Research And Application On Expressway Intelligent Prediction Model Of Traffic Congestion

Highway operating management is developing towards the direction of intelligence and automation.It's necessary that governors know well about the formation and development of traffic congestion accurately and in time and analyze massive traffic information scientifically,in order to meet traffic demand and keep traffic safe.The ultimate causes of expressway traffic congestion are traffic volume greater than the capacity and traffic accidents.When daily traffic congestion occurs,the congestion can be avoid by traffic flow control measures before becoming the congestion state.When traffic accidents occurs,the congestion cannot be avoid.In order to develop a contingency plan scientifically and improve the efficiency of emergency rescue,governors need to grasp the spread severity of accidents and congestion duration to control the impact of traffic accidents.This paper forecasts the traffic congestion in two conditions.The core contents contain the study of short-term traffic state forecasting model and the prediction model of time and space effect.The main contents include:(1)Ten kinds of traffic flow forecasting methods were introduced and the advantages and disadvantages of various methods were summarized.Appropriate forecasting model for this study could be selected based on those advantages and disadvantages.(2)The input traffic parameters and data sources of short-term traffic state forecasting model were determined and the pretreatment methods of data were introduced.While time characteristics of freeway traffic flow were analyzed,KNN nonparametric regression prediction method was chosen to predict traffic flow time series with highly uncertain but regular characteristics.Short-term traffic state forecasting model based on KNN algorithm was proposed.(3)Indicators that can represent accident impact was determined.The influences of different factors on the diffusion and dissipation mechanism of congestion caused by traffic accidents were analyzed.According to the factors,congestion diffusion and dissipation process in accident section was simulated by VISSIM.After acquiring simulation results,nonlinear regression analysis was used to establish function relation between different influence factors and accident impact indicators.The prediction model of time and space effect based on traffic simulation was proposed.(4)Measured data of Kaiyang highway with Guangzhou direction acquired by microwave equipment was used to analyze prediction results of short-term traffic state forecasting model and the prediction model of time and space effect.Two prediction models were applied in highway intelligent traffic congestion prediction system.System demand and framework were expounded.Above all,highway intelligent traffic congestion prediction system was established and it was proved that the system was reasonable,feasible and effective.