Rollover State Prediction Of Vehicle Based On Driver Preview And Path Prediction

All the time,vehicle rollover accidents have occurred frequently,causing serious casualties and property damage.Among them,non-tripping high-speed steering rollover accidents account for a large proportion.In such accidents,rollover risk prediction can effectively reduce rollover accidents.Data show that a one-second warning can reduce accidents by 60-90%.Therefore,conducting research on rollover risk prediction will have important theoretical and engineering application value.Predicting the risk of vehicle rollover earlier is the key to further improving vehicle safety performance in assisted driving systems,but it is necessary to solve the problem of unknown prior information of "road" in traditional "human-vehicle-road" coupling systems and the resulting difficulty in predicting "human" input.The project proposes a roll state prediction method that combines the driver’s steering model and the vehicle’s nonlinear dynamics model,utilizing the forward preview path information provided by the advanced driving assistance system.Firstly,the problem of steering input prediction in the "human-vehicle-road" closed-loop feedback system is solved,and then the current vehicle speed and predicted steering are used as inputs to the roll dynamics model to achieve roll state prediction following the forward path.The main work of this dissertation is as follows:(1)Considering tire load transfer,tire nonlinear characteristics and tire force coupling characteristics during steering,a three-degree-of-freedom nonlinear rollover dynamics model was established including vehicle yaw motion,lateral motion and body roll motion,laying the foundation for rollover research.(2)Establish a directional control driver model that can reflect the characteristics of the driver,convert the following path information into steering wheel angle input,and validate the established driver model using dual lane change and serpentine conditions.(3)By using the method of lane line detection,the preview path of the driver’s visible area is obtained.By obtaining the historical waypoint and pre-viewing waypoint of the vehicle,the quintic polynomial fitting is used to obtain the longer distance forecast path of the vehicle.(4)By conducting in-depth analysis of vehicle rollover prediction,a long-term and short-term rollover prediction system is established,forming a short-term rollover state prediction system based on the current vehicle state and a long-term rollover state prediction system based on path preview prediction following.