Driving mode mining of pure electric taxi based on kinematic segments
In the context of the gradual transition to comprehensive electrification of taxis, and address-ing the current shortcomings in evaluating the driving state of pure electric taxis, a method for mining driving characteristic patterns based on kinematic segments of pure electric taxis is established. This study aims to explore the driving characteristics of pure electric taxis. Firstly, utilizing GPS data from driving trajectories, 13 feature indicators, including overspeed ratio, acceleration and deceleration fre-quency, driving speed, and idling time ratio, are determined from 3 aspects in terms of speed charac-teristics, acceleration and deceleration, and driving conditions to characterize kinematic segments. This establishes a method for extracting kinematic segments of pure electric taxis and subsequently studying their driving characteristics. Subsequently, based on the eigenvalues and cumulative contribu-tion rates of principal components derived from driving characteristic indicators, key feature indicators are identified. Through integration with the K-means clustering algorithm, a method is proposed for mining driving characteristic patterns based on kinematic segments of pure electric taxis, allowing the identification of driving characteristic modes in various time and space scenarios. Finally, leveraging 7 million GPS tracking data points from pure electric taxis in Shenzhen, with a sampling interval of 1 second over 9 days, 1757 kinematic segments of pure electric taxi are extracted. Employing eight key feature indicators related to safety, efficiency, and comfort for cluster analysis, a feature pattern li-brary is generated, encompassing 27 classes of driving states for pure electric taxis on main roads, sec-ondary roads, and local roads during morning peak, normal hours, and evening peak periods. The re-search findings indicate that, combining the three aspects of safety, efficiency and comfort, pure elec-tric taxis travelled better during the morning peak than during the flat peak and evening peak hours. A pure electric taxi driving feature pattern mining method based on kinematic segmentation, principal component analysis and spatial-temporal scenario clustering analysis can effectively reflect and evalu-ate the driving status of pure electric taxis and provide reasonable driving suggestions to drivers.
traffic engineeringdriving characteristic patternkinematic segmentpure electric taxi
万方数据
维普
