考虑不规则驾驶行为的城市隧道事故率预测

准确预测交通事故风险概率是实现隧道内交通安全主动管理的先决条件。本文从交通状态和驾驶员行为2个方面分析影响事故发生的主要因素,提出“不规则行为率”作为特征变量。对特征变量显著性检验,选用Logistic概率模型构造城市隧道事故率预测模型,并提出利用ROC曲线寻找阈值的方法。最后基于实测的青岛胶州湾隧道数据进行实例验证,研究结果表明,阈值确定为0.36较为合适,此时基于不规则驾驶行为的事故率预测模型准确度为78.4%,能够较为准确预测隧道内事故发生的可能性,研究成果可以为智能交通安全发展奠定理论基础。

Urban Tunnel Accident Rate Prediction Considering Irregular Driving Behavior

Accurately predicting the probability of traffic accident risk is a prerequisite for the active management of traffic safety within the tunnel. Firstly, the main factors affecting the accident are analyzed in terms of traffic status and driver behavior, and the "irregular behavior rate" is proposed as the characteristic variable. After the saliency test of the characteristic variables, the Logistic probability model is used to construct the urban tunnel accident rate prediction model, and the method of using the ROC curve to find the threshold is proposed. Finally, the Qingdao Jiaozhou Bay Tunnel data is taken as an example to verify the model. The results show that the threshold is determined to be 0.36. The accuracy of the accident rate prediction model based on irregular driving behavior is 78.4%. It can accurately predict the possibility of accidents in the tunnel, and the research results can lay a theoretical foundation for the development of intelligent traffic safety.