Real-Time Intelligent Recognition of Chaos in Traffic Flow Using Reduced Support Vector Machine

Abstract

The real-time recognition problem of chaos in traffic flow was studied using reduced support vector machine. Based on analyzing the demand of intelligent transportation system and the problems of the exiting recognition methods, the intelligent real-time recognition method of chaos in traffic flow was proposed. The principle and the structure of the system are briefly introduced. There are online recognition subsystem and offline recognition subsystem mainly. Normal methods are used in the offline recognition model. The online recognition model was established using reduced support vector machine, which the wavelet packet energy features vector of the anterior time series of traffic flow in every training samples were used as input variables. The reduced samples are treated as the final training sample. When the new samples are appended, the reduced set may be updated. The rules of models can be updated and the problem of large sample size could be solved. The simulation result shows that the method is correct and feasible. And it can satisfy the real-time requirement of chaos recognition.