Risk assessment method for left-side off-ramp traffic of urban expressway based on risk field

Abstract

This invention discloses a traffic risk assessment method for left-side exit ramps of urban expressways based on a risk field, relating to the field of traffic safety technology. The method includes the following steps: collecting driving data from the left-side exit ramps of the urban expressway to be assessed; constructing a composite risk field model and calculating the instantaneous comprehensive risk value; the composite risk field model is as follows: obtaining the instantaneous comprehensive risk through objective physical risk and driver physiological-visual state adjustment factors; obtaining the risk peak value of the instantaneous comprehensive risk based on the instantaneous comprehensive risk value calculated by the model; and classifying the risk into safe, dangerous, and extremely dangerous levels based on the comparison relationship between the risk peak value and preset first and second thresholds, thereby obtaining the risk level assessment result. This invention enables a comprehensive quantitative assessment of traffic risks at left-side exit ramps, significantly improving the sensitivity and precision of risk identification.

Description

Technical Field

[0001] This invention relates to traffic safety technology, and more particularly to a method for assessing traffic risks at left-side exit ramps of urban expressways based on a risk field. Background Technology

[0002] Exit ramps on urban expressways are high-risk areas for traffic accidents, with a significantly higher collision rate per mile than other road sections. Left-hand exit ramps, in particular, are more complex to navigate and pose greater safety hazards because their design contradicts drivers' usual right-hand exit habits. With the continuous growth of urban traffic flow, conducting scientific and accurate traffic risk assessments of exit ramp areas has become an urgent need to improve road safety management.

[0003] Existing safety assessment technologies, such as those relying on traditional time-of-collision (TTC) metrics, often fail to comprehensively capture the dynamic evolution of risks and do not fully integrate the physiological and visual load changes experienced by drivers during decision-making and operation. For example, Chinese patent CN111311093A discloses a method for risk assessment and early warning at urban road intersections based on driver physiological data. This method establishes a vehicle risk assessment model, calculates the risk value of a target vehicle by combining it with driver physiological data, and determines the risk level based on a risk level classification matrix. Although this method considers driver physiological data, its risk assessment model is mainly based on traditional vehicle kinematic parameters, introducing physiological data only as a weighting coefficient, and fails to deeply integrate the dynamic impact of the driver's real-time physiological and visual load on risk perception.

[0004] Therefore, there is a need for a traffic risk assessment technology that can comprehensively consider objective physical risks and the driver's subjective state, and has higher risk identification sensitivity and precision for the specific scenario of the left-side exit ramp. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a traffic risk assessment method for the left-side exit ramp of an urban expressway based on a risk field, addressing the deficiencies in the existing technology.

[0006] The technical solution adopted by this invention to solve its technical problem is: a traffic risk assessment method for left-side exit ramps of urban expressways based on a risk field, comprising the following steps: 1) Collect driving data from the left-hand exit ramps of the urban expressway to be evaluated; the driving data includes ramp configuration parameters, driver characteristic data, and driver physiological and visual data; 2) Construct a composite risk field model and calculate the instantaneous comprehensive risk value; Composite risk field model: The instantaneous comprehensive risk at time t is obtained by combining objective physical risk with the driver's physiological-visual state modulation factor. ; 3) Obtain the peak value of the instantaneous comprehensive risk based on the instantaneous comprehensive risk value calculated by the model. ; ; 4) Based on the above With the preset first threshold Second threshold By comparing the relationships, risks are classified into safe levels, dangerous levels, and extremely dangerous levels; thus, risk level assessment results are obtained.

[0007] According to the above scheme, in step 1), the ramp configuration parameters include the length of the deceleration lane and the necessary number of lane changes; the driver characteristic data includes the driver's age, gender, and driving experience; the driver's physiological and visual data includes the driver's average heart rate, pupil dilation, fixation duration, and saccade duration indicators.

[0008] According to the above scheme, in step 2), ; in, For objective physical risks, It is a driver's physiological-visual state regulation factor.

