Modeling Of Vehicle Collision Risk And Optimization Of Lane-Changing Strategy At Highway Diversion Area

At present,autonomous driving technology is developing rapidly.With the promotion of national policies,the market size of vehicles equipped with autonomous driving technology will continue to increase in the future,which will bring the problem of intelligent vehicle control and management,especially on highways.This area of vehicle interweaving,how to control the intelligent vehicle to change lanes safely is a very critical issue.This article studies the risk of lane change collision in the highway diversion area,aims at improving the safety of vehicle lane change in the traffic diversion area as a goal,and provides a reference for the control and management of intelligent vehicles in China.First,the vehicle lane changing behavior is studied.By studying the motion trajectory data of the vehicle before the lane change and the motion trajectory data of the surrounding vehicles,it is found that there are precursor characteristics of the vehicle lane change.Using the motion data before the vehicle lane change,the mixed Gauss-Hidden Markov model is used for parameter training.The verification results show that the hybrid Gaussian-Hidden Markov recognition model can achieve higher recognition accuracy,and the accuracy of the left and right lane change behavior recognition is respectively 91.4% and 92.4%,the accuracy of lane keeping behavior recognition reached 85.8%.Secondly,the risk of lane change collision is studied.Through the study of the impact factors of collision risk,it is found that collision risk is mainly composed of persistent factors and serious factors.Using the fault tree theory FTA to fuse the two indices,the real-time lanechange collision risk index LCRI is obtained.The actual data was used to verify the index,and the results show that the real-time lane-change collision risk index LCRI can accurately describe the changing trend of vehicle lane-change collision risk.In the end,the dynamic optimization of vehicle lane changing strategy is studied.Based on the research of vehicle lane change collision risk,a collision risk domain is constructed,with the goal of minimizing collision risk,and using rolling optimization to solve and control the optimization problem.The actual control data was used to verify the optimization control model.The results show that the lane change strategy dynamic optimization model can significantly reduce the risk of vehicle lane change collisions.The average percentage of collision risk reduction during vehicle lane change is about 56.5%.The double optimization effect reached79.3%.