欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
针对山岭重丘区高速公路基本路段的事故预测问题开展研究。依据平纵几何线形对路段单元进行划分,基于粗糙集理论中可辨识矩阵的约简算法选择出了对事故发生有突出影响的几何线形指标变量。依据筛选出的线形指标,界定了事故预测路段单元并给出了预测单元空值项的赋值方法。针对事故率与几何线形指标、交通量之间复杂的非线性关系,建立了Elman神经网络事故预测模型,可对事故预测单元进行事故预测。应用标定出的预测模型进行敏感性分析,确定出了各线形指标、AADT等与事故率的关系。通过与基于实际事故数据统计得到的关系进行对比,验证了该模型在交通安全机理上的可靠性。模型应用结果表明：该模型具有较大的可移植性和对山岭重丘区高速公路的通用性。
英文摘要：
The research of accident prediction issue is carried out for basic sections of expressways in mountainous and rolling areas. A series of subsections are obtained by section division based on horizontal and vertical geometric alignments, and the geometric alignment indicator variables which have prominent effect on accident happening are selected by reduction algorithm of discernible matrix according to rough sets theory. Based on the selected alignment indicators, the accident prediction section is determined, and a method of assigning values to null value items is provided. In view of the complex nonlinear relationship among accident rate, geometric alignment and traffic volume, Elman neural network accident prediction model is set up, it can be used to predict accident in the prediction section. The relationship among accident rate, aligmnent indicators and AADT are determined by sensitivity analysis based on the calibrated accident prediction model. Compared with the relationship obtained by statistics of real accident data, the reliability of the model on the traffic safety mechanism is proved. The application of the model shows that the model is transferable and could be applicable to the expressways in mountainous and rolling areas.
同期刊论文项目
同项目期刊论文
期刊信息
位置：