Urban Freeway Network Density Estimation Based On Dynamic Graphhybrid Automata

In order to inform the traffic traveler freeway congestion in a timely manner andto avoid the traffic demand increases the more serious traffic safety problems,it is necessary to monitor the freeway traffic conditions timely.However,due to the large scale of urban freeway network,it will have a high cost if each section has sensor installation.At the same time it is likely to introduce noise in the process when sensorsare detecting of traffic information or transfering data to the computing center.Therefore,it is necessary to propose a fast and accurate method for estimating the traffic state of large-scale urban freeway network.The traffic data collected by a limited number of sensors are used to study the status of the whole freeway network,so that the traffic managers can grasp the whole road network operating conditions,and take reasonable management measures to improve the capacity of the freeway better.Compared with other traffic parameters,the traffic density can reflect the state of the road network more intuitively and adaptively.Therefore,this paper use freeway traffic density estimation model from the perspective of macro,which is based on the cell transition model(CTM)and dynamic graph hybrid automata(DGHA)model.And the Kalman filter is used to estimate the traffic density,and it is applied to the example of Beijing third ring road freeway.And the distributed optimization of the sensor network is given in order to achieve accurate freeway network estimation and reduce the resource loss caused by redundant sensor layout at the same time,reduce the calculation pressure of calculation center.The main research contents include:(1)Research on traffic network model based on cell transition model and dynamic graph hybrid automata.The traffic flow dynamics of the road are selected by the cell transition model.Because of the non-linear characteristics of the cell transition model,the computational complexity is increasing greatly when the large-scale road network is built.Therefore,the cell transition model is combined with the dynamic graph hybrid automata model,and the CTM model is discretized into segmental linear affine system.And we choose it as the theory basement of our paper.Build a modular and easy-expanding freeway mode algorithm to realize the theory.Example for application is given combining with the Beijing third ring freeway.(2)A method of traffic density estimation based on centralized Kalman filter is proposed.For the section where the sensor is arranged,the measurement error is reduced by Kalman filter.For the section where the sensor is not arranged,using the value derived by model as a status estimation.(3)Establishment of distributed traffic density estimation model based on consensus Kalman filter.In order to reduce the computational burden of the computing center,improve the state estimation speed,dividing the function and coverage of the calculation center,use a number of roadside computing units instead of the computing center,and use the consensus Kalman filter instead of the centralized Kalman filter for the traffic state estimation.Experiment shows that the distributed estimation model consumes less computation time under the premise of ensuring the accuracy of estimation.(4)Build dynamic traffic status display platform.In order to show the results of the traffic density estimation,the dynamic traffic status display platform is constructed in the multi-autonomous Netlogo environment according to the actual situation of Beijing third ring freeway.The freeway is divided into different sections,road density estimation results are represented by the degree of congestion of different color visual presentation,and monitoring historical trends of sections' density,and intuitively display the evolution of traffic flow in freeway.