Cooperation Of Signal Control And Route Guidance Of Urban Traffic Congestion Based On Dynamic Traffic Flow Information

As a negative externality product of urbanization spillover,the response and governance of traffic congestion on a global scale has always been the focus area of the society.In the process of continuous in-depth research,intelligent traffic system(ITS)has gradually entered the vision of traffic management functional departments and scholars,especially the intersection signal control and path guidance,which are the core part of ITS,are gradually being popularized and applied.However,with the rapid development and improvement of ITS hardware and facilities,the problem of low intelligence of road traffic management system still exists.Traffic management departments are faced with the problem of how to obtain useful knowledge and information from massive traffic flow data,especially the traffic data generated by emergency congestion,and how to make scientific traffic management decisions by making full use of information advantages.Therefore,on the basis of real-time traffic big data,it is of great significance for coping with urban traffic congestion to carry out research on dynamic traffic flow information processing as well as control and guidance strategies of congestion.In order to formulate a special traffic management strategy for traffic emergency congestion,which is different from conventional congestion,the characteristics and propagation laws of urban traffic emergency congestion are summarized and analyzed,and making use of the ability of deep learning which learn the inherent characteristics of data quickly,traffic status is identified and short-term dynamic traffic flow information from a large number of traffic flow data is predicted.On this basis,combined with the traffic flow transmission model,the prior guidance and coordination control strategies of traffic emergency congestion are discussed,from the intersection signal control system and traffic guidance system respectively.The effect of the collaborative scheme are analyzed and evaluated under the simulation environment.The specific research work is as follows:Firstly,the characteristics of emergency congestion in urban traffic are analyzed,including the causes of congestion,the evaluation indexes of congestion and the spatial and temporal distribution characteristics of traffic flow in emergency congestion.The traffic wave theory is used to analyze the law of traffic congestion transmission under different circumstances,and the response scheme process of traffic management departments under traffic emergency congestion is introduced,which provides theoretical basis for formulating specific traffic management strategies.The spatial and temporal correlation of traffic flow parameters of urban traffic emergency congestion is analyzed,and a classified pre-training module is constructed.In view of the characteristics of traffic flow data complexity,non-linearity and uncertainty,combined with the ability of deep learning which learn the inherent characteristics of data quickly and fully,the deep belief network of classified pre-training is constructed to make unsupervised feature study of traffic flow data,and the Logistic regression is introduced to realize effective and real-time identification of traffic status.Experimental results show that the method of traffic flow status discrimination and short-term traffic information prediction based on deep learning can effectively realize automatic congestion identification and short-term traffic flow prediction.Again,considering that the micro-simulation software is restricted by the software function during the analysis,the signal control of the intersection is combined with the queuing situation of the section.In order to accurately and concisely describe the queuing problem,the dual-queuing model of urban traffic burst congestion is constructed.And the theoretical characteristics of the model are proved.Furthermore,a signal control optimization model is established,and the discretized model is solved by genetic algorithm.Numerical examples show that the model can provide a correct description of the dynamic characteristics of traffic flow for the formulation of signal control strategy and realize effective traffic jam control.Again,when traffic events occur,reasonable traffic guidance strategies can avoid or alleviate traffic emergency congestion.The effect of using dynamic traffic flow information on traffic assignment is analysed.Then a route guidance scheme based on prediction information is designed to balance the optimal user and the optimal system.In order to make full use of the prediction information and improve the guidance efficiency,an improved A* algorithm under time-varying road network is proposed.Considering users' personal perception of congestion,the alternative path allocation scheme based on users' degree of urgency is constructed.Again,given that most of the current collaborative optimization strategies require repeated iterative implementation of target optimization and the calculation amount is too large,a collaborative optimization scheme of signal control and guidance system based on predictive information,which uses artificial intelligence and big data depth to fully learn the traffic flow characteristics of traffic congestion is proposed,and transfers the main computing problems to offline,so as to realize the due rapidity of online management decision-making.The time delay of intersection is analyzed,and the signal control and en-road guidance strategies proposed above are improved to facilitate the implementation of the coordinated control scheme.Finally,in order to characterize the effect of different control strategies,a typical traffic simulation environment is established using the micro simulation software SUMO.The O-D matrix data is divided into two groups,and one of the groups of data is used to develop signal control strategy,en-route guidance strategy and collaborative optimization strategy respectively.Meanwhile,the traffic data information obtained is used for training to learn the prediction model parameters,and another group is used as the evaluation experimental data.The simulation results show that the optimization of existing traffic management resources can improve the efficiency of the road network.