A Method for Identifying Temperament Propensity of Drivers Based on AutoNavi Navigation Data and a FOA-GRNN Model
-
摘要: 为提升汽车主动安全功能，研究了1种基于高德导航数据的低成本、高精度驾驶倾向性辨识方法。基于高德软件开发工具构建动态驾驶数据采集应用程序，并融入个人智能终端以实现对行车数据的实时采集、处理与网络化存储。通过驾驶员生理、心理测试和实车实验获取不同驾驶倾向性驾驶员在导航行驶过程中由时间、速度和加速度推演的驾驶行为信息，采用主成分分析法（PCA）提取驾驶倾向性主要因子，并将驾驶倾向分为激进型、普通型和保守型这3类。构建基于果蝇优化算法（FOA）和广义回归神经网络（GRNN）的高精度驾驶倾向性辨识模型，利用特征变量集对模型进行训练和验证。验证结果表明：该模型总体准确率可达94.17%，对激进型、普通型和保守型的驾驶倾向性的辨识精确度分别为95.06%，92.5%，94.93%；进一步对比发现，该模型比单一的GRNN模型总体准确率提高5%~10%，与现有基于惯性传感器数据和离散小波变换结合自适应神经模糊推理系统的方法相比，该方法更具实用性且模型总体辨识准确率提升了2.17%。Abstract: In order to improve the capacity of automobiles in active safety, a method for identifying driving propensity with a low-cost and high accuracy based on AutoNavi navigation data is proposed. An application to collectdriving data is developed based on Amap software development tool, which is further integrated into an intelligentterminal for data collection, procession, and storage in real time. Driver behavior data inferred from the time, speed, and acceleration of vehicles controlled by drivers with different temperament propensity are obtained through physiological, psychological and driving experiments. The principal component analysis(PCA)technique is used to extract the important factors for studying the temperament propensity of drivers, and the drivers are grouped into threedriving propensities: radical, common and the conservative. A Fruit-fly optimization algorithm(FOA)and a generalized regression neural network(GRNN)are integrated to establish a high-precision model for driving propensity identification, which is further trained and verified using collected data. The verification results show that: the overall accuracy of the identification model is 94.17%, and the identification precision of the radical, common and theconservative types are 95.06%, 92.5% and 94.93%, respectively; compared to the simple GRNN model, the overallprecision of the proposed model is improved by 5%~10%; and compared to the previous method based on inertialsensor data and the integrated method of discrete wavelet transformation and adaptive neuro fuzzy inference system, the FOA-GRNN model is more practical, and its overall precision is improved by 2.17%.
-
表 1 驾驶特征变量和表示符号
Table 1. Driving characteristic variables and representation symbols
变量 符号 变量 符号 年龄/岁 a 加速次数 Nacc 驾龄/年 DA 减速次数 Ndec 性别 G 加速时间/s Tacc 行程时间/s T 减速时间/s Tdec 平均速度（/m/s） Vave 平均加速度（/m/s2） Aave 最大速度（/m/s） Vmax 最大加速度（/m/s2） Amax 
下载: 导出CSV
表 2 驾驶倾向性预判结果
Table 2. Preliminary judgment result of driving propensity
驾驶倾向性类型 驾驶员编号 激进型 03，08，12，13，20，23，27，30，33，35 41，44，46，47，50 普通型 02，06，07，09，11，15，16，17，22，25 28，29，34，37，38，40，43，48 保守型 01，04，05，10，14，18，19，21，24，26 31，32，36，39，42，45，49
下载: 导出CSV
表 3 各主成分总方差解释
Table 3. Interpretation of total variance of each principal component
成分 初始特征值 提取载荷平方和 总计 方差百分比 累计/% 总计 方差百分比 累计/% 1 6.640 55.333 55.333 6.640 55.333 55.333 2 1.964 16.367 71.700 1.964 16.367 71.700 3 0.922 7.683 79.382 0.922 7.683 79.382 4 0.594 4.947 84.329 0.594 4.947 84.329 5 0.478 3.983 88.311 0.478 3.983 88.311 6 0.439 3.656 91.967 7 0.309 2.574 94.541 8 0.267 2.225 96.766 9 0.181 1.506 98.272 10 0.130 1.080 99.352 11 0.073 0.610 99.962 12 0.005 0.038 100.000
