Study Of Vehicle Active Steering Control Based On Driver Model

The rapid developments of automobile industry and highway are resulting in increasing vehicle speed and higher density of traffic in roads.While many ordinary drivers' skills are insufficient which may lead to high rate of traffic accidents.In recent years,as one of key technologies for active safety control,active steering control has become an effective measure to improve vehicle handling performance and solve the problem of losing stability in critical operation conditions.Active steering controller can modify steering angle input,being independent of the steering wheel input by driver,to improve the vehicle handling stability.However,most previous and existing active steering systems have been designed mainly based on the stability of vehicle itself but do not consider the individual difference of driver input,e.g.driving styles and skills.Hence they can not provide better assistance for the driver to control the vehicle.Therefore,in this dissertation,an active steering system which combines a driver model is creatively proposed.The driver factor and external disturbances can be sufficiently considered by the proposed controller,and based on an optimal desired path the coupling interaction between driver and active steering system is thoroughly studied.The active steering system controller is designed based on differential game theory,being able to effectively trace vehicle path in different conditions for ensuring vehicle safety.First,on the basis of preview compensation optimization neural network driver model which includes the states of vehicle and feedback information of driver,an active steering controller is designed based on Generalized Predictive Control(GPC)to guarantee vehicle tracking capability in the presence of external disturbances.Considering the complexity and uncertainty of vehicle parameters,the Controlled Auto-Regressive Integrated Moving-Average(CARIMA)model is used as vehicle predictive model.By electing path centerline as desired trajectory,and respectively taking lateral tracking error and corrected steering wheel angle as input and output,the steering wheel angle from driver is corrected,so the modified steering wheel angle is finally obtained.Second,in order to further expand the application of the active steering system to different trajectories and to solve the design problem for desired trajectory for active steering system,a path optimization method based on both vehicle kinematics and dynamics is proposed.Similar to racing car dynamics,this method can learn designing experience within shortest lap time.Taking minimum maneuver time and actuator workload as cost function,and also considering the constraints of vehicle limiting motion states and road boundary,the following optimal trajectory for active steering system is obtained by transforming the solution of optimal trajectory to a nonlinear constrained optimization problem.This thus also provides the conditions for subsequent driver-vehicle interaction steering control.In particular,by introducing curve coordinate system,the time-based vehicle dynamics system can be transformed into spatial dynamics system,which can effectively simplify the calculation and avoid the problem of complex coordination overlapping resulted from twisted path.Finally,with the steering wheel angle inputs from both driver and active steering system,a collaborated control structure is established.Considering the different intervention extents of active steering system,and based on different driving skills identified by critical driver model parameters,the steering wheel angle of active steering controller is calculated through differential game theory.With different intervention extents,the active steering system can ensure that drivers with different skill levels have a better path tracking ability and also can avoid the active steering system applying unnecessary intervention to driver steering input,which may promote making driver uncomfortable and loss driving sense.