A Driving Fatigue Model for Extra-long Tunnels Based on Multi-source Data
-
摘要: 为研究驾驶人在特长隧道内驾驶疲劳演变过程及其影响因素,基于实车试验采集的多源数据,对特长隧道内驾驶疲劳分类判别以及驾驶疲劳影响因素关系模型展开了研究。通过差异显著性分析和相关性分析筛选出闭眼百分率P80、瞳孔直径变异系数和加速度作为疲劳敏感性指标,并分析了各指标随行驶时间累积的变化规律。为构建驾驶疲劳分类判别模型,基于卡罗林斯卡嗜睡量表(Karolinska sleeping scale,KSS)主观疲劳检测结果,将疲劳程度划分清醒状态、半疲劳状态和疲劳状态,采用构造多类分类器的方法将不同疲劳状态样本进行组合分类,利用网格搜索法进行分类模型的参数寻优,并将筛选出的疲劳敏感性指标作为分类模型的输入变量,建立了基于网格搜索法的多分类支持向量机疲劳状态判别模型(GS-M-SVMs模型)。然后根据疲劳状态分类判别模型,利用有序多分类Logistic模型建立了特长隧道疲劳程度与影响因素的关系模型,对特长隧道内驾驶疲劳影响因素进行了探究。研究结果表明:疲劳敏感性指标变化规律可有效表征特长隧道内驾驶疲劳演变过程,而GS-M-SVMs模型分类检测准确率达到90.75%,对疲劳程度的分类识别效果较好,并且累积行驶时间和隧道长度显著影响驾驶人的疲劳程度,其模型回归系数分别为2.634和0.395,表明累积行驶时间是驾驶人在特长隧道路段中疲劳程度加重的最主要因素,隧道照度和隧道线形等因素并无显著影响。
-
关键词:
- 交通安全 /
- 驾驶疲劳 /
- GS-M-SVMs模型 /
- 网格搜索法 /
- 有序多分类Logistic模型
Abstract: To investigate the evolution of driving fatigue in extra-long tunnels and its influencing factors, multi-source data from real-vehicle experiments are utilized to classify and identify driving fatigue, as well as to analyze the relationship between fatigue levels and influencing factors. Through significance tests of differences and correlation analysis, the percentage of eyelid closure over the pupil over time (PERCLOS) P80, the variable coefficient of pupil diameter, and acceleration are selected as key fatigue sensitivity indicators, and their changing patterns with accumulated driving time are examined. To construct a driving fatigue classification model, fatigue levels, based on the subjective fatigue detection results from the Karolinska sleepiness scale (KSS), are categorized into awake, semi-fatigued, and fatigued states. A multi-class classifier method is then employed to combine and classify these fatigue states. The grid-search method (GS) is utilized for parameter optimization, and the selected fatigue sensitivity indicators are used as input variables to establish a multi-class support vector machine model (GS-M-SVMs) for fatigue state classification. Following this, an ordinal multi-class Logistic model is developed to explore the relationship between driving fatigue levels and influencing factors in extra-long tunnels. The results indicate that the changing patterns of fatigue sensitivity indicators effectively capture the evolution of driving fatigue. The GS-M-SVMs model achieved a classification accuracy of 90.75%, indicating strong performance in fatigue level detection. Both accumulated driving time and tunnel length significantly influence driving fatigue levels, with regression coefficients of 2.634 and 0.395, respectively. This indicates that accumulated driving time is the primary factor contributing to increased fatigue in extra-long tunnels, while factors such as tunnel illumination and alignment do not significantly impact fatigue levels. -
表 1 2018—2022年中国特长公路隧道列表
Table 1. List of extra-long highway tunnels in China from 2018 to 2022
