基于支持向量机的管制疲劳检测模型研究
|
摘要
管制疲劳是影响民航安全的重大隐患,准确检测疲劳状态是进行疲劳预警、降低疲劳风险的关键。提出应用支持向量机模型融合多生理参数和眼动指标进行管制疲劳检测,通过MP150多导生理记录仪与眼动仪,采集模拟管制试验中正常与剥夺睡眠状态下被试的试验数据,同时记录其Karolinska疲劳等级和操作绩效。结果表明,RR间期、LF/HF、快慢波比值、PERCLOS和扫视速度均与管制疲劳呈较强相关,利用支持向量机融合五项指标构建管制疲劳检测模型,对于五级疲劳度的识别准确率为78. 1%,判断正常组与剥夺睡眠组的准确率为94. 2%。
In this paper,the author intends to apply a support vector machine model to detecting the air traffic controller's fatigue by fusing the multiple physiological parameters and the eye movement indexes for the situation of the civil air traffic. As is well kown,the air traffic controller's fatigue is likely to be a major factor leading to the hazards of the safety of the civil aviation,whereas the timely warning and reminding to detect such fatigue stae should be the key to reducing the related risk. For such a purpose,we have built up an experimental platform based on the simulated tower control software,an eye tracker and the MP150 physiological polygraph, the physiological data and the eye movement data register of the air traffic controlled practitional atthe normal or sleep-deprived state collected separately by using the MP150 physiological polygraph and eye tracker in the simulated tower control experiment,with their Karolinska fatigue level and behavior being recorded. In the process of the data analysis,it would be possible to gain the degree of fatigue of the subjects in conjunction with the Karolinska fatigue state and performance.Through the partial correlation analysis of all the physiological parameters and the eye movement indexes of the air traffic controller's fatigue,the experimental results show that significant correlation can be found with the air traffic controller's fatigue degree corresponding to the physiological parameters,the RR interval and the ratio of LF to the HF ratio of the fast and slow ware with their fatigue status-in-situ between the eye movement indexes,and the PERCLOS and saccadic velocity. Thus,the model of the air traffic controllers' fatigue detection we have built up with the support of the vector machine to fuse the above 5 indexes proves available and efficient. The sample we have prepared proves that the model of ours can be as accurate to reach up78. 1% whereas its accuracy for judging the normal and sleepdeprived group turns out to be as high as 94. 2%. Thus,the model can be expected to be used as a means for monitoring the air traffic controller's fatigue and make the fatigue early warning useful for the future,so that it can help to ensure the safety of the air traffic control and even the safety of the entire civil aviation front-line.
In this paper,the author intends to apply a support vector machine model to detecting the air traffic controller's fatigue by fusing the multiple physiological parameters and the eye movement indexes for the situation of the civil air traffic. As is well kown,the air traffic controller's fatigue is likely to be a major factor leading to the hazards of the safety of the civil aviation,whereas the timely warning and reminding to detect such fatigue stae should be the key to reducing the related risk. For such a purpose,we have built up an experimental platform based on the simulated tower control software,an eye tracker and the MP150 physiological polygraph, the physiological data and the eye movement data register of the air traffic controlled practitional atthe normal or sleep-deprived state collected separately by using the MP150 physiological polygraph and eye tracker in the simulated tower control experiment,with their Karolinska fatigue level and behavior being recorded. In the process of the data analysis,it would be possible to gain the degree of fatigue of the subjects in conjunction with the Karolinska fatigue state and performance.Through the partial correlation analysis of all the physiological parameters and the eye movement indexes of the air traffic controller's fatigue,the experimental results show that significant correlation can be found with the air traffic controller's fatigue degree corresponding to the physiological parameters,the RR interval and the ratio of LF to the HF ratio of the fast and slow ware with their fatigue status-in-situ between the eye movement indexes,and the PERCLOS and saccadic velocity. Thus,the model of the air traffic controllers' fatigue detection we have built up with the support of the vector machine to fuse the above 5 indexes proves available and efficient. The sample we have prepared proves that the model of ours can be as accurate to reach up78. 1% whereas its accuracy for judging the normal and sleepdeprived group turns out to be as high as 94. 2%. Thus,the model can be expected to be used as a means for monitoring the air traffic controller's fatigue and make the fatigue early warning useful for the future,so that it can help to ensure the safety of the air traffic control and even the safety of the entire civil aviation front-line.