下载: 导出CSV
表 4 各主成分的得分
Table 4. Score of each principal component
测试样本 第1主成分 第2主成分 第3主成分 第4主成分 样本1 1.510 3 -1.338 5 0.964 0 -3.341 8 样本2 1.136 2 -1.368 8 0.826 3 1.620 9 样本3 1.631 8 -1.339 8 1.040 7 0.120 6 ⋮ ⋮ ⋮ ⋮ ⋮ 样本1999 -1.343 9 -0.184 9 0.860 5 -0.505 4 样本2000 1.579 6 0.225 4 0.834 0 0.913 3
下载: 导出CSV
表 5 驾驶倾向性辨识模型各项评价指标
Table 5. Evaluation indexes of driving propensity identification model
单位: % 驾驶倾向性类型 准确率 精确度 召回率 F1分数 1（激进型） 94.17 95.06 96.25 95.65 2（普通型） 94.17 92.5 92.5 92.5 3（保守型） 94.17 94.93 93.75 94.34
下载: 导出CSV
表 6 实车实验20名驾驶员最终验证结果
Table 6. The final verification results of 20 drivers in the real vehicle experiment
驾驶员编号 模型验证结果（准确率/%） 驾驶倾向性预判 12 激进型（95.1） 激进型 34 普通型（92.9） 普通型 09 普通型（93.3） 普通型 42 保守型（94.5） 保守型 17 普通型（92.2） 普通型 03 激进型（96.2） 激进型 21 保守型（95.1） 保守型 28 普通型（92.9） 普通型 36 保守型（94.5） 保守型 19 保守型（95.1） 保守型 25 普通型（93.3） 普通型 44 激进型（95.1） 激进型 32 保守型（96.2） 保守型 26 保守型（92.9） 保守型 27 激进型（94.5） 激进型 40 普通型（92.9） 普通型 38 普通型（92.2） 普通型 47 激进型（96.2） 激进型 04 保守型（95.1） 保守型 15 普通型（92.9） 普通型
下载: 导出CSV
表 7 GRNN驾驶倾向性辨识结果
Table 7. Identification results of GRNN driving propensity identification model
序号 σ 准确率/% 序号 σ 准确率/% 1 50 83.3 6 0.8 87.1 2 15 84.6 7 0.5 86.3 3 10 86.7 8 0.1 89.2 4 5 85.4 9 0.05 88.7 5 1 87.5 10 0.01 87.9
下载: 导出CSV
-
[1] WANG X, LIU Y, WANG F, et al. Feature extraction and dynamic identification of drivers' emotions[J]. Transportation Research Part F: Traffic Psychology and Behaviour, 2019 (62): 175-191. [2] 王晓原, 刘亚奇, 张敬磊. 基于行程时间的人车特征动态辨识方法[J]. 汽车安全与节能学报, 2017, 8 (1): 38-45. https://www.cnki.com.cn/Article/CJFDTOTAL-QCAN201701004.htmWANG X Y, LIU Y Q, ZHANG J L. Dynamic identification method of driver and vehicle features based on travel time[J]. Journal of Automotive Safety and Energy, 2017, 8(1): 38-45. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-QCAN201701004.htm [3] JOHNSON D A, TRIVEDI M M. Driving style recognition using a smartphone as a sensor platform[C]. 14th International IEEE Conference on Intelligent Transportation Systems (ITSC). Washington, D. C., USA: IEEE, 2011. [4] FENG Y, PICKERING S, CHAPPELL E, et al. A support vector clustering based approach for driving style classification[J]. International Journal of Machine Learning and Computing, 2019, 9 (3): 344-350. doi: 10.18178/ijmlc.2019.9.3.808 [5] YAN F, LIU M, DING C, et al. Driving style recognition based on electroencephalography data from a simulated driving experiment[J]. Frontiers in Psychology, 2019 (10): 1254. https://pubmed.ncbi.nlm.nih.gov/31191419/ [6] WANG W, XI J, ZHAO D. Driving style analysis using primitive driving patterns with Bayesian nonparametric approaches[J]. IEEE Transactions on Intelligent Transportation Systems, 2018, 20 (8): 2986-2998. https://ieeexplore.ieee.org/abstract/document/8506402/ [7] WANG X Y, LIU Y Q, LIU L P, et al. Experimental research on car movement characteristics under the condition of different driving emotions[J]. Advances in Mechanical Engineering, 2018, 10 (12): 1-13. doi: 10.1177/1687814018815369 [8] 王晓原, 张敬磊, Xuegang (Jeff)Ban. 