Estimation Of Maximum Queuing Length At Signalized Intersections Based On License Plate Data

The maximum queue length of vehicles at urban intersections is one of the important indicators to measure the traffic situation.The estimation of the maximum queue length can provide evidence for urban management,signal control,route guidance,etc.,so as to ease the traffic jams in the city.In this paper,the maximum queue length at a signalized intersection is defined as the number of vehicles in queue when the rear end of the queue is farthest from the intersection in a certain lane in a cycle.In this paper,the maximum queue length of intersection is estimated based on license plates.The main research contents are as follows:Firstly,this paper analyzes and cleans the data.This paper analyzes the reasons for the data redundancy,abnormal identification of vehicle plate,incorrect travel time recording and other problems in the license plate data obtained by the electronic police,and cleans the original license plate data combined the signal timing information.On this basis,the vehicle travel time was obtained by matching the license plate number,then cleaned the noise data in the travel time referenced the latest algorithm.According to the law of travel time,complete the travel time after cleaning.Then,according to the average vehicle delay and signal timing data,the day is divided into multiple time intervals.Taking a time interval as the basic unit,the vehicle travel time distribution in the interval was fitted by Gaussian mixture model,and the model was solved by EM algorithm.For the disadvantage that EM algorithm cannot optimize the distribution number of models,this paper sets the distribution number of models by analyzing the features of travel time distribution.For the disadvantage that EM algorithm cannot optimal the global optimal solution,the distribution obtained by the model is used to judge the number of queuing vehicles,and the number of queuing vehicles is transformed into the maximum queuing length of each cycle though statistics methods.Finally,in order to test the application effect of the model and algorithm used in this paper.This paper carried out VISSIM simulation analysis and field data analysis for the case road section.In the VISSM simulation analysis,the vehicle travel time obtained by the travel time detector.In the real scene,the license plate data obtained by the electronic police and the signal timing data uploaded by the signal control system are used as the data sources.Finally,use MSE,MAE and other indicators to evaluate the maximum queue length estimation results of lane groups and single lane under different traffic conditions.Comparing with other algorithms,The maximum queue length estimation method in this paper is more effective and practical.