考虑多因素的城市道路交通拥堵指数预测研究

在分析城市道路交通拥堵指数总体变化规律的基础上,综合考虑天气、节假日、重大活动等因素对交通的影响,以未来3 h、第2天24 h每5 min的交通拥堵指数明细为预测目标函数,建立基于K近邻的城市道路交通拥堵指数预测模型,确定了模型的状态向量、距离计算方法、预测值计算方法等,并根据实际采集数据对模型各参数进行标定,实现了对广州市宏观交通拥堵指数的短期、中期预测.最后,以2016年1~2月的数据为例,对模型进行测试验证.结果表明,预测模型对于普通日、特殊日的预测效果理想,且具有较强的可操作性,基本达到工程应用效果.

Prediction of Urban Traffic Performance Index Considering Multiple Factors

Based on studying the change rules of macroscopic urban traffic, and comprehensive consideration the factors on the impact of traffic, such as: weather, holidays, major activities other factors, this paper establishes a K-Nearest Neighbor traffic performance index prediction model for the next 3 hours and 24 hours (every 5 minutes). Taking the effect of all the related factors into account, the model designs the state vectors, distance calculation and the prediction calculation method, calibrates parameters of the model with historical data, and then the traffic performance index of mid-term and short-term can be predicted. The test result with the traffic performance index of Guangzhou proves the proposed model has ideal prediction on both normal and special days with strong practicability and maneuverability