve the purpose of prevention and control of miners' unsafe behavior. In order to effectively prevent the occurrence of human accidents in coal mines, this research selected the unsafe state of the miners considerably negative impact of unsafe emotional state, the facial expression as a mood test indicators, using the modern computer technology, to build a state of rapid identification of miners pre-service system,.The main research contents and results of this paper are as follows
The theory of unsafe state and emotion is studied. Based on the analysis of unsafe behavior and accident causation theory, this paper defines the connotation of the unsafe state of miners, selects the more important unsafe emotional state in the unsafe state, and expounds the related theoretical basis of emotion and facial expression recognition.
To carry out the experiment and classification research on the unsafe state of miners.The experiment of attention ability of miners under different emotions was designed and carried out, and the unsafe status of miners was classified from the perspective of emotion.Through the analysis of subjective questionnaire and behavioral data, it is found that emotion has a significant influence on the attention ability of miners, and the attention ability can be used as the classification basis of the unsafe state.K-means clustering method was used to conduct cluster analysis on the attention indexes of miss times and target time of 20 subjects under different emotions. The pre-work status of miners was divided into three categories from the two dimensions of emotion type and intensity: "safe state", "less safe state" and "unsafe state"."Safe" includes "neutral" and "happy (middle)" emotions."Less secure state" includes "disgust (middle), anger (middle), fear (middle), anger (middle)";"Insecure states" include "happiness (strong), disgust (strong), anger (strong), fear (strong), anger (strong)."
The identification system of miners' pre-work status is constructed.Firstly, the user interface is designed based on PyQT5 to complete the interaction between the human and the system.Secondly, CNN and SVM algorithms are combined to build the miners' unsafe state recognition model. The RAF-DB data set is used to train the CNN model. The data set Emotion is made according to the facial expressions of 11 emotions collected during the experiment, with the samples in the Emotion data set as input data.Taking the classification result of miners' pre-post status as the data label, the emotion recognition of miners is realized through CNN-SVM algorithm, and the emotion recognition result is output to the status classification module to obtain the level of miners' pre-post status, which is output to the user interface and stored.Finally, through the example verification, the overall accuracy of the system is 66%, the average recognition speed is 3.8ms, has a good recognition effect.
In conclusion, this paper has realized the classification of the unsafe status of miners. Based on the status classification results, an identification system of the unsafe status of miners based on facial expression recognition has been built. This system has high accuracy and good robustness, and can be applied to the monitoring of the status of miners before they work in coal mining enterprises.

[1]国家煤炭工业网.安全生产[EB.OL].[2021-1-26].www.coalchina.org.cn

[2] 张慧,王冬雪,王启飞.2005—2016年我国较大及以上煤矿事故特征分析[J].安全与环境学报,2019,19(05):1847-1852.

[3] 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.

[4] Randolph R F, Cecala A B.. Bureau of Mines human factors and management researchprogram[C].International Conference on Reliability, Production, and Control in Coal Mines, Wollongong,1991.

[5] Zhang S. Study and exploration on the relationship between miners' psychology safety in production[C].Proceedings of the 1998 International Symposium on Safety Science and Technology, ISSST, Beijing, China, 1998.

[6] 郑莹. 煤矿员工不安全行为的心理因素分析及对策研究[D].河北理工大学,2008.

[7] 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].华中科技大学,2019.

[8] Helbing D,Farkas I,Vicsek T. Simulating dynamical features of escape panic.[J]. Nature,2000,407(6803).

[9] Jing-hong Wang,Wen-yu Yan,You-ran Zhi,Jun-cheng Jiang. Investigation of the Panic Psychology and Behaviors of Evacuation Crowds in Subway Emergencies[J]. Procedia Engineering,2016,135.

[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] 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.

[12] 李乃文,冀永红,唐水清等.高危岗位矿工情绪稳定性影响机制SD仿真[J].中国安全生产科学技术,2018,14(05):167-173.

[13] 崔宏.矿工反生产行为演化机理仿真研究[D].辽宁工程技术大学,2016.

[14] 田水承,杨鹏飞,李磊等.矿工不良情绪影响因素及干预对策研究[J].矿业安全与环保, 2016,43(6): 99-102.

[15] 田水承,孔维静,况云等.矿工心理因素、工作压力反应和不安全行为关系研究[J].中国安全生产科学技术,2018,14(08):106-111.

