Analysis And Prediction Of Short-term Traffic Flow In Road Network

Short-term traffic flow forecasting is a question of the first importance for the road traffic control and induction systems. At present, traffic flow analysis and forecasting have become a principal topic in the field of traffic engineering.It is also known that, due to the influence and constraint of various complex conditions, traffic flow time series show the special characters, such as lack of information and uncertainty, nonlinearity and complexity. At the same time, road traffic is a web-like structure and various sections are interrelated and interactional.So traffic flow data of multiple sections has correlation, cyclical fluctuations and time-lag. Thus, conventional traffic flow forecasting models can not obtain satisfactory results.Grey generating space model takes the system of insufficient data and uncertainty as research object. First, grey generating advances the level of original information, then it control the system through establishing model which contains difference and differential. In addition, new grey system models, such asGM(τ,r) series not only adapt the lack of information and uncertainty of traffic flow time series but also fit the nonlinearity, periodicity and time-lag. Hence, grey forecasting models come up to the basic characteristics of short-term traffic flow time series.In view of the section of single and multiple, this article does some work about traffic flow prediction as follows:(1) Predict short-term traffic flow of single section. Although the existing single sectional traffic flow forecasting models have reached a certain precision, but only one model hard to avoid limitations and incompleteness. So chinese and foreign scholars combine different forecasting models to predict traffic flow. For the cyclical fluctuations of traffic flow time series, this paper builds composite pattern of grey-cycle extension. Meanwhile, we apply the model to forecast the traffic flow of single section. The results show that composite pattern has high accuracy.(2) Establish multiple sectional short-term traffic flow forecasting model. Firstly, we analyse and prove the basic nature of MGM(1,N) model. Then apply it to the multiple sectional short-term traffic flow prediction.Then for the delayed, periodicity and insufficient data of traffic flow, the MGM(1, N|tan(κ-τ)p, sin(κ-τ)p) model is put forward in order to enhance forecast accuracy. Then we study modeling process, parameter estimation and so on.In the end, MGM(1,N|tan(κ-τ)/p,sin(κ-τ)/p)model is applied for multiple sectional short-term traffic flow forecasting comparing with the results of MGM(1,N) model, it is proved that MGM(1,N|tan(κ-τ)p,sin(κ-τ)/p) model is an effective method of analysing and forecasting.