Research On Urban Traffic Flow Prediction Using Fractal Theory And Vehicle Path Guidance Technology

Traffic Guidance System plays an important role in the Intelligent Transport System, and the key to its establishment is whether traffic flow parameters can be dealt with effectively. Among them there are short-time traffic flow prediction, traffic status judgments and individual guidance service and so on. Traffic flow prediction could provide extension information in time-scale for traffic status space-time judgments, and traffic status space-time judgments provides dynamic information for obtaining individual vehicle path.In order to solve the problem that real-time performance and accuracy couldn't be obtained simultaneous in practical prediction; this paper puts forward a forecasting model based on fractal and weekly-similarly. For the accuracy is too low in catastrophe point when forecasting using traditional fractal model, weekly-similarly is used to remedy the error and configuration parameters. Via this method, accuracy and real-time performance have been improved simultaneously. Furthermore, how many time steps of this method is studied.In practical, manual traffic judgment exists defects, in order to solve this problem, a traffic status judgments model based on parzen-window is proposed. in this method, traffic status center characteristic is obtained via data mining. Then parzen window function is used to get probility of the status using real data; simultaneously, via similarity comparison of road cross in road network and combining with forecasting, a model based on spatial information and time scale is put forward.Owing to the diversity of travel demands in individual service, especially pursuit the optimal route under the condition of variable aim causes some difficulties. Aimed to this problem, an optimal road condition model and multi-objective optimal model has been established on the condition of using GIS forecasting data and traffic status as weight. Then, an improved heuristic genetic algorithm is applied to obtain individual vehicle path to improve real-time performance.Finally, During the research on the Important Project for Key Subject of Zhejiang Province" Research and Application of the Key Issues in Urban Intelligence Traffic Guidance System "(2006C13100) ,these theories and methods above are put into practice, as well as, theHangzhou traffic guidance information system is developed which is consisted of threesubsystems. The practice proved that these theories and methods are available and exact.