Intelligent network connection mixed driving mountain city expressway confluence area safety evaluation method

Abstract

The invention discloses a safety evaluation method for an intelligent network connection mixed driving mountain city expressway confluence area. The method comprises the following steps: constructing a simulation scene of the mountain city expressway confluence area; confluence behaviors in a mixed driving scene are simulated, and rear-end collision and lane change collision data are extracted; constructing a traffic conflict prediction model under different intelligent network connection permeability, and outputting a traffic conflict number prediction value through the traffic conflict prediction model; the traffic conflict rate is used as an evaluation index to construct a safety evaluation index system, the comprehensive weight of the evaluation index is calculated, and the safety level is divided. By constructing the mountain city expressway confluence area traffic conflict prediction model and the conflict rate evaluation index system and comprehensively integrating the empowerment and improving the gray clustering model, the method accurately reflects the traffic safety condition, adapts to the intelligent network connection mixed driving environment, provides a scientific basis for traffic optimization, effectively improves the safety level, reduces accidents and improves the traffic safety. And resident travel safety is guaranteed.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent transportation systems, and in particular to a safety evaluation method for a merging area of an intelligent connected mixed-driving mountainous urban expressway. Background Art

[0002] Expressways, the backbone of mountainous urban road networks, not only connect communities to meet residents' travel needs but also serve as the primary gateway for external traffic. These roads are characterized by steep slopes, complex interchanges, and large nonlinear coefficients. Merging areas, connecting expressways with other roads, are prone to traffic accidents due to frequent lane changes. With the advancement of information technology, intelligent connected vehicles (ICVs), featuring real-time information sharing and intelligent decision-making, can effectively reduce traffic conflicts and improve traffic safety at merging areas on mountainous urban expressways. However, mixed traffic, where ICVs and human-driven vehicles share road rights, will likely persist in the future, creating new traffic safety issues. Therefore, studying the relationship between various factors and traffic conflicts at varying ICV penetration rates is crucial for road safety assessment.

[0003] Existing research primarily focuses on mixed-driving traffic flow stability, capacity, vehicle routing, environmental and economic benefits, and ethical and legal considerations. Relatively little research has been conducted on the impact of intelligent connected vehicles on road safety in mixed-driving environments. Most research focuses on merging areas on plain urban expressways, with little research on merging areas on mountainous urban expressways. To address these issues, a safety assessment method for merging areas on mountainous urban expressways in mixed-driving environments is urgently needed. By combining microsimulation technology with an improved grey clustering model, we can accurately analyze the factors influencing traffic conflicts and assess safety levels under varying intelligent connected vehicle penetration rates. Summary of the Invention

[0004] The purpose of the present invention is to provide a safety assessment method for merging areas on intelligent connected mixed-driving mountainous urban expressways to address the problems raised in the above background technology. To achieve the above purpose, the present invention provides the following technical solutions:

[0005] A safety evaluation method for merging areas on mountainous urban expressways with mixed intelligent connected vehicles comprises the following steps: constructing a simulation scenario for the merging area on mountainous urban expressways, including parameters such as intelligent connected vehicle penetration rate, mainline traffic volume, and entrance flow ratio; simulating merging behavior in the mixed driving scenario using microscopic traffic simulation technology, and extracting rear-end collision and lane change conflict data in combination with traffic conflict technology; constructing a traffic conflict prediction model under different intelligent connected vehicle penetration rates based on a negative binomial regression model, and outputting a predicted value of the number of traffic conflicts through the traffic conflict prediction model; constructing a safety evaluation index system using the traffic conflict rate as an evaluation index, and calculating the comprehensive weight of the evaluation index using an improved grey clustering model comprehensive entropy method and a DEMATEL method to divide the safety levels.

[0006] Preferably, the construction of a simulation scenario for a merging area of a mountainous urban expressway includes the following steps: selecting a typical merging area of a mountainous urban expressway, numbering the lanes and dividing the sections according to the functions of the sections; collecting traffic operation data, traffic conflict data and road geometry parameters of the merging area of the mountainous urban expressway; performing traffic operation analysis based on the collected data, and drawing a road network for the merging area of the mountainous urban expressway; dividing the simulation scenarios according to different penetration rates of intelligent networking into mixed driving scenarios based on manual driving and mixed driving scenarios based on intelligent networking; setting the main line traffic volume and entrance flow ratio, and simulating the merging behavior under different traffic loads; comparing the output data with the measured data, adjusting the model parameters, and verifying the credibility of the model.

