论文部分内容阅读
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。
Aiming at the accident prediction of the basic section of freeway in mountain heavy hill area, According to the flat vertical geometric line, the section elements are divided. Based on the reduction algorithm of recognizable matrix in rough set theory, the geometrical linear index variables that have prominent influence on the accident are selected. Based on the linear indicators screened, the accident prediction road segment is defined and the assignment method of forecasting unit null value is given. According to the complex non-linear relationship between accident rate, geometrical linear index and traffic volume, an Elman neural network accident prediction model is established to predict the accident of the accident prediction unit. The sensitivity analysis was carried out by using the calibrated prediction model to determine the relationship between each linear index, AADT and accident rate. By comparing with the data based on actual accident statistics, the reliability of the model in traffic safety mechanism is verified. The results of model application show that this model has more portability and versatility to expressway in mountainous hilly area.