Real-Time Freeway Traffic State Estimation Based on Cluster Analysis and Multiclass Support Vector Machine

Urban traffic state analysis plays an important role in the solution of traffic congestion problem. To estimate traffic state effectively is a foundational work for improving traffic condition and preventing traffic congestion. In this paper, a novel pattern-based approach is proposed to model the clustering and classification of traffic state. First, fuzzy-set clustering method is utilized to divide the traffic state into a number of patterns. Then multiclass support vector machine (MSVM) is applied to estimate these states with real-time traffic data. The result shows that the proposed approach is promising for the dynamic estimation of road traffic state and can provide forecasted congestion information for the traffic control system and traffic guidance system.