[0007] Preferably, the traffic operation data includes data such as traffic flow, vehicle speed, and vehicle type; the traffic conflict data includes the number and severity of rear-end conflicts and lane change conflicts; and the road geometry parameters include data such as slope, lane width, and acceleration lane length.

[0008] Preferably, the simulation scenarios are divided according to different intelligent network penetration rates into mixed driving scenarios dominated by manual driving and mixed driving scenarios dominated by intelligent network, including the following steps: selection, parameter calibration and verification of the following and lane changing models of manually driven vehicles; selection, parameter calibration and verification of the following and lane changing models of intelligent networked vehicles; determination of the intelligent network penetration rate division standard, and division of the mixed driving scenarios into two types: manually driven and intelligent network.

[0009] Preferably, the comparison of the output data with the measured data, the adjustment of the model parameters, and the verification of the model credibility include the following steps: determining the correction indicators, including the average travel time, the average speed, the number of lane changes, the number of rear-end collisions, and the number of lane change conflicts; calculating the relative error between the simulation data and the actual data by the relative error analysis method, and if the error exceeds 5%, adjusting the model parameter settings until the relative errors of all correction indicators are less than 5%; performing parameter calibration based on the default parameters and value ranges of the following and lane changing models, and inputting the calibration parameter values and the actual traffic flow characteristic values in the simulation scenario, running the simulation to obtain the correction index values, and if the relative error is still greater than 5%, continuing to modify the parameter values and repeating the simulation until the error requirements are met; verifying the validity of the calibration results by comparing the relative errors of the simulation values and the measured values, and completing the model credibility verification. The relative error calculation formula is as follows: Where a i Indicates the current status data of the i-th parameter; b i Represents the simulation data of the i-th parameter.

[0010] Preferably, the simulation simulates the merging behavior in the mixed driving scenario and extracts rear-end collision and lane change conflict data, including the following steps: based on the road network file and traffic demand file of the mountain city expressway merging area simulation scenario, setting the simulation duration, starting the simulation and outputting the vehicle trajectory file, recording the real-time position, speed and acceleration data of all vehicles; converting the FCD file into an SSAM-compatible trajectory file, retaining the vehicle ID, coordinates, speed, acceleration and lane information; setting the rear-end collision TTC threshold, lane change conflict PET threshold and conflict angle range detection parameters, running the SSAM trajectory file, automatically identifying and classifying rear-end collision and lane change conflict events, and counting the number of general rear-end collisions, severe rear-end collisions, general lane change conflicts and severe lane change conflicts per hour; comparing the conflict data output by the simulation with the historical accident data, calculating the absolute error and relative error, and if the error exceeds the limit, recalibrating the following model parameters or adjusting the conflict threshold.

[0011] Preferably, the construction of a traffic conflict prediction model under different intelligent network penetration rates and outputting a traffic conflict number prediction value through the traffic conflict prediction model include the following steps: dividing the extracted rear-end collision and lane change conflict data into a mixed driving scenario data set dominated by manual driving and a mixed driving scenario data set dominated by intelligent network; extracting the number of rear-end collisions, the number of lane change conflicts and independent variable data from the two types of data sets respectively, and screening the independent variables significantly correlated with the number of conflicts as model input; constructing a negative binomial regression model, with the number of conflicts as the dependent variable, and the independent variables are input into the negative binomial regression model after logarithmic transformation, and then The regression coefficients were solved using the maximum likelihood estimation method, and traffic conflict prediction models were established for the two types of mixed driving scenarios. The Akaike Information Criterion and the Bayesian Information Criterion were used to evaluate the goodness of fit of the traffic conflict prediction models. An independent simulation scheme was designed, and the predicted number of conflicts was compared with the simulation results. The absolute error and relative error were calculated. If the error exceeded the limit, the traffic conflict prediction model parameters were recalibrated or the independent variable combination was optimized. The traffic parameters of the target merging area were input, and the traffic conflict prediction model output the predicted value of the number of traffic conflicts, including the number of general rear-end collisions, the number of severe rear-end collisions, the number of general lane change conflicts, and the number of severe lane change conflicts.

[0012] Preferably, the traffic conflict rate includes a general rear-end collision rate, a severe rear-end collision rate, a general lane change conflict rate, and a severe lane change conflict rate. The conflict rate calculation formula is: Where Zi represents the number of traffic conflicts occurring within an hour, i = 1, 2, 3, and 4 represent general rear-end collisions, severe rear-end collisions, general lane-changing conflicts, and severe lane-changing conflicts, respectively; Q represents the traffic volume within an hour.

