快速路交通事件持续时间预测模型研究

Traffic Incident Duration Forecasting Model of Expressway

摘要:

针对城市快速路交通事件持续时间影响因素的复杂性和不确定性，结合贝叶斯网络和非参数回归方法，提出了一种新的快速路交通事件持续时间预测模型。采用上海市快速路监控中心数据，经过降噪处理，生成样本数据；在分析样本数据特征基础上，确定了贝叶斯网络的结构学习方法与参数学习方法；之后对贝叶斯网模型的结果用非参数回归算法生成持续时间预测值。最后对模型预测精度进行了验证，发现模型预测效果较好。

Abstract:

According to the complexity and uncertainty of impact fact of traffic event duration on urban expressway, a new forecasting model using Bayesian network and non-parametric regression for traffic incident duration was proposed. Sample database, provided by Shanghai Expressway Monitoring Center, was generated by noise reduction. This paper determined the algorisms of structure learning and parameter learning based on data characteristics, and got forecasting results with non-parametric regression. Finally, the forecasting model was tested with new data and the results indicate the accuracy of the model.