基于交通安全数字画像的国省干道预警系统研究与实现  

Research and Implementation of Early Warning System for National and Provincial Arterial Roads Based on Digital Portrait of Traffic Safety

作　　者：王淼 楼欧阳 杨立成 金家宏 WANG Miao;LOUOu-yang;YANG Li-cheng;JIN Jia-hong(CCCC Highway Consultants Co.Ltd.,Beijing 100010,China;Jinhua Jiaotou Transportation Engineering Construction Investment Management Co.Ltd.,Jinhua 321015,China;Suzhou Hongmu Information Technology Co.Ltd.,Suzhou 215000,China)

机构地区：[1]中交公路规划设计院有限公司,北京市100010 [2]金华交投交通工程建设投资管理有限公司,金华市321015 [3]苏州弘目信息技术有限公司,苏州市215000

出　　处：《公路》2025年第2期289-295,共7页Highway

摘　　要：为进一步提升国省干道交通安全水平,针对穿村、镇区域的国省道干线公路,提出了基于交通安全数字画像的预警系统。针对人、车、路、环境等4个方面安全隐患因素采用基于雷视融合的交通安全专精感知算法对检测目标进行标定,并利用yolov5深度学习算法进行目标分类和特征提取,提出并实现基于单目低位监控的3D运动参数提取和交通冲突精确计算。提出了交通安全预警与分析闭环流程,并结合交通安全数字画像、车辆轨迹等信息形成了交通安全分析与反馈机制。系统在235国道金东安里至楼店段进行了有效应用,大幅降低了互通区及辅道平面交叉口严重的交通冲突率及交通事故率。To improve traffic safety on national and provincial trunk roads that pass through village and town areas,an early warning system is proposed in the paper based on digital imaging of traffic safety.According to the hidden danger factors of people,vehicles,roads and environment,a specialized traffic safety perception algorithm based on the fusion of thunder and vision is used to calibrate the detection target,and YOLOv5 deep Learning Algorithm is used for target classification and feature extraction.The 3D motion parameter extraction and accurate calculation of traffic conflicts based on monocular low-level monitoring is proposed and implemented.A closed-loop process for traffic safety warning and analysis is used for target classification and feature extraction.This process forms a traffic safety analysis and feedback mechanism by combining traffic safety digital portraits,vehicle trajectories,and other relevant information.The system has been effectively applied in the section from Jindong Anli to Loudian of National Highway No.235.As a result,the rate of serious traffic conflicts and accidents in the interchange area and the intersection of auxiliary roads has been significantly decreased.

关 键 词：交通冲突 国省干道 隐患识别 数字画像 安全预警 

分 类 号：U491.4[交通运输工程—交通运输规划与管理]