引文
[1] MITLER M M,CARSKADON M A,CZEISLER C A,etal. Catastrophes,sleep,and public policy:consensus re-port[J]. Sleep,1988,11(1):100-109.
[2] WILLIAMSON A,LOMBARDI D A,FOLKARD S,etal. The link between fatigue and safety[J]. AccidentAnalysis&Prevention,2011,43(2):498-515.
[3] SUN Tao(孙涛),CHEN Yu(陈宇). The managementand prevention of ATC fatigue[J]. Air Traffic Manage-ment(空中交通管理),2005(5):4-10.
[4] DELLA ROCCO P S. The role of shift work and fatigue inair traffic control operational errors and incidents[R].Oklahoma City:Federal Aviation Administration DC Of-fice of Aviation Medicine,1999:1-21.
[5] LAL S K,CRAIG A. A critical review of the psychophys-iology of driver fatigue[J]. Biological Psychology,2001,55(3):173-194.
[6] ROSKE-HOFSTRAND R. Raising awareness for fatigue a-mong air traffic controllers[C]//JENSEN R S,RAK-OVAN L A. Proceedings of the Eighth International Sym-posium on Aviation Psychology. Columbus, OH:OhioState University,1995:985-991.
[7] WILLIAMSON A,LOMBARDI D A,FOLKARD S,etal. The link between fatigue and safety[J]. Accident A-nalysis&Prevention,2011,43(2):498-515.
[8] Federal Aviation Administration Press Office. Pressrelease—The FAA announces additional staffing at 27control tower[EB/OL].[2011-04-13]. http://www. faa. gov/news/press_releases/news_story. cfm?newsId=12664
[9] DASARI D,CROWE C,LING C,et al. EEG patternanalysis for physiological indicators of mental fatigue insimulated air traffic control tasks[C]//Proceedings of theHuman Factors and Ergonomics Society Annual Meeting.Santa Monica,CA:Sage Publications,2010:205-209.
[10] WIYOR H D,NTUEN C A,STEPHENS J D W,et al.Classifying visual fatigue severity based on neurophysio-logical signals and psychophysiological ratings[J]. Inter-national Journal of Human Factors and Ergonomics,2013,2(1):11-32.
[11] CHARBONNIER S,ROY R N,BONNET S,et al. EEGindex for control operators’ mental fatigue monitoringusing interactions between brain regions[J]. Expert Sys-tems with Applications,2016,52(C):91-98.
[12] CHEN Z,XU X,ZHANG J,et al. Application of LC-MS-based global metabolomic profiling methods to humanmental fatigue[J]. Analytical Chemistry, 2016, 88(23):11293-11296.
[13] WANG Lei(汪磊),SUN Ruishan(孙瑞山). Study onface feature recognition-based fatigue monitoring methodfor air traffic controller[J]. China Safety Science Journal(中国安全科学学报),2012,22(7):66-71.
[14] FUKUI T,MORIOKA T. The blink method as an as-sessment of fatigue[J]. Ergonomics,1971,14(1):23-30.
[15] DI-STASI L L,MCCAMY M B,CATENA A,et al. Mi-crosaccade and drift dynamics reflect mental fatigue[J].European Journal of Neuroscience, 2013, 38(3):2389-2398.
[16] SUN Ruishan(孙瑞山),MA Guangfu(马广福),YUAN Leping(袁乐平). Analysis of risk of controllerfatigue based on characteristics of speech reaction time[J]. China Safety Science Journal(中国安全科学学报),2016,26(12):7-12.
