驾驶人分心状态判别支持向量机模型优化算法  

Identification of Driver Distraction States with Optimized Support Vector Machine Method

作　　者：张辉[1] 钱大琳[1] 邵春福[1] 钱振伟[1] 菅美英[2] ZHANG Hui;QIAN Da-lin;SHAO Chun-fu;QIAN Zhen-wei;JIAN Mei-ying(1. School of Traffic and Transportation, Beijing Jiaotong University, Beijing 100044, China; 2. Department of Civil Engineering, Tsinghua University, Beijing 100084, Chin)

机构地区：[1]北京交通大学交通运输学院,北京100044 [2]清华大学土木系交通研究所,北京100084

出　　处：《交通运输系统工程与信息》2018年第1期127-132,共6页Journal of Transportation Systems Engineering and Information Technology

基　　金：国家重点研发计划资助(2017YFC0804800);国家自然科学基金(51678044;91746201);中央高校基本科研业务费(2017JBM307)~~

摘　　要：驾驶人分心状态判别是分心驾驶预警系统的重要基础.建立以径向基为核函数的驾驶人分心状态判别SVM模型,采用遗传算法(GA)优化SVM模型惩罚参数C和核函数参数g,并利用模拟驾驶器实验平台采集的驾驶绩效数据对模型进行验证.结果表明,采用GASVM模型能够准确识别自由流和拥挤流场景下驾驶人分心状态,判别精度分别为94.5%和96.3%.与决策树C4.5和交叉验证(CV)-SVM对比表明,GA-SVM在准确率、精准率、召回率和F1值等模型性能方面均优于其他2种方法.本文建立的模型能够有效地判别驾驶人分心状态,可为驾驶人分心预警系统和分心控制策略提供依据.The identification of driver distraction states is the important component of driver distraction warning system. This paper establishes Support Vector Machine（SVM） model based on Radial Basis Function to identify driver distraction states. The genetic algorithm（GA） is used to optimize the parameters of SVM model. The optimized SVM model is used in the identification of driver distraction states, and the effectiveness of the model is validated by experimental data that used the driving performance data. The results show that the recognition accuracies of driver distraction states in free flow condition and crowed flow condition are 94.5% and 96.3%,respectively. Compared to the C4.5 and Cross Validation（CV）-SVM, the performances of GA-SVM are superior to the other two methods. This model can effectively determine the driver distraction state, which can provide data support for driver distraction warning system and control strategy.

关 键 词：交通工程 分心状态判别 支持向量机 遗传算法 驾驶绩效 参数优化 

分 类 号：U491[交通运输工程—交通运输规划与管理]