Abstract: In order to explore the driver characteristics through the risk distribution of various factors in car-following condition, and complete the driving style classification under vehicle-road interaction, this paper proposes a driving style classification method based on improved fuzzy comprehensive evaluation method. Firstly, the test data are collected through driving simulator. The motion parameters of the vehicle and the driving safety field are selected as characteristic parameters of driving style classification. Secondly, the combination weighting method is used to optimize the weight of the fuzzy comprehensive evaluation method, so that the corresponding weight is given to each characteristic parameter, and by using the improved fuzzy comprehensive evaluation method, drivers are divided into cautious drivers, common drivers and aggressive drivers. Finally, K-means clustering algorithm is used to verify the rationality of the above method. The result shows that the difference rate between the K-means clustering and the improved fuzzy comprehensive evaluation method is only 2%, and when the number of clusters is 3, the silhouette coefficient is up to 0.685, that is to say, just like unsupervised learning algorithm, the fuzzy comprehensive evaluation method can effectively classify driving styles.
Keywords: driving style classification; driving safety field; fuzzy comprehensive evaluation method; combination weighting method; K-means clustering algorithm
合肥工业大学学报自科版