The traditional calculation method can only determine the minimum length of the warning zone. In this paper, it proposes a method to determine the length of freeway warning zone using VISSIM simulation model and surrogate safety assessment model (SSAM). The simulation model of freeway maintenance work zones is constructed based on VISSIM, and calibrates the VISSIM simulation model. Through improving the time to collision (TTC) model, the traffic conflict discrimination threshold of the freeway maintenance work zone was determined. Based on the average absolute percentage error (MAPE) of traffic conflict, the validity of VISSIM simulation model and threshold are verified. Meanwhile, the SSAM is used to determine the changing trend of the number of traffic conflicts; a linear regression model is established to analyze the relationship between simulated and observed traffic conflicts. In addition, under different traffic conditions, the relationship between travel time, delay, traffic conflict and the warning zone length is analyzed. Based on traffic conflict number and safety evaluation index, the optimal length of warning zone is determined, and the relationship between warning zone length and safety evaluation index is studied. The results show that the simulated traffic conflict is reasonable consistent with the observed traffic conflict; under different traffic conditions, travel time and delay increase slowly with the increase of warning zone length; traffic conflicts decrease with the increase of warning zone length, when the warning zone length is more than 2200 m, traffic conflicts show a certain convergence state; When the warning zone length is 2200m, the number of traffic conflicts, delay and safety evaluation index are the minimum, and the safety is the best. Therefore, the method can determine the optimal length of the warning area and improve the safety of the warning zone.

At present, the total length of accident blackspot accounts for 0.25% of the total length of the road network, while the total number of accidents that occurred at accident black spots accounts for 25% of the total number of accidents on the road network. This paper describes a traffic accident black spot recognition model based on the adaptive kernel density estimation method combined with the road risk index. Using the traffic accident data of national and provincial trunk lines in Shanghai and ArcGIS software, the recognition results of black spots were compared with the recognition results of the accident frequency method and the kernel density estimation method, and the clustering degree of recognition results of adaptive kernel density estimation method were analyzed. The results show that: the accident prediction accuracy index values of the accident frequency method, kernel density estimation method, and traffic accident black spot recognition model were 14.39, 16.36, and 18.25, respectively, and the lengths of the traffic accident black spot sections were 184.68, 162.45, and 145.57, respectively, which means that the accident black spot section determined by the accident black spot recognition model was the shortest and the number of traffic accidents identified was the largest. Considering the safety improvement budget of 20% of the road length, the adaptive kernel density estimation method could identify about 69% of the traffic accidents, which was 1.13 times and 1.27 times that of the kernel density estimation method and the accident frequency method, respectively.

Due to the randomness and weak symmetry of traffic accidents occurring in the expressway maintenance operation area, it is difficult to use the number of traffic accidents to evaluate the safety of maintenance operation areas. In this paper, the traffic characteristics and traffic conflicts of the maintenance operation area with the lane closed on the outside of the two-way four-lane expressway are studied. By using the statistical method, the distribution of vehicle speed and time headway in different areas of the maintenance operation area are analyzed, and the queuing characteristics of vehicles in the upstream transition zone of the expressway are determined. Based on improved time to collision (TTC) model, the traffic conflict severity of expressway maintenance operation area is divided. The negative binomial distribution is used to establish a traffic conflict prediction model for the enclosed maintenance area of the outer lane of the expressway, and the validity of the traffic conflict prediction model is verified based on the average absolute error percentage (MAPE). The research results show that: when the 0 < TTC < 1.3 s, the traffic conflict is serious conflict; when 1.3 s < TTC, the traffic conflict is non-serious conflict. Furthermore, the traffic conflict prediction model has high accuracy, the MAPE of the warning area and the upstream transition area are 10.8% and 5.0%, respectively.

This paper aims to explore the effects of different cognitive loads on driver eye movement and ECG, and construct the BP (Back Propagation) neural network prediction model of driver takeover performance optimized by genetic algorithm (GA). In this paper, the simulation software UC-win/road was used to construct the highway driving scene, and the N-back tasks of different difficulty were selected to set different levels of cognitive load for testing. Using the driver eye movement data and ECG data collected during the test, combined with the NASA-TLX load scale collected after the driving simulation test, the subjective and objective data were analyzed. We determined the cognitive load level of drivers under different cognitive tasks based on the K-means clustering algorithm. We selected the significant objective indicators that affect the cognitive load of drivers, constructed a takeover performance prediction model based on BP neural network, and verified the effectiveness. Compared with the BP prediction model, the GA-BP prediction model established in this paper has different degrees of improvement in each evaluation index under different time window lengths. Among them, the improvement effect is the most obvious under the length of 10s time window, the accuracy rate is increased by 5.51%, the recall rate is increased by 7.08%, the accuracy rate is increased by 6.19%, and the F1 score is increased by 9.71%. The findings indicate that as the difficulty of the cognitive sub-task escalates, the driver's tension increases and the cognitive load increases. The GA-BP prediction model established in this paper has higher prediction accuracy.