Research On Vehicle Longitudinal Driving Assistance System Based On Driver Characteristics

As one of the research hotspots in the field of automated driving,vehicle longitudinal driving assistant system has the functions of auto-following driving,conforming to driver's behavior characteristics and low fuel consumption,which are of great significance in improving driving safety,driving comfort and fuel economy.However,most of the existing studies focus on the realization of single function,and tend to focus on the following and safety of the system,but ignore the restrictive relationship among them.A few studies have been carried out on the multi-functional longitudinal driving assistance system,considering comfort,fuel economy and safety,but not fully considering the differences of drivers' behavior characteristics and treating different types of drivers individually.Because the driver has strong complexity and variability in the closed-loop system of man-vehicle-road,it is difficult to meet the personalized requirements of the whole driver group if only considering the behavior characteristics of the general type of driver.Therefore,in the research of longitudinal driving assistance system,for different types of drivers,personalized control strategy of longitudinal driving assistance system is designed,and fuel economy and safety are regarded as the control objectives of longitudinal driving assistance system,which is the focus of personalized intelligent driving assistance system research at this stage.Based on the project of National Natural Science Foundation of China(No.51575223),a new type of steering-by-wire system control mechanism and evaluation method based on driver's characteristics is studied in this paper.The unsupervised clustering algorithm is used to effectively classify the driver's behavior characteristics,and the identification model of driver's behavior characteristics is established based on artificial neural network to accurately identify the driver's type.According to the identification results,the corresponding control strategy parameters of the longitudinal driving assistance system are matched to improve the driver's acceptance.On this basis,the driver's comfort,follow-up and the driver's acceptance are synthetically improved.The multi-objective coordinated control strategy based on model predictive control is designed to improve the comfort and safety of the system,and at the same time improve the fuel economy of the system.Finally,a multi-condition experiment is designed to verify the control algorithm.The main contents of this paper are as follows:(1)Dynamic Model of Vehicle Longitudinal SystemAs the basis of realizing the control function of the longitudinal driving assistant system,in this paper establishes a vehicle longitudinal dynamic system model based on MATLAB/Simulink,including the vehicle inverse longitudinal dynamic model,the driving and braking switching control strategies and the Car Sim vehicle model.It is proved that the model can better simulate the response characteristics of vehicle longitudinal dynamics and serve as the subsequent control strategy.The development lays the foundation.(2)Analysis and Identification of Driver Behavior CharacteristicsIn order to study the driver's behavior characteristics and establish a behavior identification model,28 drivers are selected to collect the data of car-following experiments and extract the characteristic parameters reflecting the driver's behavior.Then,based on unsupervised Kmeans algorithm,the characteristic parameters are clustered and the driver types are classified effectively.Finally,the most widely used artificial neural network is used.The identification model of driver's behavior characteristics is established based on BP neural network,and the validity of the model is verified.(3)Research on Control Strategy of Personalized Longitudinal Driving Assistance SystemBased on the analysis of the data of different types of driver's following behavior,the following distance model of fixed workshop time distance is established by choosing the parameters that can reflect the driver's behavior characteristics,and the longitudinal acceleration parameters of different types of drivers are taken into account in the control strategy of longitudinal driving assistant system to realize different control strategies of three types of drivers.A model predictive control(MPC)based multi-objective vehicle longitudinal control algorithm is designed considering fuel economy and follow up.(4)System testing and analysisCar Sim and MATLAB/Simulink joint simulation platform are built,cruise mode,follow mode and front vehicle cycle mode are designed to simulate and validate,and compared with LQR-based longitudinal driving assistance system.Finally,semi-real vehicle experiments are carried out using simulator.Experiments show that the control strategy of the longitudinal driving assistance system in this paper can meet the behavior characteristics of different types of drivers,improve the comfort of driving,and meet the requirements of safety,follow-up and economy.