骑行过程中智能手机分心行为检测模型研究

Research on detection model of smartphone distraction behavior during riding

为了探寻电动自行车骑行过程中手机分心行为的检测方法,设计了骑行过程中手机操作测试任务,通过九轴姿态传感器、GPS获取电动自行车横纵向运行指标,并对各项指标进行差异性分析以提取特征参数。采用遗传算法优化的随机森林模型(GA-RF)构建骑行过程中智能手机分心行为检测模型,利用测试集证明模型有效性。结果表明,GA-RF能够有效地对骑行者手机分心状态进行检测,GA-RF的检测精确率为92.8%,F1值为89.8%,均优于支持向量机(SVM)结果,该检测模型可为分心预警系统提供依据。

In order to explore the detection method of mobile phone distraction behavior during riding of electric bicycle,the mobile phone operation test task during riding was designed.The horizontal and vertical operation indexes of electric bicycle were obtained by nine-axis attitude sensor and GPS,and a difference analysis of each index was carried out to extract characteristic parameters.Using GA-RF(GA-RF)to construct the distraction behavior detection model of smart phone during riding,the validity of the model is proved by the test set.The results show that GA-RF can effectively detect the distraction state of mobile phones.The accuracy of GA-RF is 92.8%,and the F1 value is 89.8%,which are superior to the results of support vector machine(SVM).This detection model can provide theoretical basis for distraction early warning system.