Research On The Longitudinal Driver Assistance System Based On Hybrid Model Predictive Control

Along with continuously development of social economy nowadays,the rise of automotive industry has promoted the process of motorization and urbanization.The problems of traffic safety and jams,environmental pollution and energy shortages should not be overlooked.Taking human as the core of the research,the longitudinal driver assistance system based on intelligent transportation is devoted to reducing the factitious accidents and gaining wide public concern.However,in pursuit of the vehicle safety performance,the individual differences and state fluctuation of driver characteristics result in the coordinated interaction between human and system.So,with the proposed concept of intelligent environmental-friendly vehicle,it has become popular to give consideration to both fuel economy and riding comfort of the longitudinal driver assistance system.This paper combined with the National Natural Science Foundation of China," Control Mechanism and Evaluation Method of New Type of Steering System Based on Driver Characteristics "(No.51575223).Based on the research of situation on driver car-following behavior and driver assistance controlling at home and board,this paper summarizes the deficiencies existing in the current control algorithm of longitudinal driver assistance system.So as to fulfill the demand of different drivers on car-following situation,meanwhile the goal of intelligent control which is friendly towards traffic environment,the driver model based on hybrid system is established.Then with the goal of coordinated performances of the tracking,safety,comfort and fuel-economy,the multi-objective control algorithm based on the framework of Model Predictive Control(MPC)method is constructed.Finally,the verification on the driving simulator platform is carried out.This paper is organized as follows:(1)Establishing the Generalized Vehicular Longitudinal Dynamics(GVLD)including inverse dynamics model,switch logic between braking and driving and the CarSim software model.The veracity of GVLD system has been verified.Then the inter vehicle longitudinal dynamic system under the situation of car following is built based on the desired vehicle spacing strategy,which lays the foundation of system function achievement.(2)Concerning the adaptation of driver characteristics on car following to longitudinal driver assistance system,the Piece Wise Affine system(PWA)is adopted to mimic driver's expertise based on the concept of hybrid system.The high-dimensional data space clustering analysis method is given to deal with the dynamic car-following behavioral data collected through the driving simulator.Therefore,the identification of driver car-following characteristics is realized based on the Piece Wise Auto Regressive Exogenous(PWARX)model.(3)In order to realize the control requirement of a new generation of multi-target coordinated longitudinal driver assistance,the performance of tracking,fuel-economy,safety and comfort is quantized by the form of 2-norm after the analysis of the overall architecture on control strategy.Considering the solvability problem of optimal control variable,the constraint management method is used to soften the hard constraints,thus the MPC optimization problem can be achieved which lays the foundation of the coordinated control of tracking,comfort and fuel-economy.(4)For the verification of the proposed multi-target coordinated longitudinal driver assistance system based on hybrid system,an Adaptive Cruise Control system based on Linear Quadratic Regulator(LQR)is adopted as the experimental control group.The experiments are carried out under the conditions of cut-in,departure,acceleration and braking of the front vehicle.The results show that the proposed algorithm can not only reduce effectively fuel consumption on the premise of maintaining necessary tracking capacity,but also satisfy the desired driver characteristics of riding comfort and dynamic car-following performance.