Research On Intersection Signal Fuzzy Control Based On Traffic Flow Prediction With SVM

In view of the traffic flow at single intersection with high complexity and randomness, a fuzzy control method to intersection based on support vector machine to predict the traffic flow is proposed in this paper.The traditional intersection control method is that, when the traffic police is leading traffic, he decided whether to delay this green light at the time of the green light will turn off, in which the delay time depended on the comparison of the current phase traffic flow and the next phase traffic flow, then the traffic police decided the time of green light on experience.In this paper, we predicted the traffic flow firstly, which the key problem is how to build a suitable forecasting model. Giving support vector machine is a network model resolving nonlinear regression which has strong prediction capability, high prediction accuracy and fast convergence speed; we adopted SVM and LS-SVM to establish the predicting model. We compared the prediction results, and choose the LS-SVM model with high accuracy finally. Then, we used the traffic flow data as the input of fuzzy controller.Secondly, taking the intersection as the research object, we set up intersection four-phase fuzzy controller and the fuzzy control principles by fuzzy toolbox. Then we used the traffic flow data as the input of fuzzy controller and get the length of time of green light in the next phase by the fuzzy control principles, in order to let the traffic flow which will not line up in front of the intersection totally pass the intersection in next phases.At last, compared with the traditional intersection timing controller, the simulation and analysis are carried out in the Matlab Simulink, which results show that the average vehicle delay time in fuzzy controller based on prediction data are less than the timing controller. At the same time, the improved traffic capacity made the traffic flow pass through the intersection more effectively.