| Prediction of queue length of vehicles caused by abnormal events of expressway is an important basis for traffic managers to implement effective control of expressway sections.Most previous related researches are suitable for ideal road environment with collecting data conveniently.Due to the randomness of location of abnormal events of expressway and the limitation of data acquisition methods of expressway,it is difficult to meet the conditions of the existing queue length prediction model in the actual road environment.Therefore,it is very important for improving the operating efficiency of expressway by making full use of existing data sources to establish an effective queue length prediction model.In this thesis,the queue situation of the vehicles caused by the abnormal events happened on the road downstream intersection of the entrance ramp and the main road is studied.A flow estimation method for the target cross section and a speed estimation method for the target area by analyzing the impact of ramp flow on the main road flow is proposed.By analyzing the impact of the tunnel on the main road on the actual capacity of the road,a method for estimating the actual capacity of the road considering the influence of the tunnel is proposed.By comparing the applicability of the queuing length prediction model based on the classic theory,combining the characteristics of the target scene and the actual data conditions,a suitable model is selected as the basic model and improved,so as to realize the effective prediction of the change in the queue length of the target scene.The work of this thesis includes:(1)Estimation of cross-section traffic parameters after the intersection of ramp entrance and main line.To estimate the traffic flow in the target area,we first need to analyze the impact of ramp traffic on the main road traffic.On this basis,a traffic flow estimation method for the target cross section based on the OD matrix and exponential smoothing model is proposed.By considering the specific application of data conditions and speed parameters in the target area in queuing length prediction,an average speed estimation method based on toll data and homogeneous traffic flow theory is proposed.The estimation of density parameters is mainly based on the macro traffic flow model.Firstly,compare and analyze the characteristics of several commonly used macro traffic flow models.Then,the Van Aerde model as the estimate model of the target road vehicle density is selected by combining the characteristics of the demonstration road.The results show that the estimation methods in this thesis is applicable and effective.In the meantime,the models can lay part of the research foundation for the subsequent queuing length prediction.(2)Estimation of the capacity of the main highway tunnel section of the expressway.This thesis aims at the actual capacity of the road section with a certain length of tunnel in the road.By considering the differences in the road characteristics between the tunnel that is easily overlooked at the macro level and the road in the field and analyzing the influence of tunnel on the actual capacity of the road combing with the traditional actual capacity estimation method,a model based on the time-space consumption method for estimating the actual traffic capacity of road section which contains different levels is proposed.Trans Modeler simulation platform is used for simulation experiments.The results show that the model is effective.In the meantime,the model can lay part of the research foundation for the subsequent queuing length prediction.(3)Queue length prediction under the condition of sparse highway detection equipment.Aiming at the queuing situation of vehicles caused by the incident that occurred downstream of the intersection of the ramp entrance and the main line.Under the condition of sparse detection equipment and based on relative position of the car detector and ramp toll station of the demonstration road section,a method that can divide the road into different modalities in order to solve the problem that is hard to obtain the traffic parameters of the incident section caused by the randomness of the location of the abnormal event to a certain extent is proposed.By comparing the applicability of the two prediction models based on classic theory,combining with the limited data conditions of the target scene and the relatively complex actual road environment,the MAEQL model is selected as the basic model for improvement.The validity and stability of the model is verified by the actual data in the target road section. |