名称 地点 长度/m 开通年份 公路 米仓山隧道 陕西,南郑-四川,南江 13 833 2018 G85银昆高速 狮子坪隧道 四川,理县 13 156 2020 蓉昌高速 秦岭天台山隧道 陕西,宝鸡-凤县 15 560 2021 银昆高速 城开隧道 重庆,城口-开州 11 437 2022 银百高速 盐源隧道 四川,盐源 14 150 在建 都香高速 木寨岭隧道 甘肃,漳县-岷县 15 710 在建 兰海高速 武汉两湖隧道 湖北,武汉 19 450 在建 秦园路-三环线 天山胜利隧道 新疆,乌鲁木齐-和静 22 035 在建 乌若高速 金塘海底隧道 浙江,宁波-舟山 10 990 拟建 甬舟复线高速 海太长江隧道 江苏,南通-常熟 11 185 拟建 海太过江通道 五宝山隧道 云南,云龙 13 410 拟建 云泸高速 哀牢山隧道 云南,新平-镇沅 20 275 拟建 天猴高速 表 2 试验路段信息
Table 2. Test section information
工况 名称 长度/m 路段类型 1 瓦店子-铁峰山明线 13 800 明线路段 2 大树隧道群 5 825 隧道群路段 3 龙井隧道群 13 448 隧道群路段 4 瓦店子隧道 3 356 特长隧道路段 5 铁峰山隧道 11 362 特长隧道路段 6 南山隧道 4 828 特长隧道路段 表 3 Bartlett球形检验表
Table 3. Bartlett's spherical test
取样足够度的KMO度量 0.768 Bartlett球形度检验 近似c2 476.419 df 15 Sig 0.000 表 4 各检验指标分析汇总
Table 4. Summary of the analysis of each test index
检验指标 差异显著分析 相关性分析 眨眼频率 显著 中等相关 平均眨眼时间 不显著 不相关 闭眼百分率P80 显著 较强相关 平均注视时间 不显著 不相关 瞳孔直径变异系数 显著 较强相关 速度 不显著 不相关 加速度 显著 较强相关 表 5 GS-M-SVMs模型分类检测结果统计
Table 5. GS-M-SVMs model classification detection statistics
样本数量 样本类型 总样本 清醒状态样本 半疲劳状态样本 疲劳状态样本 样本总数 270 37 140 93 正确识别数 245 30 125 90 识别准确率/% 90.75 81.08 89.29 96.77 表 6 疲劳状态数据混淆矩阵
Table 6. Fatigue state data confusion matrix
疲劳状态 预测值 清醒状态 半疲劳状态 疲劳状态 实际值 清醒状态 30 7 0 半疲劳状态 12 125 3 疲劳状态 0 3 9 表 7 变量选取及编码
Table 7. Variable selection and encoding
变量类型 变量名称 单位 编码 解释变量 闭眼百分率P80 x1 % 协变量 瞳孔直径变异系数x2 协变量 加速度x3 m/s2 协变量 影响变量 累积行驶时间x4 s 协变量 工况行驶时间x5 s 协变量 隧道长度x6 m 协变量 隧道照度x7 Lux 协变量 CCR x8 °/m 协变量 是否为停车港x9 0=否,1=是 是否为明线路段x10 0=否,1=是 注:协变量为连续变量。 表 8 有序多分类Logistic模型回归结果
Table 8. Regression results of ordered multiclassification Logistic model
变量 估算值 标准误 瓦尔德 自由度 显著性 Exp(β) 阈值 [疲劳程度=1] 1.270 0.642 3.917 1 0.048 [疲劳程度=2] 6.568 0.859 58.502 1 0.000 位置 闭眼百分率P80 12.136 5.183 10.930 1 0.000 186 465.218 瞳孔直径变异系数 0.073 0.048 60.723 1 0.001 1.075 累积行驶时间 2.634 3.855 15.135 1 0.000 13.929 工况行驶时间 0.264 1.913 13.260 1 0.000 1.302 隧道长度 0.395 0.036 28.014 1 0.000 1.484 [是否为明线=0] -0.293 0.566 0.116 1 0.000 0.746 [是否为明线=1] 0a 0 注:“a”因为此参数冗余,所以将其设为0。 -
[1] 中华人民共和国交通运输部. 2022年交通运输行业发展统计公报[N]. 中国交通报, 2023-6-16(2).Ministry of Transport, People's Republic of China. 2022 Statistical bulletin on the development of the transportation Industry[N], China Communication News, 2023-6-16(2). (in Chinese) [2] 倪娜, 杨少伟, 潘兵宏, 等. 终南山公路隧道交通事故特征及预防措施研究[J]. 现代隧道技术, 2018, 55(4): 25-32.NI N, YANG S W, PAN B H, et al. Traffic accident characteristics and prevention measures for the Zhongnanshan highway tunnel[J]. Modern Tunnelling Technology, 2018, 55(4): 25-32. (in Chinese) [3] 黄发明. 基于"黑洞效应"的公路驾驶员视错觉改善方法试验研究[D]. 武汉: 武汉理工大学, 2014.HUANG F M. Experimental study on the improvement of optical illusion at highway tunnel entrance based on the"Black Hole Effect"[D]. Wuhan: Wuhan University of Technology, 2014. (in Chinese) [4] 杜志刚, 梅家林, 倪玉丹, 等. 基于视觉需求的城市水下特长隧道光环境评价与优化综述[J]. 交通运输工程学报, 2020, 20(6): 48-61.DU Z G, MEI J L, NI Y D, et al. Review on evaluation and optimization of light environment of extra-long urban underwater tunnel based on visual demands[J]. Journal of Traffic and Transportation Engineering, 2020, 20(6): 48-61. (in Chinese) [5] 段儒禹. 长距离连续隧道群运营安全预测方法及措施研究[D]. 成都: 西南交通大学, 2021.DUAN R Y. Research on operation safety prediction method and measures of long-distance continuous tunnel group[D]. Chengdu: Southwest Jiaotong University, 2021. (in Chinese) [6] 杨香丽. 