[0009] According to the above scheme, in step 2), the objective physical risk is calculated as follows:

[0010] In the formula, Indicates the risk load of the main vehicle; Indicates the risk load of the interactive object; It is the environmental risk capacitive reactance or dielectric constant, which characterizes the attenuation or enhancement effect of road geometry, pavement conditions and weather factors on risk propagation; It is a risk action function that describes the relative speeds between the main vehicle and each interacting object. Effective distance and relative azimuth The degree of risk is determined jointly.

[0011] According to the above scheme, in step 2), the risk load of the interactive object is calculated as follows: ; Where m is the vehicle mass; This refers to the vehicle's real-time speed. The speed exponent; This represents the risk conversion coefficient for undetermined risks.

[0012] According to the above scheme, in step 2), the driver's physiological-visual state regulation factor The calculation is as follows:

[0013] in, Represents the k-th real-time monitored physiological or visual indicator at time t; The representative standardized the indicators; The risk weight coefficient for the k-th indicator; The standardized processing method is as follows: , and Let be the baseline mean and standard deviation of the k-th indicator.

[0014] According to the above scheme, in step 3), the threshold is calibrated using the statistical distribution definition method based on extreme value theory EVT. and .

[0015] According to the above scheme, the assessment method also includes step 5) using a multi-logit model to establish a quantitative relationship between the risk level and each influencing factor, and to identify the key risk factors of the left exit ramp of the urban expressway to be assessed.

[0016] The present invention also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described scheme.

[0017] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the method in the above-described scheme.

[0018] The beneficial effects of this invention are: 1. The composite risk field model proposed in this invention integrates objective physical risks based on vehicle kinematics and road environment with subjective risk adjustment factors based on the driver's real-time physiological and visual state, thereby achieving a comprehensive quantitative assessment of traffic risks at the left-side exit ramp.

[0019] 2. By introducing a physiological-visual risk moderating factor, this method model can capture potential risk increases caused by driver fatigue, distraction, tension, or excessive cognitive load. It can also issue warnings in hidden risk situations where vehicle kinematic indicators appear normal, thus improving the sensitivity and precision of the model's risk identification.

[0020] 3. By combining multiple Logit models to analyze the causes of risk levels, it can quantitatively reveal the specific impact of different factors (ramp geometric parameters, driver individual characteristics, and even fluctuations in physiological visual indicators) on the probability of occurrence of different risk levels, providing a scientific basis for safety intervention and design optimization of left-side exit ramps. Attached Figure Description

[0021] The present invention will be further described below with reference to the accompanying drawings and embodiments. In the accompanying drawings: Figure 1 This is a flowchart of a method according to an embodiment of the present invention; Figure 2 This is a schematic diagram simulating an experimental scenario according to an embodiment of the present invention; Figure 3 This is a schematic diagram of a vehicle lane changing according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the risk value distribution in the time-space dimension of the composite risk field model according to an embodiment of the present invention. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0023] Example 1 like Figure 1 As shown, a traffic risk assessment method for left-side exit ramps of urban expressways based on a risk field includes the following steps: 1) Collect driving data from the left-hand exit ramps of the urban expressway to be evaluated; the driving data includes ramp configuration parameters, driver characteristic data, and driver physiological and visual data; The ramp configuration parameters include the length of the deceleration lane and the necessary number of lane changes; Driver characteristic data includes the driver's age, gender, and driving experience; driving experience is characterized by the driver's years of driving experience and mileage. Driver physiological and visual data were collected using an eye tracker and physiological sensors, including average heart rate, pupil dilation, fixation duration, and saccade duration. 2) Construct a composite risk field model and calculate the instantaneous comprehensive risk value; The composite risk field model is constructed as follows: Through objective physical risks With driver physiological-visual state modulators Obtain the instantaneous comprehensive risk at time t ; .

[0024] This model structure allows objective environmental threats to be dynamically adjusted based on the driver's real-time status, reflecting the interactive characteristics of risk between the subjective and objective elements. The objective physical risks are represented as follows:

[0025] In the formula, It represents the inherent risk potential energy or load of the main vehicle, reflecting its ability to generate risk, and is related to driver reaction characteristics and vehicle performance (acceleration and deceleration);

[0026] in, This indicates the driver's reaction characteristics, specifically their level of response to traffic conflicts; v represents the vehicle's speed. Indicates acceleration; The risk load representing the interactive object obj reflects its response to the host vehicle's risk field or the threat it poses itself, and is related to the type, quality, and speed of the interactive object.