[0013] Preferably, the calculation of the comprehensive weight of the evaluation index comprises the following steps:

[0014] The objective weight is determined by using the entropy method: the traffic conflict rate data is normalized using the max-min normalization method; the index value proportion of the i-th option under the K-th index is calculated using the following formula: And ∑λ ik =1; calculate the information entropy of each indicator, the formula is Calculate the objective weight of each indicator, the formula is

[0015] Use the DEMATEL method to determine the subjective weight: construct the direct impact matrix between the evaluation indicators, quantify the logical relationship between the indicators, calculate the comprehensive impact matrix, and obtain the influence degree D of each indicator. i , influence degree C i , centrality M i and causal degree R i , the subjective weight of each indicator is determined according to the centrality and causality, and the calculation formula is as follows:

[0016]

[0017] M i =D i +C i

[0018] R i =D i -C i

[0019]

[0020] In the formula, tij represents the degree of direct impact plus indirect impact of indicator i on indicator j, that is, the degree of comprehensive impact;

[0021] The comprehensive weight of the evaluation indicators is calculated using the comprehensive integrated weighting method based on game theory: by calculating the objective weight and subjective weight of each evaluation indicator, the weight vector β1 is obtained T and β2 T ; Linearly combine the subjective and objective weights, the formula is β=a1β1 T +a2β2 T , where a1 and a2 are subjective and objective weight coefficients (a1+a2=1, a1>0, a2>0); based on the principle of game theory, to minimize the combined weight vector β and the subjective and objective weight vector β1 T and β2 T The deviation of is the goal, and the overall optimal countermeasure model is obtained. The formula is min||β-β1 T -β2 T ||2=min||a1β1 T +a2β2 T -β1 T -β2 T ||2, by solving the optimization problem, we can obtain the first-order optimality condition According to the optimality condition, the subjective and objective weight coefficients a1 and a2 of each evaluation index are obtained to obtain the comprehensive weight vector β, and the comprehensive weight p* of different traffic conflict rate evaluation indicators is calculated.

[0022] Preferably, the safety level division includes: using the cumulative frequency curve method to divide the traffic conflict rate into four safety levels, with level one being the safest and level four being the most dangerous; introducing a sine function to improve the whitening weight function of the grey fixed-weight clustering model, calculating the clustering coefficient of each simulation scheme under the four safety levels, and constructing a clustering estimation value matrix, and selecting the level corresponding to the maximum clustering coefficient as the final safety evaluation result.

[0023] The beneficial effects of the present invention are: by constructing a traffic conflict prediction model for the merging area of mountain urban expressways, a conflict rate evaluation index system, a comprehensive integrated weighting and an improved gray clustering model, the present invention accurately reflects the traffic safety situation, adapts to the intelligent network mixed driving environment, provides a scientific basis for traffic optimization, effectively improves the safety level, reduces accidents, and ensures the safety of residents' travel. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0025] Figure 1 This is a flowchart of the safety evaluation method for the intelligent connected mixed driving mountain urban expressway merging area of the present invention.

[0026] It should be noted that the drawings are not necessarily drawn to scale, but are merely shown in a schematic manner that does not affect the reader's understanding. DETAILED DESCRIPTION

[0027] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0028] It should be further understood that the term "and / or" used in the present description and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0029] like Figure 1 As shown, the safety assessment method for the intelligent connected mixed driving mountain urban expressway merging area based on traffic conflict in the embodiment of the present application is specifically implemented as follows:

[0030] (1) Step S1. Construct a simulation scene of a merging area of a mountainous urban expressway.

[0031] S1.1. The Chongqing Inner Ring Expressway-Yanggong Bridge Interchange merge area was selected as the survey area. To facilitate subsequent accurate analysis of traffic parameters such as flow, speed, and lane-changing behavior, lanes were numbered and divided into sections. Based on the physical layout of the mainline lanes and on-ramps, the lanes were numbered from the inside outward in the direction of travel. The mainline lanes were numbered L1, L2, L3, and L4, and the on-ramps were numbered L5, L6, and L7. Each lane was 4 meters wide, and the total length of the survey section was 160 meters. Based on the different functions of the road section, the survey area was divided into five areas (A1-A5) along the direction of travel. The lengths of each area were 40 meters, 30 meters, 30 meters, 30 meters, and 30 meters, respectively. The specific functional divisions are as follows:

[0032] A1: Merge area, the main function is to allow vehicles to merge from the entrance ramp lane into the main lane;

[0033] A2: Buffer zone, used for vehicles to adjust speed and distance after merging into the area;

[0034] A3: Acceleration zone, used for vehicles to accelerate to match the speed of mainline traffic;

[0035] A4: Buffer zone, further adjusting vehicle speed and distance;

[0036] A5: Stable area, the vehicle travels at a stable speed in this area.

