The Short-time Prediction Model Of Traffic Flow Based On Taxi Tragectory Data

With the development of the city and transportation industry, traffic jam phenom-enon become more and more serious, showing the trend from the point to the line, from the local to a wider range. Traffic jams not only results in the decrease of road traffic capacity, traffic speed drops, at the same time bring the environment pollution, traffic safety and a series of economic problems, seriously restricting the development of cities. The essence of urban traffic jams is the condition of traffic resource under a specific spare and time serious imbalance between supply and demand. According to the charac-teristics and propagation rules of space-time of traffic flow, traffic control and traffic guidance as the main means to realize the reasonable distribution of traffic flow between different road, coordination of urban road traffic of liquidity and safety and economy, is considered to be an effective way to prevent and ease traffic jams, improve the efficiency of the entire road, traffic flow prediction technology as the basic research of traffic management has been widely concerned.At the base of referencing and fusing classical speed-flow model, the research of traffic flow aspects related home and aboard, on the basis of using VISSIM simulation experiment platform, the paper do the research of traffic flow evolution at the condition of interrupted traffic of primary and secondary road of the city and give a short-time forecasting method of traffic based on the taxi data.Traffic flow prediction method by taxi consists of floating car data preprocessing, variety fitting the vehicle speed and road traffic flow prediction model. Data preprocessing is to remove potential anomalies in the data sources and get valid data; The main task of varied vehicle velocity matching stage is to estimate the speed of the traffic flow by dealing the effective data through an adaptive exponential smoothing algorithm. Comparing with the classical traffic flow model, the main task of the traffic flow prediction model is to establish three phase speed-flow model divided by speed, and gives the method of parameter enactment on the base of three phase speed-dentity model. In the last, the paper reckons the traffic flow of the particular section of VISSIM simulation and validates the efficiency of the model by adopting the method putted forward in the article.