[17] MAO Zhe(毛喆),CHU Xiumin(初秀民),YAN Xin-ping(严新平),et al. Advances of fatigue detecting tech-nology for drivers[J]. China Safety Science Journal(中国安全科学学报),2005,15(3):108-112.
[18] GRANDJEAN E P,WOTZKA G,SCHAAD R,et al.Fatigue and stress in air traffic controllers[J]. Ergonom-ics,1971,14(1):159-165.
[19] ZHAO C,ZHAO M,LIU J,et al. Electroencephalo-gram and electrocardiograph assessment of mental fatiguein a driving simulator[J]. Accident Analysis and Preven-tion,2012,45(1):83-90.
[20] FAN Xiaoli(范晓丽),NIU Haiyan(牛海燕),ZHOUQianxiang(周前祥),et al. Mental fatigue characteris-tics based on EEG analysis[J]. Journal of Beijing Uni-versity of Aeronautics and Astronautics(北京航空航天大学学报),2016,42(7):1406-1413.
[21] LAL S K,CRAIG A,BOORD P,et al. Development ofan algorithm for an EEG-based driver fatigue counter-measure[J]. Journal of Safety Research,2003,34(3):321-328.
[22] DIAZPIEDRA C,RIEIRO H,SUREZ J,et al. Fa-tigue in the military:towards a fatigue detection testbased on the saccadic velocity[J]. Physiological Meas-urement,2016,37(9):62-75.
[23] HU Shuyan(胡淑燕),ZHENG Gangtie(郑钢铁).Driver fatigue prediction with eyelid related parametersby support vector machine[J]. Journal of Beijing Uni-versity of Aeronautics and Astronautics(北京航空航天大学学报),2009,35(8):929-932.
[24] XU Chuan(胥川),WANG Xuesong(王雪松),CHENXiaohong(陈小鸿). Evaluating performance of non-intrusive indicators on drowsy driving detection[J].Journal of Southwest Jiaotong University(西南交通大学学报),2014,49(4):720-726.
[25] GUO Mingwei(郭明玮),ZHAO Yuzhou(赵宇宙),XIANG Junping(项俊平),et al. Review of object de-tection methods based on SVM[J]. Control and Decision(控制与决策),2014(2):193-200.
[2] WILLIAMSON A,LOMBARDI D A,FOLKARD S,etal. The link between fatigue and safety[J]. AccidentAnalysis&Prevention,2011,43(2):498-515.
[3] SUN Tao(孙涛),CHEN Yu(陈宇). The managementand prevention of ATC fatigue[J]. Air Traffic Manage-ment(空中交通管理),2005(5):4-10.
[4] DELLA ROCCO P S. The role of shift work and fatigue inair traffic control operational errors and incidents[R].Oklahoma City:Federal Aviation Administration DC Of-fice of Aviation Medicine,1999:1-21.
[5] LAL S K,CRAIG A. A critical review of the psychophys-iology of driver fatigue[J]. Biological Psychology,2001,55(3):173-194.
[6] ROSKE-HOFSTRAND R. Raising awareness for fatigue a-mong air traffic controllers[C]//JENSEN R S,RAK-OVAN L A. Proceedings of the Eighth International Sym-posium on Aviation Psychology. Columbus, OH:OhioState University,1995:985-991.
[7] WILLIAMSON A,LOMBARDI D A,FOLKARD S,etal. The link between fatigue and safety[J]. Accident A-nalysis&Prevention,2011,43(2):498-515.
[8] Federal Aviation Administration Press Office. Pressrelease—The FAA announces additional staffing at 27control tower[EB/OL].[2011-04-13]. http://www. faa. gov/news/press_releases/news_story. cfm?newsId=12664
[9] DASARI D,CROWE C,LING C,et al. EEG patternanalysis for physiological indicators of mental fatigue insimulated air traffic control tasks[C]//Proceedings of theHuman Factors and Ergonomics Society Annual Meeting.Santa Monica,CA:Sage Publications,2010:205-209.