[0037] S1.2. Collecting Traffic Operation Data, Traffic Conflict Data, and Road Geometry Parameters at Merging Areas of Mountain Urban Expressways: Using a DJI Mini 2SE drone, we collected traffic operation data at merging areas of mountain urban expressways. This data included traffic flow, vehicle type ratios, and traffic conflict data, including mainline traffic volume, entrance flow ratio, and the proportion of large vehicles. Road geometry parameters, including slope, lane width, and acceleration lane length, were obtained through manual on-site observation and topographic map measurements. Tracker software was used to process this data to obtain traffic operation and conflict parameters, including the number and severity of tailback and lane-change conflicts.

[0038] S1.3. Conduct traffic operation analysis based on the collected data and use SUMO software to map the mountain urban expressway merging area network.

[0039] S1.4. Construct models for following and changing lanes for both manually driven vehicles and connected vehicles. Divide the simulation scenarios into mixed driving scenarios dominated by manual driving and mixed driving scenarios dominated by connected vehicles, based on the penetration rate of connected vehicles. The specific implementation process is as follows:

[0040] S1.4.1. Selection and parameter calibration of following and lane-changing models for manually driven vehicles: For manually driven vehicles, the IDM model is used to simulate the following behavior of vehicles on expressways, including acceleration and deceleration, and headway adjustment. The LC2013 model is used to simulate the lane-changing behavior of vehicles due to speed differences or path requirements. The average travel time, average speed, number of lane changes, number of rear-end collisions, and number of lane-changing conflicts are selected as correction indicators for parameter calibration. Parameter calibration is performed based on the value range of the driving behavior parameters of the following model and lane-changing model in the SUMO software. The relative error between the simulation data and the current data of each correction indicator is calculated. If the relative error exceeds 5%, the simulation parameter settings are adjusted until the relative error of each correction indicator is less than 5%, and the parameter calibration is completed.

[0041] S1.4.2. Validation of the manually driven vehicle following and lane-changing model: Construct a mountain urban expressway merging area network and traffic demand file in the SUMO software. Input the calibrated parameters, run the simulation through the SUMO-GUI, output the vehicle trajectory file, and use the SSAM software to extract traffic conflict data (such as TTC and PET thresholds). By comparing the simulation results with the measured data, verify the consistency of the model with actual traffic behavior.

[0042] S1.4.3. Model selection and parameter setting for following and lane changing of intelligent connected vehicles: For intelligent connected vehicles, the CACC model is adopted to support real-time information sharing and collaborative control between vehicles. The LC2013 model is used as the lane changing model, but the collaborative decision-making capabilities of intelligent connected vehicles are reflected through parameter adjustment. With reference to the CoEXist project and literature data, the CACC model parameters are set, including the minimum headway and headway time. In view of the characteristics of intelligent connected vehicles with short reaction time and stable headway, the acceleration and emergency braking parameters are adjusted to complete the setting of relevant parameters.

[0043] S1.4.4. Verification of the intelligent connected vehicle following and lane-changing model: Set the intelligent connected vehicle penetration rate (e.g., 40%, 80%) in the SUMO software, simulate the traffic flow characteristics under different mixed driving ratios, analyze the performance of intelligent connected vehicles in rear-end collisions and lane-changing conflicts through SSAM, and verify the effectiveness of the model in reducing traffic conflicts.

[0044] S1.4.5. Determine penetration rate categorization criteria and further adjust relevant dynamic parameters: Using a 40% intelligent connected vehicle penetration rate as a threshold, categorize mixed driving scenarios into two types: predominantly manual (≤40%) and predominantly intelligent connected vehicle (>40%). Dynamically optimize model parameters as the intelligent connected vehicle penetration rate changes through orthogonal experimental design (e.g., combining independent variables such as mainline traffic volume, entrance flow ratio, and slope).

[0045] S1.5. Set the mainline traffic volume and entrance flow ratio, and simulate merging behavior under different traffic loads using SUMO software.

