An Efficient Method for Driver Abnormal Lane-Changing Identification Based on Trajectory Data

Risky lane-changing (LC) behavior of vehicles on the expressway has adverse effects on traffic safety. This paper performs a method of risky LC behavior recognition based on the particle swarm optimization-based support vector machine (PSO-SVM) combined with cluster analysis. In this study, the CitySim data set is processed using time windows and extracted driving behavioral characteristic parameters. Then a clustering approach is used to identify sequences of risky LC behavior. The PSO algorithm is adopted to find the optimal penalty factors and kernel parameters, and then the samples are applied to train and test the proposed model. The accuracy of the model is 98.2%, which can identify the LC behavior high performantly and support future research on risky LC behavior.