基于时空关联规则挖掘的城市交通拥堵传导预测

Forecasting urban traffic congestion conduction based on spatiotemporal association rule mining

对城市交通拥堵进行准确预测是智能交通领域的重要研究问题.为更准确地预测道路拥堵状态和挖掘拥堵传导规则,本文提出了一个新的基于时空关联规则的交通拥堵传导预测模型.该模型使用基于遗传网络规划(GNP)的时空关联规则挖掘算法识别交通拥堵在不同时间、不同地点的共现规则,揭示了交通拥堵的时空传导模式.最后,本文基于北京市交通状态实测数据的实证结果验证了该模型具有较高的预测性能.该模型突破已有研究“先流量预测,再状态分析”的技术路径,将交通拥堵状态作为直接研究对象,揭示了交通拥堵的时空动态传导规律,从而可支持城市交通主管部门提前采取更加系统化的应对措施,提高交通拥堵的前瞻性和动态性处置能力.

Accurate prediction of urban traffic congestion is a significant research in the field of intelligent transportation.In order to more accurately predict the road congestion and mine congestion transmission rules,this paper proposes an urban traffic congestion conduction forecasting model based on spatiotemporal association rules.The model first constructs a new spatiotemporal association rule mining algorithm based on genetic network programming(GNP) to identify traffic congestion rules and forms a traffic congestion rule database.On this basis,the spatiotemporal conduction prediction of traffic congestion is generated.Finally,for the validation purpose,the empirical study based on the data of Beijing ’s traffic is carried out,and the results show high performance of the proposed model.This model breaks through the technical path of "prediction of traffic first,then analysis of state",and uses traffic congestion status as the direct research object,reveals the temporal and spatial dynamic transmission rules of traffic congestion.Thus,it can support urban transportation authorities to take more systematic response measures in advance to improve the forward-looking and dynamic handling capabilities of traffic congestion.