| With the development of computer technology,autonomous vehicles have gradually become an important role in the transportation system.Mixed autonomous vehicles and conventional vehicles traffic will be a common traffic pattern in the future.VISSIM and SSAM simulation software were used to build a simulation environment for highway mixed autonomous vehicles and conventional vehicles traffic in this dissertation.The traffic conflict risk mechanism of the special traffic environment was analyzed by TCT.Combined with the derived conflict track data,a mixed traffic conflict prediction model was constructed.By analyzing the occurrence of mixed traffic conflicts,the identification model of mixed traffic conflicts and the conflict risk degree model of mixed traffic were established,and the conflict risk degree of vehicles with different proportions of autonomous vehicles were quantitatively analyzed.Specific research contents were as follows:(1)Firstly,by comparing the research on micro-driving characteristics of autonomous vehicles,driving response time,critical safety distance and driving randomness were selected as indicators to quantitatively compare the differences of micro-driving characteristics between conventional vehicles and autonomous vehicles.Then based on the assumption of ideal conditions and Co EXist research results of the project,scene of the mixed traffic in highway was realized in VISSIM.Finally,the distribution of traffic volume and headway were selected to test the validity of the established model.(2)SSAM was used to calculate the trajectory data in VISSIM to obtain the traffic conflict data during the simulation period.Lane changing times,mixed traffic volume,proportion of autonomous vehicles and expected speed were selected to establish the number of traffic conflicts relationship model.The results show that the number of traffic conflicts had a certain linear positive correlation with lane changing times,mixed traffic volume and expected speed,and the total number of traffic conflicts will increase first and then decrease with the increase of the proportion of autonomous vehicles.When the proportion of autonomous vehicles were 20%-30%,the maximum value of the number of conflicts was taken.Among the four parameters,the linear fitting degree of the total number of conflicts and the number of lane changes was the highest.(3)The conflict prediction model of highway mixed traffic is established which is based on BP neural network.Lane changing times,mixed traffic volume,proportion of autonomous vehicles and expected speed were used as input of the model,and the output was the number of traffic conflicts.640 sets of data were screened as training samples and 10 sets of data as test samples.Through several validations,the total number of conflicts,rear-end conflicts and lane-changing conflicts were predicted by BP neural network,and the results of the relative error E_iless than 30%,the determination coefficient R~2were 96.069%,92.363%,93.025%.It was shown that the BP neural network prediction model had good performance and can predict the traffic conflicts under the condition of mixed autonomous vehicles and conventional vehicles traffic in highway.(4)According to the behavior of following and lane change in highway,the identification model of mixed traffic conflict,the probability model of mixed traffic conflict,the model of conflict severity and the risk model of mixed traffic conflict were established.Through the simulation example,the overall conflict risk of the road section under different proportion of autonomous driving vehicles were calculated.The results show that:under the same TTC condition,the probability of conflict accident of self-driving vehicles were much lower than that of traditional vehicles.The overall accident risk of road section increased first and then decreased.When the proportions of AV were 20%～30%,the total risk degree of conflict of mixed traffic got the maximum value. |
PDF Full Text Request