摘要
为了完善路侧信息系统,弥补传统环境感知技术的不足,基于深度学习的原理,将支持向量机应用于路面积水危险水位的判断。通过考虑不同道路属性组合的各类路况下路面积水量对于车辆涉水行驶安全性的影响,利用网格搜索选取最优参数组合,构建水位判断模型,模型精度达到97.66%。该模型能够较好地针对不同路况进行水位划分,将路面积水水位划分为安全水位和危险水位,有利于车辆根据此信息调整路径及车速等决策,促进智能网联交通系统的发展和完善。
In order to improve roadside information system and make up for the shortcomings of traditional environmental perception technology,the support vector machine is applied to the judgment of water level on the pavement,based on the principle of deep learning.By considering the impact of ponding quantity of the pavement on safety of vehicle wading under various road conditions with different road attribute combinations,a grid search method is used to select the optimal parameter combination,and a water level judgment model is constructed.The model accuracy reaches 97.66%.This model can better divide the water level according to different road conditions,dividing ponding level of the pavement into safe level and dangerous level,which is beneficial for vehicles to adjust the path and speed and other decisions based on this information,and contributes to the development and improvement of the intelligent and connected transportation system.
作者
王思祺
吴悠
王世妍
WANG Siqi;WU You;WANG Shiyan(School of Transportation,Jilin University,Changchun 130022;School of Civil Engineering,Fuzhou University,Fuzhou 350000)
出处
《公路交通技术》
2022年第5期151-155,共5页
Technology of Highway and Transport
关键词
水位判断
支持向量机
网格搜索法
智能网联交通
深度学习
judgment of water level
support vector machine
grid search method
intelligent and connected transportation
deep learning