[0027] Where m is the vehicle mass; This refers to the vehicle's real-time speed. The velocity exponent is typically set to 2 to characterize the physical relationship between risk potential energy and vehicle kinetic energy. In the high-speed flow environment at the left exit, a value of 2 is used. It can non-linearly amplify the risk value brought by high vehicle speed, thereby more keenly capturing the potential threat of fast lane diversion; The risk conversion coefficient is used to map physical kinetic energy into a dimensionless risk potential energy value. It is the environmental risk capacitance or dielectric constant, which characterizes road geometry such as slope and curvature; as well as the attenuation or enhancement effect of road surface conditions and weather factors on risk propagation; The higher the value, the lower the risk; It is a risk action function that describes the relative speeds between the main vehicle and each interacting object. Effective distance and relative azimuth The degree of risk is determined jointly.

[0028] Objective physical risks The objective collision threat posed by the surrounding traffic environment to the main vehicle is quantified. The calculation process is as follows: First, the interactive objects around the main vehicle are identified, including other vehicles and fixed obstacles. Then, the relative speed, effective distance, and relative azimuth angle between the main vehicle and each interactive object are calculated and substituted into the calculation formula to obtain the risk value posed by each interactive object to the main vehicle.

[0029] Among them, physiological-visual risk moderating factors The objective risk assessment results, used to dynamically adjust based on the driver's real-time internal state, reflect the crucial role of human factors in risk formation; the calculation is as follows: ; in, Represents the k-th real-time monitored physiological or visual indicator at time t; The representative standardized the indicators; The standardized processing method is as follows: ; and These are the baseline mean and standard deviation of the indicator; These are the risk weight coefficients for each indicator, reflecting the amplification effect of the indicator deviating from the normal baseline on risk. Weight The calibration was performed using a risk association analysis method based on logistic regression.

[0030] This embodiment constructs a composite risk field model. This model aims to go beyond the traditional TTC-based static threshold method by integrating objective physical threats with the driver's subjective perceived load to achieve a more refined assessment of conflict risk.

[0031] 3) Based on the instantaneous comprehensive risk value calculated by the model, obtain the peak value of the instantaneous comprehensive risk at any given time or within a certain event window. ;

[0032] 4) Based on the above With the preset first threshold Second threshold Based on the comparative relationship, risks are classified into safe level, dangerous level, and extremely dangerous level; In this embodiment, a statistical distribution-based threshold selection method based on extreme value theory (EVT) is used. and .

[0033] Ultimately, the risk level assessment results for the left-hand exit ramp of the urban expressway to be evaluated were obtained.

[0034] Example 2 A traffic risk assessment method for left-side exit ramps of urban expressways based on a risk field includes the following steps: 1) Collect driving data from the left-hand exit ramps of the urban expressway to be evaluated; the driving data includes ramp configuration parameters, driver characteristic data, and driver physiological and visual data; The ramp configuration parameters include the length of the deceleration lane and the necessary number of lane changes; Driver characteristic data includes the driver's age, gender, and driving experience; driving experience is characterized by the driver's years of driving experience and mileage. Driver physiological and visual data were collected using an eye tracker and physiological sensors, including average heart rate, pupil dilation, fixation duration, and saccade duration. In this embodiment, the aforementioned driving data was obtained by setting up an experimental platform.

[0035] This embodiment employs a high-fidelity driving simulator platform, which provides an immersive driving view and records vehicle dynamic data (including speed, acceleration, and position) in real time, with a sampling frequency of 10Hz. The experimental scenario simulates the left-side exit ramp of an underground interchange on an urban expressway. The simulation device for the experimental scenario is shown below. Figure 2 To comprehensively explore the impact of different factors, the experiment employed an orthogonal design method, generating nine different driving scenarios. These scenarios were achieved by changing at least one of the following: speed bump length, number of lane changes, and traffic density. Speed ​​bump lengths of 145 meters, 170 meters, and 195 meters were set; drivers were required to perform one, two, or three different lane change maneuvers to exit the ramp, adjusting the traffic density of the mainline lanes, especially those closest to the ramp. A lane-changing diagram is shown below. Figure 3 .

