Investigating the Economical Classification of Driving Behavior Utilizing Dynamic Thresholds

This research introduces a methodology for classifying the economic efficiency of driving behavior by employing dynamic thresholds and utilizing real-world vehicle operation data. Initially, the real-world vehicle data are divided into short trips based on the speed segments from two instances of vehicle stopping, and the traffic conditions during these trips are classified using the K-means clustering algorithm, thereby obtaining the traffic state for each short trip. Following this, characteristic variables are selected to represent the driving behavior characteristics of the drivers, and the eco-scores along with the feature variable values for all short trips are computed. Spearman correlation analysis is then used to identify feature variables that exhibit strong correlations with eco-scores, serving as economic evaluation indicators. Subsequently, dynamic thresholds for the evaluation indicators under varying traffic conditions are calculated, thereby establishing a method for classifying driving behavior based on these dynamic thresholds. The proposed method effectively eliminates the influence of traffic conditions on driving behavior, consequently yielding more objective classification outcomes. This study provides a solid theoretical foundation for developing personalized energy-saving driving strategies tailored to different driver types. In addition, it offers critical insights for the design of intelligent transportation systems, the development of advanced driver assistance systems, and the formulation of environmental policies. In practical applications, the dynamic threshold-based classification method for economic driving behavior can deliver real-time feedback to drivers, aiding in the optimization of driving behavior and further enhancing driving efficiency.