单调环境下驾驶疲劳生成机理及预防对策[D]. 西安: 长安大学, 2020.YANG X L. The mechanism and preventive measures of driving fatigue under monotonous environment[D]. Xi'an: Chang'an University, 2020. (in Chinese) [7] ORON-GILAD T, RONEN A, SHINAR D. Alertness maintaining tasks(AMTs)while driving[J]. Accident Analysis & Prevention, 2008, 40: 851-860. [8] 谢晓莉. 驾驶疲劳生成机理研究[D]. 北京: 北京工业大学, 2010.XIE X L. Study on driving fatigue formation mechanism[D]. Beijing: Beijing University of Technology, 2010. (in Chinese) [9] 王连震. 机动车驾驶员疲劳累积特性及预防措施研究[D]. 哈尔滨: 哈尔滨工业大学, 2014.WANG L Z. Characteristics and prevention measures of fatigue accumulation of vehicle drivers[D]. Harbin: Harbin Institute of Technology, 2014. (in Chinese) [10] 李庆臣. 基于面部特征的疲劳驾驶检测系统设计[D]. 郑州: 郑州大学, 2019.LI Q C. Design of fatigue driving detection system based on facial feature[D]. Zhengzhou: Zhengzhou University, 2019. (in Chinese) [11] 阎莹, 袁华智, 杨香丽, 等. 基于多源数据的单调道路环境与驾驶疲劳关系模型[J]. 中国公路学报, 2021, 34(5): 156-167. doi: 10.3969/j.issn.1001-7372.2021.05.015YAN Y, YUAN H Z, YANG X L, et al. Relationship model of monotonous road environment and driving fatigue based on multi-source data[J]. China Journal of Highway and Transport, 2021, 34(5): 156-167. (in Chinese) doi: 10.3969/j.issn.1001-7372.2021.05.015 [12] 蔡素贤, 杜超坎, 周思毅, 等. 基于车辆运行数据的疲劳驾驶状态检测[J]. 交通运输系统工程与信息, 2020, 20(4): 77-82.CAI S X, DU C K, ZHOU S Y, et al. Fatigue driving state detection based on vehicle running data[J]. Journal of Transportation Systems Engineering and Information Technology, 2020, 20(4): 77-82. (in Chinese) [13] 白婧荣, 唐伯明, 孙宗元, 等. 毗邻互通立交特长隧道驾驶负荷研究[J]. 交通运输系统工程与信息, 2022, 22(1): 301-310.BAI J R, TANG B M, SUN Z Y, et al. On driving load of super-long tunnel adjacent to interchanges[J]. Journal of Transportation Systems Engineering and Information Technology, 2022, 22(1): 301-310. (in Chinese) [14] 朱兴林, 克然木·司马义, 姚亮, 等. 基于SSA-BP极端高温天气驾驶疲劳的检测[J]. 科学技术与工程, 2023, 23(3): 1254-1261. doi: 10.3969/j.issn.1671-1815.2023.03.044ZHU X L, SIMAYI K, YAO L, et al. Research on driving fatigue detection in extreme high temperature weather based on SSA-BP[J]. Science Technology and Engineering, 2023, 23(3): 1254-1261. (in Chinese) doi: 10.3969/j.issn.1671-1815.2023.03.044 [15] 梁波, 刘涛, 肖尧, 等. 特长隧道出口处交通标志与驾驶人感知关系试验[J]. 长安大学学报(自然科学版), 2020, 40(5): 48-55.LIANG B, LIU T, XIAO Y, et al. Experimental on relationship between traffic signs and drivers' perception at exit of extra-long tunnel[J]. Journal of Chang'an University(Natural Science Edition), 2020, 40(5): 48-55. (in Chinese) [16] LAI S K, CRAIG A. A critical review of the psychophysiology of driver fatigue[J]. Biological Psychology, 2001, 55(3): 173-194. doi: 10.1016/S0301-0511(00)00085-5 [17] PHILLIPS R O. A review of definitions of fatigue - and a step towards a whole definition[J]. Transportation Research Part F: Psychology & Behaviour, 2015, 29: 48-56. [18] 祝荣欣. 基于生理信号的驾驶疲劳检测研究综述[J]. 