摘要
为解决驾驶行为引起的交通冲突预测误差大的问题,采用LabVIEW构建了驾驶员驾驶行为数据集,并利用LSTM(Long Short-Term Memory)算法对驾驶行为进行风险评估,得出综合风险因数;在实车测试中,将风险因数与跟驰模型结合,建立了驾驶员碰撞风险评估模型,并验证其精度。研究结果表明:1)综合风险因数是道路风险预测的主要参数,其分值越高则事故发生率越高;2)基于LSTM的碰撞风险评估准确率为96.26%,高风险驾驶行为的识别速度可缩短到1.72 s;3)基于LSTM驾驶行为的碰撞风险评估模型精度高、评估稳定、反应迅速,可为预防交通事故、促进智能交通发展提供及时可靠的安全预警信息。
In order to solve the problem of large error in predicting traffic conflicts caused by driving behavior,LabVIEW is used to construct a dataset of driver driving behavior.The LSTM(Long Short-Term Memory)algorithm is used to assess the risk of driving behavior and obtain a comprehensive risk factor.In the actual vehicle test,the risk factor and the following model are combined to establish the driver collision risk assessment model and verify its accuracy.The results show that the comprehensive risk factor is the main parameter of road risk prediction,and the higher the score,the higher the accident rate.The accuracy of collision risk assessment based on LSTM is 96.26%,and the recognition speed of high-risk driving behavior can be shortened to 1.72 s.The collision risk assessment model based on LSTM driving behavior has high accuracy,stable evaluation and rapid response,which can provide timely and reliable safety early warning information for preventing traffic accidents and promoting the development of intelligent transportation.
作者
周士谦
高永强
齐龙
尹迁齐
ZHOU Shiqian;GAO Yongqiang;QI Long;YIN Qianqi(School of Automotive Engineering,Shandong Jiaotong University,Jinan 250357)
出处
《公路交通技术》
2024年第1期163-170,共8页
Technology of Highway and Transport
关键词
道路交通安全
驾驶行为
LSTM识别
风险评估
road traffic safety
driving behavior
LSTM identification
risk assessment