Driving Behavior Analysis Based On Data Mining Technology

The car have being the most popular means of transport,which led to millions of traffic accidents each year,resulting in a huge casualties and property losses.Therefore,the prevention of traffic accidents is of great significance.In the past,improved vehicle design and road construction had been implemented,but still without a significant breakthrough.Nowadays more and more attention has been paid to the driver's driving behavior.However,there is still lack of a unified quantitative indicators to evaluating the driving behavior and even only a handful of quantitative indicators to judge the level of driving safety.Therefore,based on the OBD technology,our research aimed at analyzing the driving behavior from five dimensions: mileage,the types of operation,the driving speed,the acceleration and the engine rotational speed.Due to the long-term stability of driving habits,we collected the vehicle driver data and then extracted the quantitative index of driving behavior by some calculation methods.The data of this study came from the so called “Intelligent box” which collected the driving data more than 2 years and the main driving area of the driver is Chongqing.“Intelligent box” should connect the OBD interface which commonly used in automobile to acquire the driving data,then display the driving operation data in the mobile phone and delivery it to the server via Bluetooth,so that the vehicle running data of each driver can continuously record.Currently on the market most of the car using a unified international standard OBD II,expediently “intelligent box” can get real-time access to different manufacturers,different models of vehicle operation data through the OBD interface.With the development of OBD technology,the data of vehicle running can also be increased.In this paper,the vehicle running data related to driving behavior are extracted from five dimensions,and obtain 57 indicators after transformation,then utilize the stability analysis of time series in financial field to test the stability of the 57 indexes,finally we got 17 stable indexes which can be used as the quantitative indicators to evaluate driving behavior.On the basis of the quantitative index of driving behavior,the driver is classified.The results show that drivers can be distinguished by the use of quantitative indicators of driving behavior.Our study provides a theoretical basis for the quantitative study of driving behavior and provides a new way for accident-prone research.The data obtained by the OBD method is easier than the traditional experimental methods,and the results are more authentic.