基于运行数据的电动客车误踩加速踏板状态识别研究  

Step faultily on the accelerator state detection based on vehicle running data

作　　者：赵登峰[1] 高岩 刘朝辉[1] 秦朝峰 高立杰 ZHAO Dengfeng;GAO Yan;LIU Zhaohui;QIN Chaofeng;GAO Lijie(Zhengzhou Yutong Bus Company Ltd.,Zhengzhou 450016,China;Key Laboratory of Ministry of Public Security for Road Traffic Safety,Wuxi 214000,China)

机构地区：[1]郑州宇通客车股份有限公司,中国郑州450016 [2]道路交通安全公安部重点实验室,中国无锡214000

出　　处：《汽车与安全》2021年第3期80-85,共6页Auto & Safety

摘　　要：驾驶人误踩加速踏板导致的电动客车交通事故时有发生,影响因素复杂。为准确识别误踩加速踏板行为,提醒电动客车驾驶人安全驾驶,基于电动客车车载终端采集的车辆运行状态数据,提取9项与误踩加速踏板行为相关的驾驶特征,采用随机森林算法模型对电动客车误踩加速踏板状态进行识别,结果表明整体的识别准确率为99%,召回率为0.99,表明99%的误踩加速踏板行为可被识别出来。实验表明采用随机森林算法进行误踩加速踏板状态识别具有较好的效果,且与逻辑回归、Adaboost等识别算法相比,具有准确率、召回率和精确率高等特点。该方法利用电动客车的运行数据进行分析,简单高效成本低,具有一定的可行性和实用性。Fault stepping the Accelerator is one of the main causes of traffic accidents.It is of great importance to detect fault stepping precisely and remind drivers of electrical bus to concentrate on driving safely.Based on the vehicle running data collected by electrical bus data acquisition system,this paper extracts 9 features relevant to driving behaviors and uses random forest algorithm to identify fault stepping.The results show that the overall recognition accuracy is 99%,and the recall rate is 0.99 which means 99%of fault stepping conditions can be successfully identified.Experiments show that fault stepping detection based on vehicle running data with Random forests is the most effective.Compared with other fault stepping detection methods,the proposed method is simple and convenient,the cost is relatively low.

关 键 词：驾驶行为 交通事故 误踩加速踏板识别 随机森林 电动客车 

分 类 号：U471.3[机械工程—车辆工程]