Urban Road Traffic Condition Pattern Recognition Based on Support Vector Machine

As an important part of the modern intelligent transportation system, urban transport condition recognition is the base of intelligent control, guidance and synergy system. This paper establishes a three-dimensional space with traffic volume, average speed and occupation ratio. It then classifies transportation condition patterns in terms of blocking flow, crowded flow, steady flow and unhindered flow based on wide literature review. Furthermore, this paper presents the algorithm with the MATALB LiBSVM toolbox. To process the data, this paper compares the classification result of different SVM kernel functions and thus realizes the transport condition pattern recognition via the support vector machine (SVM). The results reveal that the selected indexes effectively reflect the characteristics of the traffic conditions. The SVM kernel function can separate different patterns from traffic flows with high classification accuracy, and the data normalization has a significant influence on the result of classification.