[0046] S1.6. Compare the output data with the measured data, adjust the model parameters, and verify the model credibility, including the following steps: determine the correction indicators, including average travel time, average speed, number of lane changes, number of rear-end collisions, and number of lane change conflicts, calculate the relative error between the simulation data and the actual data through the relative error analysis method, and if the error exceeds 5%, adjust the parameter settings until the relative error of all correction indicators is less than 5%. Based on the default parameters and value ranges of the following model (IDM) and lane change model (LC2013) in the SUMO software, perform parameter calibration, and input the calibration parameter values and actual traffic flow characteristic values in the simulation scenario, and run the simulation to obtain the correction indicator values. If the relative error is still greater than 5%, continue to modify the parameter value and repeat the simulation until the error requirements are met. Finally, by comparing the relative error between the simulation value and the measured value, verify the validity of the calibration result and complete the model credibility verification. The relative error calculation formula is as follows: Where a i Indicates the current status data of the i-th parameter; b i Represents the simulation data of the i-th parameter.

[0047] (2) Step S2: Simulate merging behavior in a mixed driving scenario and extract rear-end collision and lane change conflict data.

[0048] Based on the SUMO software-built mountain urban expressway merging area road network and traffic demand file, set the simulation duration, start the simulation and output the vehicle trajectory file, record the real-time position, speed and acceleration data of all vehicles, use the traceExporter tool to convert the FCD file into an SSAM-compatible trajectory file, retaining the vehicle ID, coordinates, speed, acceleration and lane information; set the rear-end collision TTC threshold, lane change conflict PET threshold and conflict angle range detection parameters, run the SSAM trajectory file, automatically identify and classify rear-end collisions and lane change conflict events, count the number of general rear-end collisions, severe rear-end collisions, general lane change conflicts and severe lane change conflicts per hour, compare the conflict data output by the simulation with historical accident data, calculate the absolute error and relative error, and recalibrate the following model parameters or adjust the conflict threshold if the error exceeds the limit.

[0049] (3) Step S3. Construct a traffic conflict prediction model under different intelligent network penetration rates, and output a predicted value of the number of traffic conflicts through the traffic conflict prediction model.

[0050] The extracted rear-end collision and lane-changing collision data are divided into a mixed driving scenario data set dominated by manual driving and a mixed driving scenario data set dominated by intelligent networking: that is, divided into two types of data sets according to the intelligent networking penetration rate of 40%, namely, the mixed driving scenario of "dominated by manual driving" with a penetration rate ≤ 40% and the mixed driving scenario of "dominated by intelligent networking" with a penetration rate > 40%; the number of rear-end collisions, the number of lane-changing conflicts and independent variables (mainline traffic volume, entrance flow ratio, proportion of large vehicles, slope, etc.) under the two types of scenarios are extracted respectively, and the independent variables significantly correlated with the number of conflicts are screened, and the parameters with unstable rules are eliminated, and the intelligent networking penetration rate, mainline traffic volume, entrance flow ratio, proportion of large vehicles and slope are retained as model inputs; a negative binomial regression model is constructed to convert the general number of rear-end collisions into the number of lane-changing conflicts. , serious rear-end collisions, general lane-changing conflicts and serious lane-changing conflicts are used as dependent variables; the independent variables are logarithmically transformed and input into the model, and the regression coefficients are solved by the maximum likelihood estimation method. Models are established for the two types of mixed driving scenarios respectively, and the Akaike Information Criterion and Bayesian Information Criterion are used to evaluate the model fit; an independent simulation scheme is designed, the predicted number of conflicts is compared with the simulation results, and the absolute error and relative error are calculated. The error is less than 10%, which proves that the model is effective. If it exceeds the limit, the model parameters are recalibrated or the independent variable combination is optimized; the traffic parameters of the target merging area are input, and the predicted values of the four types of conflicts are directly output through the model, including the general number of rear-end collisions, the serious number of rear-end collisions, the general number of lane-changing conflicts and the serious number of lane-changing conflicts, which are used for safety evaluation or management strategy optimization.

[0051] (IV) Step S4: Construct a safety evaluation index system using the traffic conflict rate as an evaluation index, calculate the comprehensive weight of the evaluation index, and divide the safety levels.

[0052] S4.1. Select the general rear-end collision rate, severe rear-end collision rate, general lane change collision rate, and severe lane change collision rate to construct a traffic safety evaluation index system for mountainous urban expressway merging areas and form an evaluation matrix. The conflict rate calculation formula is: Where Zi represents the number of traffic conflicts occurring within an hour, i = 1, 2, 3, and 4 represent general rear-end collisions, severe rear-end collisions, general lane-changing conflicts, and severe lane-changing conflicts, respectively; Q represents the traffic volume within an hour.