[16] 王浩,蒋承林,史莉莉.煤矿井下噪声危害分析及对策[J].中国安全生产科学技术,2011,7(12):183-187.

[17] 代丽芳.矿工情绪管理研究[D].阜新：辽宁工程技术大学，2012.

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

[19] 张良斐. 恐慌情绪下矿工应急行为研究[D].辽宁工程技术大学,2014.

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

[21] 李乃文,张文文,牛莉霞.领导非权变惩罚对矿工不安全行为的影响研究[J].中国安全科学学报,2019,29(12):1-6.

[22] 李乃文,牛莉霞.矿工工作倦怠、不安全心理与不安全行为的结构模型[J].中国心理卫生杂志,2010,24(03):236-240.

[23] Friesen E, Ekman P. Facial action coding system: a technique for the measurement of facial movement[J]. Palo Alto, 1978, 3.

[24] Mase K,Pentland A.Automatic lipreading by by optical-flow analysis[J].Systems and Computers in Japan,1991,22(6):67-76.

[25] Li H , Roivainen, P , Forchheimer, R . 3D motion estimation in model based facial image coding[J]. IEEE Transaction on Pettern Analysis and Machine Intelligence,1993,15(6): 545-555.

[26] K Mase.Recognition of Facial Expression from Optical Flow, IEICE Transactions, 1991,(74): 3474-3483

[27] 周书仁，梁昔明，朱灿等.基于ICA与HMM的表情识别[J].中国图象图形学报A, 2008, 13(12).

[28] Shan C,Gong S,Mcowan P W.Facial expression recognition based on Local Binary Patterns:A comprehensive study[J].Image and Vision Computing,2009,27(6):803-816.

[29] 付晓峰,韦巍.基于多尺度中心化二值模式的人脸表情识别[J].控制理论与应用,2009,26(6): 629-633.

[30] M. V. Jonnalagedda,D. D. Doye. An Efficient and Simple Approach to Extract Features for Facial Expression Recognition[J]. International Journal of Mathematics & Computation™,2015,27(4).

[31] Asim Munir,Ayyaz Hussain,Sajid Ali Khan,et al. Illumination invariant facial expression recognition using selected merged binary patterns for real world images[J]. Optik,2018,158.

[32] Wang Shangfei,Pan Bowen,Chen Huaping,Ji Qiang. Thermal Augmented Expression Recognition.[J]. IEEE transactions on cybernetics,2018,48(7).

[33] Aditya Harit,J. C. Joshi Col,K. K. Gupta. Facial Emotions Recognition using Gabor Transform and Facial Animation Parameters with Neural Networks[J]. IOP Conference Series: Materials Science and Engineering,2018,331(1).

[34] Cohn J F,Zlochower A J,Lien J,et al.Automated face analysis by feature point tracking has high concurrent validity with manual FACS coding[J].Psychophysiology,1999, 36(1):35-43

[35] Doranaika F,Davoine F.Simultaneous facial action tracking and expression recognition in the presence of head motion[J]. International Journal of Computer Vision,2008,76(3):257-281.

[36] Jagdish Lal Raheja,Umesh Kumar. Human Facial Expression Detection From Detected in Captured Image using Back Propagation Neural Network[J]. International Journal of Computer Science & Information Technology,2010,2(1).

[37] Heinrich W H,Peterson D,Roos N.Industrial accident prevention[M].New York:McGraw-Hill Book Company,1980:22-23.

[38] 孟现飞, 丁恩杰, 刘全龙, 等. 危险源概念的重新界定及与隐患关系研究[J].中国安全科学学报, 2017, 27(04): 55-59.

[39] 李健行,夏登友,武旭鹏.非常规突发灾害事故的演化机理与演变路径分析[J].安全与环境工程,2014,21(06):166-170.

[40] 傅贵,吴琼,李亚,王秀明.不安全动作与不安全物态的判定方法探讨[J].安全与环境学报,2017,17(03):994-998.

[41] REASON J．Human error［M］．Cambridge: Cambridge University Press，1990．

[42] Surry J. Industrial accident research: a human engineering appraisal[J].Journal of Occupational & Environmental Medicine, 1969, 11(9)

[43] 张海卿,闫明仁.安全疏散中人的行为研究及模糊综合评价[J].消防科学与技术,2010,29(03):252-255.

