Research On Driver Model Based On Driving Behavior Characteristics

In recent years,with the continuous development of intelligent transportation networks,people have put forward higher requirements for intelligent driving.As the carrier of intelligent transportation,intelligent vehicles play an extremely important role in smart travel,traffic safety,energy conservation and environmental protection.As a key part of the intelligent assisted driving system,the driver model is of great significance to continuously innovate its research methods,improve its simulation accuracy,and improve its test content.With the continuous development of driver models,traditional models can no longer meet the needs of indoor testing and assisted driving system design.Many domestic enterprises and research institutes are working on the study of the relationship between driving style and automatic driving or traffic games.It is urgent for them to build a driver model that can characterize driving style.Therefore,this paper uses real road test data to analyze the influence of driving behavior characteristic parameters on driving style,and establishes a driver model based on driving behavior characteristics.The main contents and innovations of this paper are as follows:(1)Based on the real road test data,a driving style recognition model based on machine learning algorithm is established,and the characteristic parameter indicators that can evaluate the driving style are summarized.Firstly,by analyzing the influence of driving behavior habits on driving style,some original data are extracted as driving style characteristic parameters;then,principal component analysis method is used to reduce the dimensionality of the characteristic parameters,and the comprehensive characteristic parameters of driving behavior that can characterize driving style are obtained.The Kmeans clustering algorithm is used to classify the driving style.Finally,based on the random forest algorithm,the driving style recognition model is established with the comprehensive parameters of the driving behavior of different driving styles as the training data.The simulation results show that the recognition model has high recognition accuracy and can be used as the basis for selecting driving style evaluation indicators.(2)According to the influence of road conditions on vehicle steering characteristics,considering lateral movement on vehicle driving,a multi-constraint vehicle dynamics model considering path curvature and lateral slope is established,which is used as model predictive control.A multi-constraint model prediction(MMPC)controller was designed to build a lateral driver model.The simulation results show that,compared with the traditional MPC controller,the MMPC controller has better path tracking accuracy and driving stability,and can better adapt to complex paths.(3)In order to solve the problem of vehicle tracking under complex paths,considering the coupling effect of longitudinal vehicle speed on lateral control,a modeling method for lateral and longitudinal control driver is proposed.On the basis of the above lateral driver model,the critical speed of vehicle rollover and sideslip is derived,and the maximum safe speed of the reference path is predicted.The MMPC controller is used to obtain the optimal front wheel angle control sequence,and the PID controller realizes the follow-up of the longitudinal speed.Through the Simulink-CarSim cosimulation,the results show that the lateral and longitudinal control driver model can predict the longitudinal vehicle speed according to the road conditions,and improve the path tracking accuracy and driving stability.(4)A driver model based on driving behavior characteristics is established by integrating the driving style evaluation indicators.The model is based on the lateral and longitudinal control driver model,and the lateral and longitudinal driving style evaluation indicators are used in driver modeling.Finally,a hardware-in-the-loop(HIL)test system is built,and the HIL simulation verifies the validity and robustness of the model under complex paths.