Short-Term Traffic State Forecasting Of Road Network Based On Nonparametric Regression

The short-term traffic state forecasting often used in the dynamic traffic guidance, the advanced traffic management and the traffic control and safety, plays an important role in the research area of traffic engineering. So study on short term traffic state forecasting is very popular in the world during the past few years. Traffic engineers have done much work in this area, and get lots of remarkable fruits. Further study is done in this thesis:(1) We expound the important role and status of short-term traffic state forecasting in Intelligent Transportation System, which result in the necessary and importance of short-term traffic state forecasting research.(2) The strength and weakness of some popular methods are introduced. Then, this thesis provides an overview of the research in short-term traffic state forecasting, which classified into single Section prediction and multi-Section prediction.(3) The three basic parametric of traffic flow are introduced and based on traffic state time and space distribution on urban traffic network, main characteristics of traffic are introduced and analyzed. Then this thesis introduces the method to repair traffic data acquired from detector.(4) After reviewing the application of nonparametric regression in short-term traffic state forecasting, this thesis introduce the achievement of nonparametric regression in this area. And then, this thesis present a new nonparametric which can be used for short-term traffic state forecasting in road network.(5) The presented model is examined by the traffic data of Beijing. By comparing the results of different K and input, we get best prediction with K=10 and the most similar state as input of prediction function, following by the analysis on result of the best prediction.At last, the strength and weakness of this new model and the development and problems of nonparametric in short-term traffic state forecasting in future are listed.