[44] Iraj Mohammadfam, Fakhradin Ghasemi, Omid Kalatpour, Abbas Moghimbeigi. Constructing a Bayesian network model for improving safety behavior of employees at workplaces[J]. Applied Ergonomics,2017,58.

[45] 田水承,孔维静,况云等.矿工心理因素、工作压力反应和不安全行为关系研究[J].中国安全生产科学技术,2018,14(08):106-111.

[46] 黄莺,瑚珊,姚思梦.施工塔吊安全管理的系统动力学分析[J].安全与环境学报,2020,20(06):2060-2068.

[47] 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.

[48] 田水承,郭方艺,杨鹏飞.不良情绪对胶轮车驾驶员不安全行为的影响研究[J].矿业安全与环保,2018,45(05):115-119+125.

[49] Dacher Keltner,Disa Sauter,Jessica Tracy,Alan Cowen. Emotional Expression: Advances in Basic Emotion Theory[J]. Journal of Nonverbal Behavior,2019,43(2).

[50] James W.What is an emotion[J].Mind,1884,9(34)188－205．

[51] Oatley K.,Keltner D,Jenkins J.Understanding emotions[M].Oxford:Blackwell publishing, 2006: 24.

[52] 孟昭兰.体验是情绪的心理实体——个体情绪发展的理论探讨[J].应用心理学,2000, 21(2): 48-52.

[53] Buck R. Prime theory: An integrated view of motivation and emotion[J]. Psychological Review, 1985, 92(92):389-413.

[54] Gaetano Valenza,Antonio Lanata,Enzo Pasaquale Scilinggo.The role of nonlinear dynamics in affective valence and arousal recognition[J].IEEE Transactions on affective computing,2012,3(2):237-249.

[55] Wundt W.Principles of physiological psychology[M].New York:Macmillan Press.

[56] Izard C E.Human emotions[M].New York:Plenum Press.

[57] Lang PJ, Bradley MM, Cuthbert B. International Affective Picture System (IAPS):Technical Manual And Affective Ratings [M]. Gainesville: University of Florida，1999．

[58] Russell J A.The psychology of facial expresssion[M].New York:Cambridge University Press,1997.

[59] Plutchik R . A general psychoevolutionary theory of emotion[J]. Emotion Theory Research & Experience, 2000, 21(4-5):529-553.

[60] 王志平. 杜邦公司的安全信念与管理实践[J].外国经济与管理,2004,26(4):20-24.

[61] 曹庆仁,李爽,宋学锋. 煤矿员工的知-能-行不安全行为模式研究[J]. 中国安全科学学

报,2007,17(12): 19-25.

[62] 周刚,程卫民,诸葛福民.人因失误与人的不安全行为相关原理的分析与探讨[J].中国安全科学学报, 2008,18(3):10-14.

[63] 冯涛.基于情绪与建筑工人不安全行为影响关系的安全管理策略研究[D].西安建筑科技大学,2017.

[64] 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]. 华中科技大学, 2019.

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

[66] 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.

[67] 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.

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

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

[70] Picard R W,Vyzas E,Healey J.Toward Machine Emotional Intelligence:Analysis of Affective Physiological State (HTML)[J].IEEE Transactions on Pattern Analysis & Machine Intelligence,2001,23(10):1175-1191.

[71] Ekman P, Friesen W V .Facial action coding system :a technique for the measurement of facial movement [ M] .Palo Alto:Consulting Psychologists Press, 1978

[72] Ekman P,Davidson R J,Friesen W V. The Duchenne smile: emotional expression and brain physiology. II.[J]. Journal of personality and social psychology,1990,58(2).

[73] Mauss Iris B,Levenson Robert W,McCarter Loren,Wilhelm Frank H,Gross James J. The tie that binds? Coherence among emotion experience, behavior, and physiology.[J]. Emotion (Washington, D.C.),2005,5(2).

[74] Fraser W. Smith,Philippe G. Schyns. Smile Through Your Fear and Sadness: Transmitting and Identifying Facial Expression Signals Over a Range of Viewing Distances[J]. Psychological Science,2009,20(10).

[75] 孟昭兰.为什么面部表情可以作为情绪研究的客观指标[J].心理学报,1987(02):124-134.

[76] 刘静,任艳萍,屈威等.中国人面孔表情强度分级图片库的建立[J].中国心理卫生杂志,2019,33(02):120-125.