[0036] To ensure the authenticity and validity of the experimental data, 30 participants with valid driver's licenses were recruited. All participants were required to have at least 3 years of actual driving experience and normal vision (or corrected vision), without color blindness or color weakness.

[0037] During the experiment, in addition to the vehicle operation data recorded by the driving simulator platform itself, visual performance data, physiological state data, and driver static characteristics were also recorded simultaneously.

[0038] 2) Construct a composite risk field model and calculate the instantaneous comprehensive risk value; The composite risk field model is constructed as follows: Through objective physical risks With driver physiological-visual state modulators Obtain the instantaneous comprehensive risk at time t ;

[0039] This model structure allows objective environmental threats to be dynamically adjusted based on the driver's real-time status, reflecting the interactive characteristics of risk between the subjective and objective elements. The objective physical risks are represented as follows:

[0040] In the formula, It represents the inherent risk potential energy or load of the main vehicle, reflecting its ability to generate risk, and is related to driver reaction characteristics and vehicle performance (acceleration and deceleration); The risk load representing the interactive object obj reflects its response to the host vehicle's risk field or the threat it poses itself, and is related to the type, quality, and speed of the interactive object.

[0041] Where m is the vehicle mass; This refers to the vehicle's real-time speed. The velocity exponent is typically set to 2 to characterize the physical relationship between risk potential energy and vehicle kinetic energy. In the high-speed flow environment at the left exit, a value of 2 is used. It can non-linearly amplify the risk value brought by high vehicle speed, thereby more keenly capturing the potential threat of fast lane diversion; The undetermined risk conversion coefficient is used to map physical kinetic energy into a dimensionless risk potential energy value. The calibration of the undetermined risk conversion coefficient adopts a regression analysis method based on historical accident data: First, historical traffic accident data of urban expressways are collected, and the accident vehicle type, corresponding mass m, and collision speed are extracted. The data includes the comprehensive accident loss (including property damage and casualties, converted as the severity value S); secondly, the data is grouped by speed range and vehicle type, and the average severity of each group is calculated; finally, based on the formula... Nonlinear least squares fitting is performed to calculate coefficients that characterize the severity of the actual accident consequences for the corresponding vehicle type. .

[0042] It is the environmental risk capacitance or dielectric constant, which characterizes road geometry such as slope and curvature; as well as the attenuation or enhancement effect of road surface conditions and weather factors on risk propagation; The higher the value, the lower the risk; It is a risk action function that describes the relative speeds between the main vehicle and each interacting object. Effective distance and relative azimuth The degree of risk is determined jointly.

[0043] Objective physical risks It quantifies the objective collision threat posed by the surrounding traffic environment to the main vehicle.

[0044] Physiological-visual risk modulators The objective risk assessment results, used to dynamically adjust based on the driver's real-time internal state, reflect the crucial role of human factors in risk formation; the calculation is as follows:

[0045] in, Represents the k-th real-time monitored physiological or visual indicator at time t; The representative standardized the indicators; The standardized processing method is as follows: , and These are the baseline mean and standard deviation of the indicator; These are the risk weight coefficients for each indicator, reflecting the amplification effect of the indicator deviating from the normal baseline on risk. Weight The calibration employed a risk association analysis method based on logistic regression. First, using objective vehicle kinematic data (such as lateral acceleration) collected during the experiment, the experimental data slices were labeled into high-risk and low-risk categories. Second, a standardized physiological-visual indicator system was constructed with the risk label as the dependent variable. Using a binomial logistic regression model with independent variables, calculate the regression coefficients of each indicator on the high-risk state. Finally, the regression coefficients that are statistically significant and positive are selected and normalized to obtain the results. This ensures that the weights objectively reflect the contribution of each physiological indicator to actual driving risks, achieving a scientific mapping from statistical data patterns to physical model parameters. The physiological-visual risk moderating factor is based on 1. When all indicators are at the baseline level, the moderating factor is 1; when certain indicators show abnormal driver conditions, such as excessive heart rate or excessive pupil dilation indicating stress or increased workload, the physiological-visual risk moderating factor is greater than 1, amplifying the risk assessment.

