Study On Identification Model And Calculation Method Of Driving Tendency Based On Dynamic Driver-Vehicle-Environment Data

Along with the development of urbanization and popularization of vehicles, automobile has brought about enormously travel convenient as one main transportation mean, but it has come up with traffic safety question simultaneously. The vehicle driving-assistance system—a relatively positive and effective measure dealing with traffic safety question, can provide protection for the person-vehicle unit and avoid accident under unsafe driving situation. Among traffic safty influential aspects, internal factor of driver is one of the key points, the vehicle-road-environment affect the traffic safety simultaneously through people—the driver. The present driving-assistance system’s (especially the auto-active-safty early warning system) capacity of accurate protection for drivers and vehicles have a relatively low level, because the ignorance of the individual. Driving tendency is one of the driver’s features, it can reflect the differences between different drivers. Provided that can be used in auto-active-safty early warning systems while identify the driver’s type via recognition model. Personalized auto-active-safty early warning system can be established according to the different driver types and the relatively accurate protection for the person-vehicle unit can be achieved.Aimed at achieving tendency recognition, driving tendency can be categorized into three types:risk-taking type, cautious type and conservative type. Research of driving tendency recognition method toward different traffic condition such as the unlimited flow and the car-following flow were conducted:driving datas of different drivers’ types under the unlimited flow and the car-following flow were obtained through psychological questionnaire tests, observed experiments, real vehicle experiments and the interactive parallel driving simulated experiments. The feature extraction methods based on BP neural network and rough set are used to extract eigenvectors with good classification abilitv of driving tendency, under the traffic condition of unlimited flow and the car-following flow. The pattern recognition method based on support vector machine is used to establish dynamic recognition model of driving tendency. Judging indicators of tendency type under car-following situation were given combined with vehicles’ kinematic theory, and dynamic identity model of driving tendency were established. Driving tendency recognition models were verified by real vehicle experiments, the interactive parallel driving simulated experiments and the traffic flow micro-simulated experiments. Results show that the design schemes of experiments, the extraction methods of eigenvectors and established identification models are reasonable and feasible, which can achieve the dynamic recognition of driving tendency.