Research On Chaotic Traffic Flow Based On Rouch Neural Network

Traffic flow forecast is an important research area of the Intelligent Transportation Systems (ITS).Accurate real-time traffic flow forecast is foundation and prerequisite for intelligent traffic guidance and control.Also,it is significance and high value to solve the traffic congestion.This thesis studied an important phenomenon in the traffic flow -the chaos. Further studied rough neural network algorithm for traffic flow forecasti. And the rough neural network algorithm is applied to the prediction of chaotic traffic flow. The contributions are as following:Firstly, introduced the theory of chaos and two important theoretical models of traffic flow.Further analysed them in theory and got the conditions and processes of chaos phenomenon generated in the traffic flow. Then based on two models, gave the analysis from micro sense and concluded five points about the chaotic traffic flow. On this basis, the chaotic traffic flow forecast presented the general steps.Secondly, established a single-crossing and adjacent intersection traffic flow forecast model.From the current traffic flow forecast methods, this paper selected neural network algorithm as the research object, and improved the three neural network algorithms with combination of rough set theory. The improved algorithms had better generalization ability and forecasting accuracy. The improved algorithm had better generalized ability and higher forecasting accuracy. Last gave the simulation analysis.Thirdly,mproved the rough neural network algorithms were applied to the forecasting of chaotic traffic flow,and gave the simulation analysis respectively, It's proved that the improved RBF network had better forecasting effect.Finally,gave a summary and prospect of this paper. Summed up the gains and achievements,instructed issues needed to be further explored.