A Traffic State Assessment Based on Grey Clustering Analysis

A gray clustering analysis based on traffic state assessment model is proposed in this paper, in which the traffic states are divided into four gray types: free, slight congestion, congestion and serious congestion. According to the methodology proposed in this paper, the whitening function of three traffic parameters, including volume, speed and occupancy corresponding to each gray type, is used to calculate membership of each parameter and then synthesize the clustering coefficient of each gray cluster. Finally, the best classification results are further computed. Special cases are used to check the methodology by collecting traffic data of Beijing. The case study shows that it is effective to identify the traffic state by gray clustering analysis, which can be objectively verified by the gray coefficient entropy in the model. As a result, the subjectivity of manual discrimination and influence of sample noise are reduced. The methodology provides basic information for traffic management decision making.