[0053] S4.2. Use the max-min normalization method to standardize the conflict rate data. Use the entropy method to determine the objective weights, the DEMATEL method to determine the subjective weights, and the game theory integrated weighting method to calculate the comprehensive weights of the evaluation indicators. The specific implementation process is as follows:

[0054] S4.2.1. Determine objective weights using the entropy method: Calculate the traffic conflict rate for each option using the above formula. Normalize the data using the max-min method to eliminate the effects of inconsistent indicator dimensions. Calculate the weight of the indicator value for option i under the Kth indicator using the following formula: And ∑λ ik =1, calculate the information entropy of each indicator, the formula is Calculate the weight of each indicator, the formula is

[0055] S4.2.2. Use the DEMATEL method to determine subjective weights: construct a direct impact matrix between indicators, quantify the logical relationship between indicators, calculate the comprehensive impact matrix, and obtain the impact degree D of each indicator. i , influence degree C i , centrality M i and causal degree R i , the subjective weight of each indicator is determined according to the centrality and causality, and the calculation formula is as follows:

[0056]

[0057] M i =D i +C i

[0058] R i =D i -C i

[0059]

[0060] Where, t ij It represents the degree of direct impact plus indirect impact of indicator i on indicator j, that is, the comprehensive impact degree;

[0061] S4.2.3. Calculate the comprehensive weight of the evaluation indicators using the comprehensive integrated weighting method based on game theory: Use the entropy method and DEMATEL method to calculate the objective weight and subjective weight of each evaluation indicator respectively, and obtain the weight vector β1 T and β2 T , linearly combine the subjective and objective weights, the formula is β=a1β1 T +a2β2 T , where a1 and a2 are subjective and objective weight coefficients (a1+a2=1, a1>0, a2>0), based on the principle of game theory, to minimize the combination weight vector β and the subjective and objective weight vector β1 T and β2 T The deviation of is the goal, and the overall optimal countermeasure model is obtained. The formula is min||β-β1 T -β2T ||2=min||a1β1 T +a2β2 T -β1 T -β2 T ||2, by solving the optimization problem, we can obtain the first-order optimality condition Finally, Python software is used to calculate the subjective and objective weight coefficients a1 and a2 of each evaluation index according to the optimality conditions, thereby obtaining the comprehensive weight vector β, and Python code is used to calculate the comprehensive weight p* of different traffic conflict rate evaluation indicators.

[0062] S4.3. Classify security levels.

[0063] Based on the orthogonal test method, a simulation experiment was conducted on the traffic operation in the merging area of a mountainous urban expressway under an intelligent connected mixed driving environment. An evaluation matrix containing four traffic conflict rates was constructed. The conflict rates were divided into four safety levels, from level one to level four, using the cumulative frequency curve method, with level one being the safest and level four being the most dangerous. A sine function was introduced to improve the whitening weight function of the traditional grey fixed-weight clustering model. The whitening function values for each level were determined to make them more consistent with the actual evaluation situation, including a typical whitening weight function, an upper limit measure whitening weight function, a moderate measure whitening weight function, and a lower limit measure whitening weight function. The comprehensive weights obtained by the comprehensive integrated weighting method based on game theory were used to calculate the clustering coefficient of each simulation scheme at the four safety levels. A clustering estimate matrix was constructed, and the level corresponding to the maximum value was selected as the final safety evaluation result. The traffic safety level corresponding to each evaluation scheme was determined. The distribution of each traffic conflict rate indicator at different safety levels was statistically analyzed to determine the thresholds for different safety levels.

[0064] Compared with the existing technology, this application accurately reflects the traffic safety situation by constructing a traffic conflict prediction model for the merging area of mountain urban expressways, a conflict rate evaluation index system, a comprehensive integrated empowerment and improved gray clustering model, adapts to the intelligent connected mixed driving environment, provides a scientific basis for traffic optimization, effectively improves the safety level, reduces accidents, and ensures the safety of residents' travel.

[0065] Regarding the embodiments of the present invention, it should also be noted that, in the absence of conflict, the embodiments of the present invention and the features therein may be combined with each other to obtain new embodiments.

