Research On Overflow Control Methods Of Urban Intersection Based On Deep Learning

At present,urban traffic congestion is one of the main bottlenecks of urban development.Saturated traffic phenomena in megacities such as Beijing and Shanghai are becoming more and more common.Counter-overflow control is the entry point to alleviate the impact of saturated and super-saturated traffic.It has important theoretical significance and application value for the improvement of urban road traffic capacity.The paper mainly aims at the identification and control of overflow traffic at urban intersections,aiming at eliminating urban road traffic overflow,and for intelligent detection of queue length under different traffic conditions,signal control of overflow signal intersections and coordinated control of overflow signal intersections.Conduct research on major issues mention above..The main work and innovations of the thesis are:(1)In-depth study of intersection overflow characteristics and intersection overflow judgment algorithm based on YOLO V3 algorithm,establish overflow traffic detection method,complete intersection-based overflow detection based on deep learning.Based on the characteristics of overflow traffic,the queuing length estimation and overflow phase judgment model of overflow traffic are established.(2)A single intersection anti-overflow control model based on fuzzy control is established.The fuzzy controller is designed with the goal of traffic flow dissipation rate and signalized intersection delay in the overflow direction of the intersection.Further,the overflow phase detection result is used as an input of the fuzzy control,thereby obtaining a green light extension time of the overflow phase.The simulation results show that the method can improve the overflow rate of overflow traffic and reduce the delay time of vehicles at signalized intersections.(3)The coordinated control model of the trunk line in associated area of the overflow intersection is established.The maximum control capacity and the minimum delay are taken as the control targets to determine the coordinated control method with the non-overflow and overflow areas.Use trunk coordination to optimize trunk capacity in associated areas.The simulation results show that compared with the traditional optimization model,the model proposed can realize multi-path coordination optimization control of the entire road network.The research results in this paper can provide theoretical support and application methods for the urban saturation control of saturated,supersaturated signalized intersections and associated regions.