论文部分内容阅读
依据我国山岭重丘区高速公路几何线形和交通事故数据,建立了基于交通流量和几何线形指标的高速公路基本路段事故预测模型。首先,基于几何线形条件对基本路段进行了划分,确定了路段单元。其次,分析并确定了理想线形条件的范围,建立了理想线形条件下的基本事故率预测模型。再次,应用BP神经网络与敏感性分析相结合的方法,确定出了对事故发生有突出影响的道路纵坡、平曲线半径和直线段长度3个线形指标,并确定了上述线形指标的事故率修正系数。依据基本事故率预测模型及事故率修正系数即可进行事故预测。模型验证结果表明:该模型能够对路段单元进行事故预测,事故总体预测值与实际值的相对误差在-5.85%~-7.87%之间。
According to the geometric shape and traffic accident data of the expressway in our country’s mountainous hilly region, a traffic accident and geometric linear indicator model is established to predict the accident of the basic section of expressway. First of all, the basic road sections are divided based on geometric linear conditions, and the road section units are determined. Secondly, the range of the ideal linear condition is analyzed and determined, and the basic accident rate prediction model under the ideal linear condition is established. Thirdly, by combining BP neural network and sensitivity analysis, three linear indicators of road longitudinal gradient, flat curve radius and linear section length that have a prominent impact on the accident are determined, and the accident rates of the above linear indicators are determined Correction factor. According to the basic accident rate prediction model and the accident rate correction coefficient can be accident prediction. The results of model verification show that the model can predict the accident of road section units, and the relative error between the predicted value of the accident and the actual value is between -5.85% -7.77%.