The Research On Characteristics Of Short-term Traffic Flow And Traffic State Change

It is necessary to predict short-term traffic flow fast and accurately in the control and induction of intelligent transportation system, so short-term traffic flow prediction is the core of intelligent transportation system, while to study the short-term traffic flow characteristics is the basis for short-term traffic flow prediction, meanwhile transportation system is a people involved, time-varying, and complex non-linear large-scale systems, making the traffic system operation status and the occurrence of congestion are constantly changing, it brings a certain degree of difficulty to forecast short-term traffic flow, to analyzes the short-term traffic flow characteristics and the operation of state changes becomes necessary.Firstly, this paper describes researches on short-term traffic flow characteristics and traffic state change both at home and abroad, introduces phase space reconstruction theory and discusses the selection of the embedding dimension and embedding delay, also introduces single-variable recurrence plot and recurrence quantification analysis and multi-variable recurrence plot and recurrence quantification analysis respectively, meanwhile an advanced traffic data collection system is introduced, where the original data which is used in this paper is download, then a massive traffic flow data analysis and processing method is introduced . On the above basis, takes 1-minute interval data as example and chooses four different parts of one day ,and then obtains the quantitative values of by utilizing the recurrence quantification analysis, Similarly takes 1-minute interval data as example, respectively visualizes the recursive property of time occupancy and traffic volume time series under recurrent congestion and non-recurrent congestion conditions by applying the cross recurrence plot, and then identifies the change time and obtains the statistical characteristics of different traffic patterns by applying the cross recurrence quantification analysis and analyzes the recurrence variables.It is found that the short-term traffic flow time series is non-linear, non-stationary and appears to be random, chaotic and determined in different time periods by analyzing a single traffic parameter from one-dimensional time series. the results shows that traffic is probably to have an inhomogeneous way of evolving concerning its statistical characteristics and shifts do not occur in a determined way with respect to time, so more accurate forecasting methods should be used in line with short-term traffic flow characteristics. it is also found that more accurate when considering four rather than three regimes of traffic flow by studying time occupancy and traffic volume time series under recurrent congestion and non-recurrent congestion conditions respectively, both free-flow and congestion exhibit strongly deterministic, while free-flow is highly instable, whereas high-speed synchronous flow and low speed synchronous flow are weakly deterministic, their structure are relatively simple, non-stationary, while the high-speed synchronous streams prone to shocks, that is accessible to the low-speed synchronous flow or blocking flow.