[0046] See the diagram of risk field distribution. Figure 4 It visually demonstrates the dynamic changes in risk over time and space as a vehicle passes through the left-side exit ramp.

[0047] This embodiment constructs a composite risk field model. This model aims to go beyond the traditional TTC-based static threshold method by integrating objective physical threats with the driver's subjective perceived load to achieve a more refined assessment of conflict risk.

[0048] 3) Based on the instantaneous comprehensive risk value calculated by the model, obtain the peak value of the instantaneous comprehensive risk at any given time or within a certain event window. ;

[0049] 4) Based on the above With the preset first threshold Second threshold By comparing risk levels, risks are classified into safe, hazardous, and extremely hazardous levels; thus, risk level assessment results are obtained. In this embodiment, a statistical distribution-based threshold selection method based on extreme value theory (EVT) is used. and .

[0050] Based on the sample experiment in step 1), all 3444 samples were obtained. The risk distribution is first analyzed by constructing a set of risk peaks from the samples and performing a distribution test to verify that it conforms to the characteristics of a long-tailed distribution, meaning that the vast majority of samples are in the low-risk range and a very small number of samples are in the high-risk range, which aligns with the safety continuum principle in the field of traffic safety. Then, a threshold overshoot (POT) model is used to select high-risk data from the tail of the sample set and fit a generalized Pareto distribution (GPD) curve, calculating the risk field value at which the cumulative probability reaches 95% as the threshold. This threshold statistically represents a critical point that occurs only in a very small percentage of extreme operating conditions (around 5%), corresponding to a physically severe loss of control or a near-collision state. For general hazard levels, the risk threshold... The calibration is based on the statistical characteristics of state transitions in traffic flow theory. The cumulative distribution function (CDF) of the sample set is calculated, and the risk field value at which the cumulative probability reaches 85% is selected as the baseline. This is based on the 85th percentile rule commonly used in traffic engineering, marking the critical point where a driver's psychological load significantly increases and evasive action is required. Ultimately, the risk level is determined based on the thresholds calculated from the two objective factors mentioned above. Divided into three cases: ; In this embodiment, multiple Logit models are used to quantitatively reveal the impact of each factor on the risk level, providing a scientific basis for safety intervention and design optimization of the left-side exit ramp.

[0051] 5) Employ a multivariate Logit model to establish a quantitative relationship between risk level and various influencing factors; calculate the relative risk ratio and average marginal effect to identify key risk factors.

[0052] A multinomial logit (MNL) model is used to analyze the impact of various factors on the selection of discrete risk levels. The MNL model is constructed based on the theory of stochastic utility maximization. In the complex traffic environment of the left-side exit ramp, the final risk state of the nth traffic sample (including the driver, vehicle, and current environment) can be regarded as the "selection" of a certain instantaneous state by the human-vehicle-road system under multiple constraints. The model assumes that each observed sample will choose a state that maximizes its potential utility. Highest risk level ; Risk level in model building Utility on sample n for: , in, For the utility of risk level i, For explanatory variables, The random part; In accordance with risk level The relevant parameter vector to be estimated; the random component of the utility The unobservable part encompasses random disturbances that are not collected (such as momentary driver inattention or minute changes in road surface friction). In the MNL model, it is assumed that... Follows independent and identically distributed Extreme value distribution. Explanatory variables. It includes factors that may affect the risk level, such as ramp configuration, driver characteristics, and statistical features extracted from physiological or visual data.

[0053] Risk level 1 is set as the reference group, i.e. Then sample n belongs to risk level. probability for: ; Using a database containing 3444 samples, the log-likelihood function is maximized. To estimate the parameter vector of sample n and :

[0054] in, It is an indicator variable, when the actual risk level of the observed sample n is It equals 1 if it is true, otherwise it is 0.

