Research On Intelligent Traffic Signal Model Based On Real-time Traffic Data

Intelligent traffic signal model(ITM)is a method to proactively change the green signal ratio of traffic signals according to real-time road conditions,which aims to give full play to the traffic capacity under existing road conditions to alleviate traffic congestion.The intelligent traffic signal model is based on real-time traffic data.However,most of the real-time traffic data have problems such as poor real-time performance,small coverage,high cost and difficult acquisition.With the rapid popularization of navigation services,it is possible to build intelligent traffic signal models using real-time traffic condition data of open source navigation platform.In addition,with the rapid development of deep learning in recent years,deep learning is also used to extract the traffic flow from traffic surveillance videos.In this paper,an intelligent traffic signal model based on real-time traffic data provided by open source navigation platform is designed by using fuzzy control method.Based on open source real-time traffic data and real-time traffic flow data extracted from traffic surveillance videos,an intelligent traffic signal model with better performance based on multisource real-time traffic data is established by using the method of deep reinforcement learning.The main research contents of this paper are as follows:(1)Automatic collection of road condition data.Through the application programming interface(API)provided by the open source navigation platform,the batch collection of open source real-time traffic data of specified roads is realized.Through YOLO algorithm(You Only Look Once)and Deep SORT(Deep Simple Online and Realtime Tracking)algorithm combined with crossing detection technology to extract the traffic flow at the intersection in the traffic surveillance video.YOLO is used to detect the vehicles in the video,Deep SORT is used to track the vehicles in the video,and the line-crossing detection technology is used to count the vehicles.(2)Studied and designed an intelligent traffic signal model based on open source real-time traffic data.The congestion evaluation method based on open source realtime traffic data is studied,and the congestion evaluation models for directional sections,single intersections and adjacent intersections are designed respectively.The intelligent traffic signal model based on open source real-time road condition data is constructed,and the intelligent signal cycle model considering congestion at neighborhood intersection is constructed by fuzzy control method.The timing model is established according to the direction of road congestion.The intelligent signal cycle model intelligently changes the signal cycle time according to the congestion at the adjacent intersection,and the timing model intelligently assigns the green light time according to the congestion at the direction of the road.(3)The intelligent traffic signal model based on multi-source real-time traffic data is studied and designed.Open source real-time road condition data and traffic surveillance video data are used as data sources,Deep Q-Learning Network(DQN)algorithm is adopted,and SUMO traffic simulation software is used to build the training environment.Congestion at single intersection and traffic flow at intersection entrance and entrance based on surveillance video are taken as the state and reward function indicators of the Network model.In order to change the signal phase,an intelligent traffic signal control network model based on multi-source real-time road condition data is established to achieve more accurate traffic signal control.(4)Simulation experiments of the two models.The timing signal control scheme was set as the control group and the queue length of vehicles was taken as the evaluation index under two traffic conditions of general congestion and severe congestion.The intelligent traffic signal model based on open source real-time traffic data is simulated and verified by C# programming on VISSIM traffic simulation platform.An intelligent traffic signal model based on multi-source real-time traffic data is simulated and verified by Python programming on SUMO traffic simulation platform.Simulation results show that both models can effectively alleviate urban traffic congestion,and the effect is better in severe congestion conditions.The intelligent traffic signal model based on open source real-time traffic data can reduce the queue length of vehicles by 29.99%.The intelligent traffic signal model based on multisource real-time traffic data can reduce the queue length by 32.05%.The results show that the existing real-time multi-source road data can be used to design a low-cost intelligent traffic signal model,which can significantly improve the capacity of the existing road network,and is expected to be applied and popularized through further testing.