Study on Accident Impact Situation Cluster Analysis

The thesis analyzes the characteristics of accidents and their impacts on freeways, and proposes the concept of accident impact scenarios. From the perspective of accident attributes, traffic attributes, and external environmental attributes, it identifies the factors affecting the accident, and builds an accident impact scenario clustering index system design at speed of 100 km/h. Based on the matter-element extension model, clustering studies are made on accident impact scenarios. Considering the large number of highway accidents and the fact that accident data is difficult to obtain, this paper selects the middle value of each indicator’s classification threshold and arranges and groups them. Each type of combination has a certain degree of representativeness. The eigenvalues of the level variables under various combinations are calculated, and the accident scenarios belong to the same level of the eigenvalues of the level variables are clustered, and four possible accident impact scenarios are obtained.