Research On Risk Assessment Of Intelligent Vehicles At Intersection

With the development of modern transportation motorization,the number of motor vehicles in China continues to increase,and traffic accidents are occurring more and more frequently.As an important part of the road traffic system,intersections are also a high risk area for traffic accidents.At intersections,drivers are unable to accurately detect the risk of collision in the driving environment,and the lack of timely reaction and improper handling are the main causes of collisions.If the vehicle can accurately identify the risk of driving in the process and send out relevant alerts to the driver in advance,it can significantly reduce the occurrence of vehicle collisions and improve driving safety.The development of vehicle network enables vehicles to obtain real-time driving environment information,which gives technical support for risk warning.Thesis is based on the vehicle network environment,from the perspective of safe driving,to study the vehicle driving risk identification and risk warning method in the intersection traffic environment.(1)The key technologies required for intersection vehicle risk assessment are studied to obtain the vehicle information required for vehicle risk assessment.Firstly,a simplified vehicle motion model is built,and the extended Kalman filter is used to improve the accuracy of vehicle position estimation;secondly,the time domain and airspace characteristics of the intersection traffic conflict area are analyzed.Then,the model is optimized by combining the airspace conditions,which solves the problem of large discrepancy between the calculated collision risk and the actual risk perception in single time domain conditions.The results show that the simplified vehicle motion model and the vehicle collision model can more accurately represent the vehicle collision process in real situations.(2)To study the application of extreme value theory in collision risk assessment.First,analyze the problems of two indicators of vehicle conflict severity: intrusion time and collision time;second,use the threshold model to model the data distribution and use the payoff level method to calculate the vehicle collision probability;then,use the mixture distribution combined with extreme value theory to model the safety level,and also model the empirical distribution function of collision time at any time;finally,propose the data separation method for vehicle The data separation method is proposed to classify the interaction mode under the intersection and decompose to calculate the vehicle collision risk.The results show that the extreme value theorem combined with the data separation method can effectively improve the estimation accuracy of the estimator.(3)Research the risk alert system components and interactions,and analyze the risk alert system requirements and performance.First,according to the two indicators of vehicle driving safety and interference to drivers,consider the possible utility of different risk alert programs,and use the interval number method to specify the program results;then,on the basis of the study of utility theory,establish the utility equation for different driving behaviors of vehicles,combine with the optimal utility principle,and determine the alarm threshold of the risk alert system.Finally,a differentiated alarm scheme is adopted according to different driving characteristics of drivers.(4)A simulation system is designed and built to conduct multiple sets of experiments for the intersection traffic scenario.First,the data separation method proposed in this study is verified to improve the estimator accuracy;second,it is verified that the risk alert system can significantly reduce the occurrence of traffic accidents in the intersection scenario;then,the experiments are compared with the collision time based warning system,and the results show that the average false alarm rate of the risk alert system can be reduced by up to 28%;Finally,thesis analyzed the impact of the risk warning system on reducing the severity of collisions,and the experimental results showed that the risk warning system can reduce the average speed of vehicles at the time of collisions to a certain extent,and reduce the loss of collisions.