Risk Characteristics And Merging Zone Length Settings In Nighttime Maintenance Work Zones On Mountainous Highways

This essay conducts a study on the traffic flow characteristics and merging area risk features of highway maintenance work zones at night based on traffic observation data in order to clarify the driving risk characteristics of highway maintenance work zones at night and explore the impact of macroscopic and microscopic feature parameters and the length of the control section on the driving risk in merging areas.This study estimates the magnitude of operational zone hazards by combining XGBoost-SHAP theory with singlevehicle trajectory data,macroscopic traffic flow data,and a set of traffic flow simulations.The regression equation of trajectory parameters under merging conflict risk and the comprehensive risk index(ICRI)is derived.Logistic regression prediction models are developed utilizing micro-trajectory parameters with high feature importance.To determine the level of variation between the measured traffic volume and the simulated output traffic volume under various parameter settings,a one-factor analysis of variance of the VISSIM parameters is carried out.Additionally,the VISSIM parameters were optimized using BP neural networks in conjunction with a genetic algorithm.After parameter calibration,a simulated model of the maintenance work area is created.To determine the law of change in the merging risk of the control section length,the analysis is done from the perspectives of the severity of traffic flow risk and collision likelihood in various control regions of the maintenance work zone.The following are the key findings of the study:(1)The thesis describes the features of the nighttime traffic flow in highway maintenance work zones,the relationship between operating zone conflicts and microtrajectory factors,and the characteristics of merging conflicts.The research examines the risk law of nighttime maintenance zones from the viewpoints of traffic volume,traffic composition,speed distribution,and vehicle trajectory characteristics using data from drone observations.Additionally,the loss of collision energy index(LEI)and accident index are combined to examine the severity of conflicts under various circumstances and the conflict law under various lane-changing characteristics in the merging zone(CI).(2)The thesis also determines the importance and evolution trend of micro-trajectory parameters and macroscopic traffic flow parameters in affecting the risk characteristics of highway maintenance work zones at night.The research results show that among the micro-trajectory parameters,DRAC,THW,average speed,1/TTC,and travel time have a higher average impact on the model output,and DRAC is always the first output impact parameter.DRAC,average speed,and THW have a positive driving effect on risk.Among the macroscopic traffic flow parameters,traffic volume,heavy vehicle proportion,average driving time in the merging zone,and 5% 1/TTC are key factors affecting the overall risk of the maintenance zone.The prediction results can provide reference for the organization and control of traffic in highway maintenance zones at night.(3)The thesis constructs a Logistic risk regression prediction model for highway maintenance work zones at night and obtains the collision probability of vehicles in different control sections of the merging zone during maintenance construction and the overall risk level probability of the road conditions.Based on the important feature parameters of the prediction model,factors such as speed,acceleration,distance to the upstream starting point,vehicle length,driving time in the merging zone,and THW are selected as independent variables,and the comprehensive risk index(ICRI)is used as the dependent variable to analyze the significance of each factor under different risk levels.(4)The thesis derives the risk indicators RSL and CID and the mean risk probability under different control sections,and thoroughly studies the collision probability and safety corresponding to different combinations of warning zones and upstream transition zones of varying lengths.The paper reveals the law of variation of collision probability and various conflict risk probabilities of highway maintenance work zones at night with different control section lengths.The safety performance of different control section deployment schemes in the merging zone is evaluated using the fitted relationship equation between Ls+Lw and the comprehensive safety indicator I.When veh>1000pcu/h,Ls and Lw have the same sensitivity to the optimal control section combination.At this time,the service level of the merging zone decreases,and the improvement in road safety level caused by the increase in Ls and Lw is not significant.The research results have certain reference significance for improving the driving safety level of highway maintenance work zones at night.