| Traffic congestion is a major problem to be solved in modern cities.As nodes of the road network,intersections are apt to become the inducement points of traffic congestion.The study of the traffic data at intersections has important theoretical value and practical significance.Intelligent and data-based transportation system has become one of the important means to solve the problem of traffic congestion.The way to efficiently obtain more valuable information from the massive traffic flow database has become a hot topic for scholars to discuss.Therefore,studying the traffic status parameters and congestion propagation rules based on the traffic data at intersections has guiding significance for achieving congestion warning and early interventionBased on intersection traffic flow data,the paper makes a study on the estimation of state parameters and the method for mining congestion propagation rule of intersections to analyze the characteristics of traffic flow in urban road systems.The research contents and main conclusions of this dissertation include:(1)The vehicle detector data errors are divided into two categories:data anomalies and missing data.The threshold method is used to process the abnormal data,and the missing data is repaired according to the correlation and periodicity of the time series traffic data.For non-detected intersections,the impact of the traffic flow at the entrances and exits on both sides of the roads associated with the intersections is considered and the data at the intersections with detectors is converted into the data at the intersection entrance.A framework for detecting traffic data at intersections without detectors based on an improved similarity clustering algorithm is proposed.It was found that the Pearson correlation coefficient increased by 37.29%before and after the improvement,the lowest was 10.8%,and the average improvement was 20.58%.(2)In the case of single traffic data at intersections,three common methods for calculating traffic state parameters at intersections,such as saturation,queuing length,and delay time,were effectively explored.An improved demand-capacity method is proposed to calculate the saturation of intersections.In terms of demand,the impact of inflows and outflows on road sections on the demand for intersections in the morning and evening peak periods is considered.In terms of capacity,the downstream intersection queuing and spread leading to the damage of upstream intersections is considered.In the process of estimating the queue length and delay time,two different calculation methods with saturation equal to 0.8 as the limit were determined.(3)The method of applying data mining technology to directly identify and analyzing massive data to obtain the evolution of congestion and diffusion at intersections in road networks is used.Then,the adaptive adjustment to the database is conducted according to the specific features of the traffic field.Meanwhile,the Apriori algorithm based on the topological constraints of the road network is proposed to establish the model for mining traffic congestion propagation rule at regional intersections.Finally,regional intersections in Suzhou is used as an example to analyze and verify the proposed model.It is found that congestion has similarity and recurrence from the time dimension analysis and two congestion propagation modes:backflow mode and outflow mode,are obtained from the spatial point of view,which are consistent with the reality.The results show that the model can effectively mining congestion propagation rule at regional intersections and predict traffic conditions.The time sequence of congestion at each intersection in the area based on the above-mentioned congestion propagation law is obtained.When congestion occurs at an intersection with a congestion source in a certain law,the subsequent traffic conditions of other intersections in the law can be predicted according to the trend of congestion propagation in the law of congestion propagation. |
