Freeway Traffic State Estimation Based On Fixed And Mobile Sensing Data

Real-time traffic state estimation(TSE)is a basis for freeway intelligent traffic control.The development of Internet of Vehicle provides unprecedented opportunities for TSE.It is expected that fixed and mobile sensing data would coexist for a long period.This paper focuses on the freeway traffic state estimation based on mixed sensing data,and proposes several general TSE methods suitable for any road network topology,fixed detector configuration and the market penetration rate(MPR)of floating cars.The proposed methods can simultaneously estimate traffic flow variables,traffic flow model parameters and the MPR of floating cars with high spatiotemporal accuracy.First,based on the CTM model and the METANET model,this paper proposes two general methods for freeway traffic state estimation under the condition of mixed sensing data,and realizes the joint online estimation of the traffic states and model parameters by means of the random walk equation.The performance of the above estimation methods and its sensitivity to the MPR of floating cars are evaluated using NGSIM data.Second,previous works have confirmed that online model parameter estimation(OMPE)is important for TSE of using fixed sensing data.Under mixed sensing case,OMPE has hardly been studied.Thus,this paper raises several important questions:(1)Does OMPE still necessary for TSE under mixed sensing case?(2)Will the increase of the MPR of floating cars reduce the dependence of traffic state estimators on OMPE?(3)will the increase of the MPR of floating cars improve the performance of traffic state estimators?(4)Will the increase of the MPR of floating cars make up for the inadequacy of the first-order model based estimation method(compared to the second-order model based estimation method)?Third,based on the assumption of speed-uniformity and the dynamic modeling of the MPR of floating cars,this paper presents two general methods for estimating freeway traffic states that do not rely on OMPE,and also uses NGSIM data to evaluate the performance of the two methods and their sensitivity to the MPR of floating cars.Finally,considering the inherent deficiencies of NGSIM data in terms of the length of the road and the span of the time,this paper relies on a high-accuracy micro-simulation platform to verify the effectiveness of the above four TSE methods.This paper reached the following conclusions:(1)Under the mixed sensing conditions,OMPE is still indispensable for freeway TSE;(2)With the increase of the MPR of floating cars,the dependence of state estimation on OMPE is weakened;(3)The increase of the MPR of floating cars can significantly improve the performance of the estimators;(4)With the increase of the MPR of floating cars,the influence of model differences on TSE is weakened.In addition,according to the different MPR of floating cars,this paper gives specific suggestions for the selection of TSE methods.The mainly contributions of this paper are as follows:(1)Four general methods for freeway traffic state estimation under mixed sensing case are proposed;(2)A general method that can accurately and effectively estimate the MPR of floating cars is delivered;(3)It is revealed that the increase of the MPR of floating cars can reduce the dependence of TSE on OMPE;(4)It is confirmed that the performance of the traffic state estimator that using mixed sensing data improved with the increase of the MPR of floating cars;(5)The compensation effect of the increase of the MPR of floating cars on the model differences involved in TSE is revealed.