Real-Time Freeway Traffic State Estimation Based On Kalman Filter

With the rapid development of the social economic and freeway system, it has plays an important role in social and economic activities. However, limited detection equipments can not provide the freeway traffic states at any time.So the premise freeway traffic state estima-tion becomes the key of developing the effective traffic management and control strate-gies,and also the key to intelligent traffic management system operation.In this paper,relying on the National Natural Foundation of highway speed discrete characteristics and freeway state estimation and safety warning system funded by Nanjing Science and Technology Bureau,the research concentrated on analysing the spatio-temporal correlation characteristics of continuous freeway segements based real measured data from Pems database,to realise the traffic state estimation of the blind detection area through the es-tablishment of freeway traffic state estimation model based on kalman filter.Fistly,through the actual survey datas,bying introduction of the spatio-temporal correla-tion coefficient to study the time correlation and spatial correlation of continuous freeway segements traffic flow parameters.Then realise the choose of estimated area and traffic states estimation of blind detection area.Secondly,research on the actual traffic state estimation effcet.By genetic algorithm for online calibration macroscopic traffic flow model with different parameters,and application the calibrated model on the strong time and spatial relevant freeway segements to estimate the traffice state of blind detection area.Then comprehensive statistical evaluation index to se-lecting the optimal traffic model as the based state estimation model. In the end,discussing the sensitivity of the choosed model parameters.Additionally,according to the "recursive-estimation-fixed" principle build freeway traffic state estimation model respectively based on Extented Kalman Filter and Unscented Kalman Filter, and the general steps of traffic state estimation is given.Finally, throuth matlab implementation the practical application evaluation of real-time traffic state estimation based on kalman filter,mainly includes the tracking capability of muta-tion between two kinds of state estimation model and constrastive analysis of state estimation error. Besides,further research on the state estimation effect of different detector layout scheme,gives the practical error of five detector layout schems and provides references basis for detection equipments layout.