Research On Lane-changing Safety Model Under The Condition Of Driving Risk Field Improved By Driving Style

The rapid development of highway in China has provided support for economic development and convenience for the people,but also caused a large number of traffic accidents.The current highway driving safety situation is in urgent need of improvement.Research on driving risk is an important basis for accident prevention measures.Drivers are considered the most important factor affecting driving risks,with 94% of traffic accidents caused by drivers.Among them,the driver’s driving style is closely related to the accident.At the same time,as one of the most common driving behaviors,accounting for 27% of traffic accidents,the research on lane change safety has gradually become the focus of scholars.Therefore,drivers’ driving style should be introduced into the study of driving risk to explore a risk assessment method that takes into account various factors and realize dynamic risk quantification in complex driving environments,which is of great significance for the study of highway lane changing safety.In view of this,this paper firstly classified and identified the driver’s driving style,then introduced driving style to improve the driving risk field.A driving risk field model was established which takes into account driving mode and vehicle driving state.Finally,the driving risk field model was introduced into the study of lane change safety model,and a lane change safety model considering driving mode was established.The main research efforts are as follows:(1)Clustering analysis of driving style based on trajectory data.Based on the high D trajectory data extraction and driving style characteristic parameter calculation,the Kmeans algorithm and K-means++ algorithm are used to cluster analysis of driving style after principal component analysis and dimensionality reduction.Finally,by using contour coefficients to compare and evaluate the clustering effect,the results show that both clustering methods have good effect.(2)Modeling of driving risk field considering driving style.After improving the static risk field of the vehicle,the vehicle motion state was corrected to obtain the vehicle kinetic energy field model;Define driving style factors to quantify the driving style of drivers;By combining the driving style factor with the vehicle kinetic energy field,a driving risk field model considering driving style was obtained.The model takes into account driving style,speed,acceleration and steering angle,and can predict and evaluate dynamic driving risk.A vehicle tracking model based on driving risk field is derived to calibrate the model parameters and particle swarm optimization used to calibrate the model parameters.(3)Based on the improved driving risk field,a lane change safety model considering driving style is developed.The proposed driving risk field model was introduced in the construction of lane change safety model,and driver safety sensitivity put forward to reflect the different needs of different driving styles for lane change distance.Finally,a lane change safety model considering driving mode was established.Numerical simulation verification analysis of the model is carried out and the results showed that the model can characterize the influence of speed,acceleration and driving mode on the minimum safe distance between lane changing vehicles and interactive vehicles.(4)Application test of the lane change safety model in lane change warning.The proposed safety model was applied to the early warning test.Sample data based on trajectory dataset was extracted to satisfy the setting of the study and compared with lane change safety model based on elliptical vehicle model.The warning results showed that compared with the former,the lane change safety model proposed in this paper showed better performance in lane change warning testing,with an accuracy rate of 89.0%,a false alarm rate of 9.9%,and a missed alarm rate of 1.1%,This indicates that the model is relatively sensitive to identifying lane changing risks and rarely misses dangerous points,which is beneficial for ensuring lane changing safety.The results of this study can help the driving system to understand human driving style and provide reference for dynamic risk assessment and lane change safety in human-machine shared driving environment.