[0055] Specifically, the utility of risk level 2 is:

[0056] The utility of risk level 3 is:

[0057] in, The length of the deceleration lane. For the number of lane changes, Driver's age; Finally, a parameter significance test is performed on the estimated parameter vector. and Each coefficient in the study was subjected to a significance test to determine whether each accepting variable had a statistically significant effect on the choice of risk level.

[0058] Explanation variables Parameter vector estimation corresponding to risk level i; calculation of relative risk ratio , representing explanatory variables For each additional unit, select the factor by which the occurrence ratio of risk level i changes. This indicates that an increase in factors will increase the relative probability of risk level i.

[0059] Calculate the average marginal utility (AME). The formula is , representing explanatory variables The average impact of a unit change on the selection probability of each risk level i. By comprehensively analyzing parameter significance, relative risk ratio, and average marginal effect, key factors influencing the traffic conflict risk of left-side exit ramps on urban expressways are identified.

[0060] The analysis results show that all these parameter estimates passed the significance test, indicating that each explanatory variable has a statistically significant impact on the choice of risk level.

[0061] It should be understood that those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.

Claims

1. A method for assessing traffic risk at left-side exit ramps of urban expressways based on a risk field, characterized in that, Includes the following steps: 1) Collect driving data from the left-hand exit ramps of the urban expressway to be evaluated; the driving data includes ramp configuration parameters, driver characteristic data, and driver physiological and visual data; 2) Construct a composite risk field model and calculate the instantaneous comprehensive risk value; The composite risk field model is as follows: through objective physical risk With driver physiological-visual state modulators Obtain the instantaneous comprehensive risk at time t ; Among them, objective physical risks The calculation is as follows: In the formula, Indicates the risk load of the main vehicle; Indicates the risk load of the interactive object; It is environmental risk tolerance, which characterizes the attenuation or enhancement effect of road geometry, pavement conditions and weather factors on risk propagation; It is a risk action function that describes the relative speeds between the main vehicle and each interacting object. Effective distance and relative azimuth The degree of risk is determined jointly; Driver physiological-visual state modulation factor The calculation is as follows: in, Represents the k-th real-time monitored physiological or visual indicator at time t; The representative standardized the indicators; The risk weight coefficient for the k-th indicator; The standardized processing method is as follows: , and Let be the baseline mean and standard deviation of the k-th indicator; 3) Obtain the peak value of the instantaneous comprehensive risk based on the instantaneous comprehensive risk value calculated by the model. ; 4) Based on the aforementioned risk peak With the preset first threshold Second threshold By comparing the relationships, risks are classified into safe levels, dangerous levels, and extremely dangerous levels; thus, risk level assessment results are obtained.

2. The traffic risk assessment method for left-side exit ramps of urban expressways based on risk fields according to claim 1, characterized in that, In step 1), the ramp configuration parameters include the deceleration lane length and the necessary number of lane changes; Driver characteristic data includes the driver's age, gender, and driving experience; driver physiological and visual data includes the driver's average heart rate, pupil dilation, fixation duration, and saccade duration.

3. The traffic risk assessment method for left-side exit ramps of urban expressways based on risk fields according to claim 1, characterized in that, In step 2), 。 4. The traffic risk assessment method for left-side exit ramps of urban expressways based on risk fields according to claim 1, characterized in that, In step 2), the risk load of the interactive object is calculated as follows: ； Where m is the vehicle mass; This refers to the vehicle's real-time speed. The speed exponent; This is the risk conversion coefficient.

5. The traffic risk assessment method for left-side exit ramps of urban expressways based on risk fields according to claim 1, characterized in that, In step 3), the threshold is calibrated using the statistical distribution bounding method based on the extreme value theory EVT. and .

6. The traffic risk assessment method for left-side exit ramps of urban expressways based on risk fields according to claim 1, characterized in that, The assessment method also includes step 5), which uses a multinomial Logit model to establish a quantitative relationship between the risk level and each influencing factor, and identifies the key risk factors of the left exit ramp of the urban expressway to be assessed.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method described in any one of claims 1 to 6.

Images