摘要
以驾驶模拟器作为数据采集平台,采集雾天高速公路不同能见度车辆的车头时距数据,对车头时距分区间统计,不做任何假设,克服了仅以天气等级或是车速和能见度作为风险评估的指标、行车安全等级以仿真得到的曲线图来定性判定的缺点。选取能见度和车头时距作为指标,对雾天高速公路的交通安全进行了风险评估。运用Fisher最优分割编写Matlab程序。将风险分为4级,以车头时距出现的概率结合损失量的大小确定风险分级标准。采用K近邻非参数对风险进行预测,训练集及验证集的分类误差均为0.7%,验证该模型具有有效性。
Considering driving simulator as data collection platform,the data of vehicle headway with different visibility in foggy weather conditions are collected,headway data with partition are counted with no assumptions,and only the rank,or the speed and visibility are considered as an indicator of the risk assessments.Traffic safety level is obtained by a simulation curve with qualitative judgement.The visibility and headway is selected as the indicators to make highway traffic safety assessment carried out in foggy weather conditions,and the idea of optimal partition of Fisher is applied and write Matlab program is writen to determine the risk which can be divided into 4levels,the probability of headway and loss of size are combined to make sure the risk classification standards,then Knearest neighbor nonparametric is used to forecast the risk,the error on the training set and validation set were 0.7%,it indicates that the model is effective.
出处
《交通科学与工程》
2015年第1期79-84,共6页
Journal of Transport Science and Engineering
基金
江西省交通运输厅科技计划项目(2013C0008)
长沙理工大学研究生科研创新项目
