Research On Car-following Control Algorithm Based On Driver Characteristics

This project relies on the Science and Technology Project of Jilin Province Education Department,Distributed Electric Vehicle Dynamic and Collaborative Control Based on Chassis-by-Wire System(Project Number: JJKH20200963KJ).With the continuous deepening of economic reforms,people’s quality of life continues to improve,and people’s needs for cars are getting higher and higher.This has led to a substantial increase in the number of cars in our country,which has led to a rise in traffic accident rates,environmental damage,congestion and energy waste.Realistic problems and demands continue to drive the development and growth of intelligent driving.As an important part of intelligent driving,car-following behavior control is one of the focus and hotspots of research.On the premise of ensuring safety,taking into account economy,followability and comfort at the same time is one of the focus in future carfollowing control research,especially comfort.Comfort is not only related to physical parameters such as acceleration or jerk,but also related to driving style.Car-following control that conforms to the driver’s driving behavior style can not only meet their psychological expectations,but is also one of the keys for the driver to accept the car-following control.This paper mainly models the driver’s behavior characteristics,and establishes a carfollowing control strategy based on this model.Through the comparative analysis of existing related research on car following control and driving style,we summarized the current carfollowing control to meet the driver’s driving personality as the goal,while considering the influence of curve curvature on longitudinal driving.Based on the piecewise affine systems,the driver’s longitudinal driving behavior model is established;then in order to ensure safety while taking into account economy,followability and comfort at the same time,a multiobjective coordination strategy is established basing on model predictive control(MPC).The MPC control makes full use of the past acceleration information of the preceding vehicle.Finally,the car-following control strategy is experimentally verified on the experimental platform.The main work of this paper is as follows:(1)This paper uses the Trucksim model as the longitudinal dynamic model,and then uses the logarithmic sampling method to select different curvatures,and establishes an experimental highway with multiple curvature curves.Then an Inverse Longitudinal Dynamics Model was established based on the Trucksim longitudinal dynamics model.The verification showed that the model can meet the research requirements.Finally,a two-car-following model is established,which lays the foundation for the model predictive control design.(2)In order to fully capture the characteristics of the driver’s longitudinal driving behavior,this paper selects the most widely used model in the hybrid system,the Piece Wise Affine(PWA)as its description method.After setting up the experimental platform,design experiments to obtain the driver’s car-following behavior data,and perform high-dimensional cluster analysis on the data.Finally,use the clustering results to identify the parameters of the subsystems.At the same time,use the clustering results to train the Back Propagation Net(BP Net),which is used to replace the switching plane in the traditional PWA model.(3)This paper analyzes the rationality of using the Autoregressive Integrated Moving Average model(ARIMA)to predict the acceleration time series of the preceding vehicle,and then discusses the model predictive control based on the ARIMA model(ARIMA-MPC).After analyzing the characteristics of the acceleration time series of the preceding vehicle,it is decided to use the fmincon function of Matlab to identify the parameters of the ARIMA model.(4)Car-following goals were analyzed,and the following performance,fuel economy,comfort and safety were divided into sub-goals and quantified design.Finally,the quantitative indicators are integrated,the relaxation factor is introduced,and the prediction result of the acceleration of the preceding vehicle is applied to the model prediction,so that the ARIMAMPC control is transformed to the optimal control problem,which provides a basis for its solution.(5)In order to verify the effectiveness of the car-following algorithm in this paper,based on the experimental platform,this paper designs two comparative car-following controllers based on PI control and Linear Quadratic Regulor(LQR)control,and also designs 5experimental conditions,which are cut-in,departure,constant acceleration,emergency braking conditions of the preceding vehicle,and curved road condition.The experimental results show that the ARIMA-MPC car-following algorithm proposed in this paper can effectively integrate the requirements of followability,comfort,safety,and fuel economy,and the relaxation factor can automatically adjust the upper and lower bounds of each constraint,which can handle the traditional MPC has no solution.