[0066] The above description is only a preferred embodiment of the present invention and does not constitute any form of limitation to the present invention. The scope of protection of the present invention shall be subject to the scope of protection of the claims. Although the present invention has been disclosed as above with preferred embodiments, it is not intended to limit the present invention. Any technician familiar with this profession can make some changes or modifications to equivalent embodiments of equivalent changes using the technical content disclosed above without departing from the scope of the technical solution of the present invention. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the scope of the technical solution of the present invention.

Claims

1. A safety assessment method for merging areas on intelligent connected mixed-driving mountainous urban expressways, characterized in that: The following steps are involved: Construct a simulation scenario of a merging area on a mountainous urban expressway; Simulate merging behavior in mixed driving scenarios and extract rear-end collision and lane change conflict data; Construct a traffic conflict prediction model under different intelligent network penetration rates, and output the predicted value of traffic conflicts through the traffic conflict prediction model; A safety evaluation index system is constructed using traffic conflict rate as an evaluation indicator, the comprehensive weight of the evaluation index is calculated, and the safety level is divided.

2. The safety evaluation method according to claim 1, characterized in that: The construction of a simulation scene of a mountainous urban expressway merging area includes the following steps: A typical mountainous urban expressway merging area was selected, lanes were numbered, and sections were divided according to their functions; Collect traffic operation data, traffic conflict data and road geometry parameters at merging areas of mountainous urban expressways; Conduct traffic operation analysis based on collected data and draw the road network of merging areas of mountainous urban expressways; The simulation scenarios are divided into mixed driving scenarios dominated by manual driving and mixed driving scenarios dominated by intelligent networking according to different penetration rates of intelligent networking. Set the mainline traffic volume and entrance flow ratio to simulate merging behavior under different traffic loads; Compare the output data with the measured data, adjust the model parameters, and verify the credibility of the model.

3. The safety evaluation method according to claim 2, characterized in that: The traffic operation data includes data such as traffic flow, vehicle speed, and vehicle type; the traffic conflict data includes the number and severity of rear-end conflicts and lane change conflicts; and the road geometry parameters include data such as slope, lane width, and acceleration lane length.

4. The safety evaluation method according to claim 2, characterized in that: The simulation scenarios are divided into mixed driving scenarios dominated by manual driving and mixed driving scenarios dominated by intelligent networking according to different intelligent networking penetration rates, including the following steps: selection, parameter calibration and verification of the following and lane changing models of manually driven vehicles; selection, parameter calibration and verification of the following and lane changing models of intelligent connected vehicles; determination of the intelligent networking penetration rate division standard, and division of the mixed driving scenarios into two types: one dominated by manual driving and the other dominated by intelligent networking.

5. The safety evaluation method according to claim 2, characterized in that: The method of comparing the output data with the measured data, adjusting the model parameters, and verifying the model credibility includes the following steps: Determine correction indicators, including average travel time, average speed, number of lane changes, number of rear-end collisions, and number of lane change conflicts; The relative error between the simulation data and the actual data is calculated by relative error analysis. If the error exceeds 5%, the model parameter settings are adjusted until the relative errors of all calibration indicators are less than 5%; Calibrate the parameters based on the default parameters and value ranges of the car-following and lane-changing models. Enter the calibration parameter values and actual traffic flow characteristics in the simulation scenario and run the simulation to obtain the corrected index values. If the relative error is still greater than 5%, continue to modify the parameter values and repeat the simulation until the error requirement is met. By comparing the relative errors between the simulated values and the measured values, the validity of the calibration results is verified and the model credibility is verified. The relative error calculation formula is as follows: Where a i Indicates the current status data of the i-th parameter; b i Represents the simulation data of the i-th parameter.

6. The safety evaluation method according to claim 1, characterized in that: The simulation simulates merging behavior in a mixed driving scenario and extracts rear-end collision and lane change conflict data, including the following steps: Based on the road network file and traffic demand file of the mountain city expressway merging area simulation scenario, set the simulation duration, start the simulation and output the vehicle trajectory file, recording the real-time position, speed and acceleration data of all vehicles; Convert FCD files to SSAM-compatible trajectory files, preserving vehicle ID, coordinates, speed, acceleration, and lane information; Set the TTC threshold for rear-end collisions, the PET threshold for lane change collisions, and the conflict angle range detection parameters, run the SSAM trajectory file, automatically identify and classify rear-end collisions and lane change conflicts, and count the number of general rear-end collisions, severe rear-end collisions, general lane change conflicts, and severe lane change conflicts per hour; Compare the conflict data output by the simulation with historical accident data, calculate the absolute error and relative error, and if the error exceeds the limit, recalibrate the parameters of the car-following model or adjust the conflict threshold.

