摘要
文中以辨识船员分心驾驶关键因素为目标,建立了船员个体特征、生理特征、心理特征为主体的分心驾驶行为细粒度指标体系,将决策树(DT)作为AdaBoost算法的基分类器构建了AdaBoost-DT模型,通过实船实验采集的225组船员身心及分心驾驶行为数据验证了AdaBoost-DT模型的高效性.结果表明:选取以决策树作为基分类器的AdaBoost-DT模型与AdaBoost-SVM模型相比辨识准确率更高,达到91.3%,且AUC值为0.9559;年龄、感知压力、疲劳程度、工作态度及驾龄是影响被试船员群体发生分心驾驶行为的关键因素.
In order to identify the key factors of distracted driving,a fine-grained index system of distracted driving behavior with individual characteristics,physiological characteristics and psychological characteristics of crew as the main body was established.Decision tree(DT)was used as the base classifier of AdaBoost algorithm to construct the AdaBoost-DT model.The efficiency of AdaBoost-DT model was verified by the data of 225 groups of crew members’mind,body and distracted driving behavior collected from real ship experiments.The results show that the AdaBoost-DT model based on decision tree is more accurate than the AdaBoost-SVM model,reaching 91.3%,and the AUC value is 0.9559.Age,perceived stress,fatigue,work attitude and driving experience are the key factors that affect the distracted driving behavior of the crew.
作者
刘清
王馨玥
王磊
吴宇航
LIU Qing;WANG Xinyue;WANG Lei;WU Yuhang(Wuhan University of Technology School of Transportation and Logistics Engineering,Wuhan 430063,China;National Key Laboratory of Waterway Traffic Control at Wuhan University of Technology,Wuhan 430063,China)
出处
《武汉理工大学学报(交通科学与工程版)》
2025年第4期699-705,共7页
Journal of Wuhan University of Technology(Transportation Science & Engineering)
基金
科技部十四五国家重点研发计划(2021YFC3001500)
国家自然科学基金面上项目(51979214)。
关键词
船员
分心驾驶
身心叠加
行为辨识
ADABOOST
crew
distracted driving
mind-body superposition
behavior identification
AdaBoost