人类工效学, 2020, 26(4): 82-86.ZHU R X. A review of research on driving fatigue detection based on physiological signals[J]. Chinese Journal of Ergonomics, 2020, 26(04): 82-86. (in Chinese) [19] 胥川, 裴赛君, 王雪松. 基于无侵入测量指标的个体差异化驾驶疲劳检测[J]. 中国公路学报, 2016, 29(10): 118-125. doi: 10.3969/j.issn.1001-7372.2016.10.013XU C, PEI S J, WANG X S. Driver drowsiness detection based on non-intrusive metrics considering individual difference[J]. China Journal of Highway and Transport, 2016, 29 (10): 118-125. (in Chinese) doi: 10.3969/j.issn.1001-7372.2016.10.013 [20] 付川云. 疲劳状态下驾驶人生理及眼动特征研究[D]. 哈尔滨: 哈尔滨工业大学, 2011.FU C Y. Research on physiological and eye movement characteristics of driver under fatigue condition[D]. Harbin: Harbin Institute of Technology, 2011. (in Chinese) [21] 秦洪懋, 刘志强, 汪澎. 基于多通道信息融合的疲劳驾驶行为分析研究[J]. 中国安全科学学报, 2011, 21(2): 115-120. doi: 10.3969/j.issn.1003-3033.2011.02.020QIN H M, LIU Z Q, WANG P. Research on drowsy driving behavior based on multi-channel information fusion[J]. China Safety Science Journal(CSSJ), 2011, 21 (2): 115-120. (in Chinese) doi: 10.3969/j.issn.1003-3033.2011.02.020 [22] 王雪松, 李飞虎. 基于驾驶模拟实验的眼部指标与疲劳分级[J]. 同济大学学报(自然科学版), 2015, 43(2): 226-231, 264.WANG X S, LI F H. Eye indicators and drowsy level analysis based on driving simulator[J]. Journal of Tongji University (Nature Science), 2015, 43(2): 226-231, 264. (in Chinese) [23] 焦方通, 杜志刚, 王首硕, 等. 城市水下特长隧道出入口视觉及舒适性研究[J]. 中国公路学报, 2020, 33(6): 147-156.JIAO F T, DU Z G, WANG S S, et al. Visual characteristic and comfort at the entrance and exit of the extra-long urban underwater tunnel[J]. China Journal of Highway and Transport, 2020, 33(6): 147-156. (in Chinese) [24] 李德才. 基于数据特征的汽车行驶速度状态与驾驶行为安全性研究[D]. 哈尔滨: 东北林业大学, 2021.LI D C. Research on vehicle speed and driving safety based on data characteristics[D]. Harbin: Northeast Forestry University, 2021(in Chinese) [25] 朱承元, 张澈, 管建华. 基于改进支持向量机的空域交通态势识别方法[J]. 交通信息与安全, 2023, 41(2): 76-85. doi: 10.3963/j.jssn.1674-4861.2023.02.008ZHU C Y, ZHANG C, GUAN J H. A method for monitoring traffic state in the airspace based on an improved support vector machine[J]. Journal of Transport Information and Safety, 2023, 41(2): 76-85. (in Chinese) doi: 10.3963/j.jssn.1674-4861.2023.02.008 [26] 付旻, 王炜, 王昊, 等. 多分类支持向量机在公交换乘识别中的应用[J]. 哈尔滨工业大学学报, 2018, 50(3): 26-32.FU M, WANG W, WANG H, et al. Application of multi-class support vector machine in public transit transfer recognition[J]. Journal of Harbin Institute of Technology, 2018, 50 (3): 26-32. (in Chinese) [27] 李清毅, 周昊, 林阿平, 等. 基于网格搜索和支持向量机的灰熔点预测[J]. 浙江大学学报(工学版), 2011, 45(12): 2181-2187. doi: 10.3785/j.issn.1008-973X.2011.12.017LI Q Y, ZHOU H, LIN A P, et al. Prediction of gray melting point based on grid search and support vector machine[J]. Journal of Zhejiang University(Engineering Edition), 2011, 45(12): 2181-2187. (in Chinese) doi: 10.3785/j.issn.1008-973X.2011.12.017 [28] 叶飞, 苏恩杰, 梁晓明, 等. 公路隧道景观设计现状和思考[J]. 中国公路学报, 2022, 35(1): 23-37.YE F, SU E J, LIANG X M, et al. Review and thinking on landscape design of highway tunnel[J]. China Journal of Highway and Transport, 2022, 35(1): 23-37. (in Chinese)