7. The safety evaluation method according to claim 2, characterized in that: The method of constructing a traffic conflict prediction model under different intelligent network penetration rates and outputting a traffic conflict number prediction value through the traffic conflict prediction model includes the following steps: The extracted rear-end collision and lane change conflict data are divided into a mixed driving scenario dataset dominated by manual driving and a mixed driving scenario dataset dominated by intelligent networking; The number of rear-end collisions, lane-changing conflicts, and independent variable data were extracted from the two data sets, and the independent variables significantly correlated with the number of conflicts were selected as model inputs. A negative binomial regression model was constructed, with the number of conflicts as the dependent variable. The independent variables were logarithmically transformed and entered into the negative binomial regression model. The regression coefficients were solved using the maximum likelihood estimation method. Traffic conflict prediction models were established for the two types of mixed driving scenarios. The goodness of fit of the traffic conflict prediction models was evaluated using the Akaike Information Criterion and the Bayesian Information Criterion. Design an independent simulation plan, compare the predicted number of conflicts with the simulation results, calculate the absolute error and relative error, and recalibrate the traffic conflict prediction model parameters or optimize the independent variable combination if the error exceeds the limit; The traffic parameters of the target merging area are input, and the traffic conflict prediction model outputs the predicted value of the number of traffic conflicts, including the number of general rear-end collisions, the number of severe rear-end collisions, the number of general lane change conflicts, and the number of severe lane change conflicts.

8. The safety evaluation method according to claim 1, characterized in that: The traffic conflict rate includes general rear-end collision rate, severe rear-end collision rate, general lane change conflict rate and severe lane change conflict rate. The conflict rate calculation formula is: Where Zi represents the number of traffic conflicts occurring within an hour, i = 1, 2, 3, and 4 represent general rear-end collisions, severe rear-end collisions, general lane-changing conflicts, and severe lane-changing conflicts, respectively; Q represents the traffic volume within an hour.

9. The safety evaluation method according to claim 1, characterized in that: The calculation of the comprehensive weight of the evaluation index comprises the following steps: The objective weight is determined by using the entropy method: the traffic conflict rate data is normalized using the max-min normalization method; the index value proportion of the i-th option under the K-th index is calculated using the following formula: And ∑λ ik =1; calculate the information entropy of each indicator, the formula is Calculate the objective weight of each indicator, the formula is Use the DEMATEL method to determine the subjective weight: construct the direct impact matrix between the evaluation indicators, quantify the logical relationship between the indicators, calculate the comprehensive impact matrix, and obtain the influence degree D of each indicator. i , influence degree C i , centrality M i and causal degree R i , the subjective weight of each indicator is determined according to the centrality and causality, and the calculation formula is as follows: M i =D i +C i R i =D i -C i Where, t ij It represents the degree of direct impact plus indirect impact of indicator i on indicator j, that is, the comprehensive impact degree; The comprehensive weight of the evaluation indicators is calculated using the comprehensive integrated weighting method based on game theory: by calculating the objective weight and subjective weight of each evaluation indicator, the weight vector β1 is obtained T and β2 T ; Linearly combine the subjective and objective weights, the formula is β=a1β1 T +a2β2 T , where a1 and a2 are subjective and objective weight coefficients (a1+a2=1, a1>0, a2>0); based on the principle of game theory, to minimize the combined weight vector β and the subjective and objective weight vector β1 T and β2 T The deviation of is the goal, and the overall optimal countermeasure model is obtained. The formula is min||β-β1 T -β2 T ||2=min||a1β1 T +a2β2 T -β1 T -β2 T ||2, by solving the optimization problem, we can obtain the first-order optimality condition According to the optimality condition, the subjective and objective weight coefficients a1 and a2 of each evaluation index are obtained to obtain the comprehensive weight vector β, and the comprehensive weight p* of different traffic conflict rate evaluation indicators is calculated.

10. The safety evaluation method according to claim 1, characterized in that: The safety level division includes: using the cumulative frequency curve method to divide the traffic conflict rate into four safety levels, with level one being the safest and level four being the most dangerous; introducing a sine function to improve the whitening weight function of the gray fixed-weight clustering model, calculating the clustering coefficient of each simulation scheme under the four safety levels, and constructing a clustering estimation value matrix, and selecting the level corresponding to the maximum clustering coefficient as the final safety evaluation result.

Images