Research On Driving Behavior Based On High Frequency Operation Data Of Pure Electric Vehicle

In recent years,with the continuous growth of domestic vehicle ownership,road traffic is increasingly congested,traffic accidents are also frequent year after year.How to reduce traffic accidents has become the focus of traffic research.It is of great significance to study the driver’s driving behavior to reduce traffic accidents and promote the development of intelligent transportation and autonomous driving.Compared with traditional fuel vehicles,the operation data collection of pure electric vehicles is more convenient,the data quantity is more comprehensive,and the action response is fast,which can provide more information for the study of driving behavior.Firstly,this paper introduces the high-frequency operation data acquisition method and characteristics of pure electric vehicle.The data is divided into multiple single driving event data according to the division principle.The dirty data in the original data is detected and processed.The driving behavior data set is constructed by screening out the signal data used in the following driving behavior analysis.The characteristic parameters cover speed,acceleration,steering wheel control and energy consumption,etc.,which can represent most of the driver’s behaviors in the driving process.Secondly,in order to better understand the driver’s driving behavior,the stable and convergent driving behavior data set is used to study the distribution characteristics of each driving behavior parameter and their mutual influence rules.This paper proposes a multi parameter combination threshold boundary line method to identify dangerous driving behavior and establishes six representative dangerous driving behavior identification models,including overspeed,rapid acceleration,rapid deceleration,rapid steering,steering overspeed and fatigue driving,which provided data support for the construction of the following driving style classification and scoring model.Thirdly,the driving style is introduced to understand the driver’s driving habits.Based on the driver’s driving behavior parameters and dangerous driving behavior recognition results,the clustering effects of various dimensionality reduction and clustering combination algorithms are compared and discussed.The driving style classification model based on dangerous driving behavior recognition is established by t-SNE and GMM combination algorithm.The results show that there are great differences in vehicle operation and energy consumption under different driving styles,which provides a reference for the design of driving behavior scoring model.Finally,according to the vehicle operation data and dangerous driving behavior recognition results,the driving behavior score index system is constructed,including driving operation,driving parameters and fatigue driving and energy consumption.A complete and scientific driving behavior score model is established by using AEW-AHP method,which can score the driving behavior and classify driving safety grade according to the driving data information.This model provides certain reference for fleet management,vehicle insurance claims and traffic safety management.