[10] WIYOR H D,NTUEN C A,STEPHENS J D W,et al.Classifying visual fatigue severity based on neurophysio-logical signals and psychophysiological ratings[J]. Inter-national Journal of Human Factors and Ergonomics,2013,2(1):11-32.
[11] CHARBONNIER S,ROY R N,BONNET S,et al. EEGindex for control operators’ mental fatigue monitoringusing interactions between brain regions[J]. Expert Sys-tems with Applications,2016,52(C):91-98.
[12] CHEN Z,XU X,ZHANG J,et al. Application of LC-MS-based global metabolomic profiling methods to humanmental fatigue[J]. Analytical Chemistry, 2016, 88(23):11293-11296.
[13] WANG Lei(汪磊),SUN Ruishan(孙瑞山). Study onface feature recognition-based fatigue monitoring methodfor air traffic controller[J]. China Safety Science Journal(中国安全科学学报),2012,22(7):66-71.
[14] FUKUI T,MORIOKA T. The blink method as an as-sessment of fatigue[J]. Ergonomics,1971,14(1):23-30.
[15] DI-STASI L L,MCCAMY M B,CATENA A,et al. Mi-crosaccade and drift dynamics reflect mental fatigue[J].European Journal of Neuroscience, 2013, 38(3):2389-2398.
[16] SUN Ruishan(孙瑞山),MA Guangfu(马广福),YUAN Leping(袁乐平). Analysis of risk of controllerfatigue based on characteristics of speech reaction time[J]. China Safety Science Journal(中国安全科学学报),2016,26(12):7-12.
[17] MAO Zhe(毛喆),CHU Xiumin(初秀民),YAN Xin-ping(严新平),et al. Advances of fatigue detecting tech-nology for drivers[J]. China Safety Science Journal(中国安全科学学报),2005,15(3):108-112.
[18] GRANDJEAN E P,WOTZKA G,SCHAAD R,et al.Fatigue and stress in air traffic controllers[J]. Ergonom-ics,1971,14(1):159-165.
[19] ZHAO C,ZHAO M,LIU J,et al. Electroencephalo-gram and electrocardiograph assessment of mental fatiguein a driving simulator[J]. Accident Analysis and Preven-tion,2012,45(1):83-90.
[20] FAN Xiaoli(范晓丽),NIU Haiyan(牛海燕),ZHOUQianxiang(周前祥),et al. Mental fatigue characteris-tics based on EEG analysis[J]. Journal of Beijing Uni-versity of Aeronautics and Astronautics(北京航空航天大学学报),2016,42(7):1406-1413.
[21] LAL S K,CRAIG A,BOORD P,et al. Development ofan algorithm for an EEG-based driver fatigue counter-measure[J]. Journal of Safety Research,2003,34(3):321-328.
[22] DIAZPIEDRA C,RIEIRO H,SUREZ J,et al. Fa-tigue in the military:towards a fatigue detection testbased on the saccadic velocity[J]. Physiological Meas-urement,2016,37(9):62-75.
[23] HU Shuyan(胡淑燕),ZHENG Gangtie(郑钢铁).Driver fatigue prediction with eyelid related parametersby support vector machine[J]. Journal of Beijing Uni-versity of Aeronautics and Astronautics(北京航空航天大学学报),2009,35(8):929-932.
[24] XU Chuan(胥川),WANG Xuesong(王雪松),CHENXiaohong(陈小鸿). Evaluating performance of non-intrusive indicators on drowsy driving detection[J].Journal of Southwest Jiaotong University(西南交通大学学报),2014,49(4):720-726.
[25] GUO Mingwei(郭明玮),ZHAO Yuzhou(赵宇宙),XIANG Junping(项俊平),et al. Review of object de-tection methods based on SVM[J]. Control and Decision(控制与决策),2014(2):193-200.