基于动态人车环境协同推演的汽车驾驶倾向性辨识[M]. 北京: 科学出版社, 2013.WANG X Y, ZHANG J L, BAN X. Identification of vehicle driving tendency based on dynamic driver-vehicle-environment collaborative deduction[M]. Beijing: Science Press, 2013. (in Chinese) [9] HOU Y, EDARA P, SUN C. Modeling mandatory lane changing using Bayes classifier and decision trees[J]. IEEE Transactions on Intelligent Transportation Systems, 2013, 15(2): 647-655. [10] TRAN D, DO H M, SHENG W, et al. Real-time detection of distracted driving based on deep learning[J]. IET Intelligent Transport Systems, 2018, 12 (10): 1210-1219. doi: 10.1049/iet-its.2018.5172 [11] 王畅, 付锐, 彭金栓, 等. 应用于换道预警的驾驶风格分类方法[J]. 交通运输系统工程与信息, 2014, 14(3): 187-193+200. https://www.cnki.com.cn/Article/CJFDTOTAL-YSXT201403029.htmWANG C, FU R, PENG J S, et al. Driving style classification method for lane change warning system[J]. Journal of Transportation Systems Engineering and Information Technology. 2014, 14 (3): 187-193+200. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-YSXT201403029.htm [12] 严利鑫, 贺宜, 糜子越, 等. 考虑生理特性的驾驶行为险态辨识研究[J]. 交通信息与安全, 2019, 37 (3): 12-19+27. https://www.cnki.com.cn/Article/CJFDTOTAL-JTJS201903002.htmYAN L X, HE Y, MI Z Y, et al. A physiological signal based method for identifying risk status of driving behaviors[J]. Journal of Transport Information and Safety, 2019, 37(3): 12-19+27. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-JTJS201903002.htm [13] 朱冰, 李伟男, 汪震, 等. 基于随机森林的驾驶人驾驶习性辨识策略[J]. 汽车工程, 2019, 41 (2): 213-218+224. https://www.cnki.com.cn/Article/CJFDTOTAL-QCGC201902014.htmZHU B, LI W N, WANG Z, et al. Identification strategy of driving style based on random forest[J]. Automotive Engineering, 2019, 41 (2): 213-218+224. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-QCGC201902014.htm [14] SUBASI A, GURSOY M I. EEG signal classification using PCA, ICA, LDA and support vector machines[J]. Expert Systems with Applications, 2010, 37 (12): 8659-8666. doi: 10.1016/j.eswa.2010.06.065 [15] SPECHT D F. A general regression neural network[J]. IEEE Transactions on Neural Networks, 1991, 2 (6): 568-576. doi: 10.1109/72.97934 [16] ROOKI R. Application of general regression neural network (GRNN) for indirect measuring pressure loss of Herschel -Bulkley drilling fluids in oil drilling[J]. Measurement, 2016, 85: 184-191. [17] PAN W T. A new fruit fly optimization algorithm: taking the financial distress model as an example[J]. Knowledge-Based Systems, 2012 (26): 69-74. [18] 潘文超. 果蝇最佳化演算法: 最新演化式计算技术[M]. 台北: 沧海书局. 2011.PAN W C. Fruit fly optimization algorithm: A new evolutionary computing technique[M]. Taipei: Canghai Bookstore. 2011. (in Chinese) [19] EFTEKHARI H R, GHATEE M. Hybrid of discrete wavelet transform and adaptive neuro fuzzy inference system for overall driving behavior recognition[J]. Transportation Research Part F: Traffic Psychology and Behaviour, 2018(58): 782-796. [20] WANG W, XI J, CHONG A, et al. Driving style classification using a semisupervised support vector machine[J]. IEEE Transactions on Human-Machine Systems, 2017, 47(5): 650-660. -
点击查看大图
图(11) / 表(8)
计量
- 文章访问数: 1167
- HTML全文浏览量: 656
- PDF下载量: 40
- 被引次数: 0