Highway emergency accident monitoring and early warning method based on road condition features

Abstract

The application discloses a highway sudden accident monitoring and early warning method based on road condition characteristics, relates to the field of highway sudden accident monitoring and early warning, and comprises the following steps: acquiring dynamic environment data of a next road section of a target vehicle in real time, combining with initial dry state of a road surface to calculate real-time road surface humidity data, correcting a basic road risk index generated based on inherent road conditions to obtain a comprehensive road risk value; collecting current vehicle driving state data in real time, and associating and matching the comprehensive road risk value to generate a special risk probability for the target vehicle, realizing accurate coupling of road risk and vehicle state, improving the pertinence and reliability of highway sudden accident early warning, effectively reducing the accident rate, and guaranteeing road traffic safety.

Description

Technical Field

[0001] This invention belongs to the field of highway monitoring and early warning technology, and relates to a method for monitoring and early warning of sudden accidents on highways based on road condition characteristics. Background Technology

[0002] Against the backdrop of rapid development in the transportation industry, highways, as the core channels for cross-regional travel and freight transport, are directly related to the operation of the social economy and the safety of people's lives and property. However, the highway driving environment is complex, and the occurrence of accidents is often random and easily affected by the combined effects of dynamic environment and inherent road characteristics. If timely monitoring and accurate warnings are not provided to passing vehicles, it can easily trigger a chain of accidents, causing traffic congestion and even casualties.

[0003] Currently, existing technologies have proposed methods for highway accident early warning. For example, the invention patent with publication number CN113611139B proposes an intelligent early warning method and system suitable for sudden road conditions. When a new road obstacle appears, it generates early warning information containing the road obstacle's address and range and sends it to the target vehicle terminal passing by the road obstacle address. It then determines whether the road segment where the road obstacle is located is a lane changeable area. If not, it acquires the road obstacle image information and calculates the safe distance between the road obstacle and the road edge. If the vehicle width is not less than the safe distance, it updates the navigation route according to preset scheduling principles. Before updating the route, it tracks the vehicle's position in real time and periodically identifies vehicle characteristic information at the road obstacle. If the characteristics change, the early warning is deemed invalid and the update is canceled. Simultaneously, it combines factors such as road intersections and traffic flow to determine whether to update the route and how to select the optimal alternative route. This helps drivers prepare for sudden road conditions in advance, thereby improving driving convenience.

[0004] However, although existing technical solutions have achieved certain results in the field of highway accident early warning, they still have the following shortcomings: First, highways often pass through remote mountainous areas, depressions and other areas with complex geographical conditions. Some sections have complex alignments such as sharp bends and steep slopes due to terrain limitations. Rainy weather will further increase the risk of slippery roads. Current technologies mainly rely on relatively static or simple dynamic parameters such as road obstacle information, vehicle size and road intersections to make route decisions. They do not adequately consider the risk superposition effect under severe weather conditions, which may lead to delayed early warning response and increase the difficulty and safety risks of vehicles avoiding obstacles in special road sections.

[0005] Secondly, existing technologies focus too much on the single judgment of whether a vehicle's physical size can pass through road obstacles, while ignoring that the vehicle's dynamic driving behavior itself is also a key variable that constitutes accident risk. Although the vehicle width and safe distance comparison model can solve the passability problem in specific scenarios, the model has extremely limited dimensions and cannot reflect the comprehensive risks under the coupling of complex road conditions and bad driving behavior. This limits the generation of warning and path planning schemes to fully cover the potential dangers in the driving process, thus reducing the effectiveness of safety protection. Summary of the Invention

[0006] In view of this, in order to solve the problems mentioned in the background art, the present invention provides a method for monitoring and early warning of sudden accidents on highways based on road condition characteristics.

[0007] The objective of this invention can be achieved through the following technical solution: a highway emergency accident monitoring and early warning method based on road condition characteristics, including: S1, real-time acquisition of rainfall conditions and road shape data of the target vehicle's current location and the next road segment of its travel route, the road shape data including the curvature amplitude value of the current location and the next road segment, curve continuity indicators and the number of tunnels.

[0008] S2. Obtain the initial dryness data of the road surface of the next road segment, and calculate the real-time road surface humidity level of the corresponding road segment in combination with the rainfall conditions.

[0009] S3. Determine the risk level based on road shape data, generate a basic road risk index, and adjust the basic road risk index based on road surface moisture level to generate a comprehensive road risk value.

[0010] S4. Real-time collection of the vehicle's speed and lane change frequency.

[0011] S5. Perform correlation analysis between the comprehensive road risk value and driving speed and lane change frequency to calculate the accident risk probability of the target vehicle in the next road segment, issue warnings based on the accident risk probability classification, and push them to the target vehicle.

[0012] Compared with the prior art, the beneficial effects of the present invention are as follows: (1) The present invention obtains the dynamic environmental data of the next road segment of the target vehicle in real time, and calculates the real-time road surface humidity data in combination with the initial dry state of the road surface, and then corrects the basic road risk index generated based on the inherent road conditions to obtain the comprehensive road risk value. It realizes the dynamic fusion assessment of the risk impact of meteorological conditions and complex road shape, and can respond to changes in the wet and slippery road surface in a timely manner, effectively improving the timeliness of risk warning and driving safety of vehicles in special road sections such as wet and slippery and winding road sections.

[0013] (2) This invention collects real-time vehicle driving status data and correlates and matches it with comprehensive road risk value to generate a unique risk probability for the target vehicle. By combining road environment risk with vehicle driving status data, it realizes a comprehensive assessment of the coupling effect of bad driving behavior and complex road conditions, thereby providing a more comprehensive and personalized risk warning and enhancing the accuracy and initiative of highway driving safety protection. Attached Figure Description

[0014] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 This is a diagram illustrating the implementation steps of the method of the present invention.

[0016] Figure 2 This is a flowchart of the curve continuity identification and curve group aggregation process of the present invention.

[0017] Figure 3 This is a flowchart illustrating the real-time road surface humidity calculation process of this invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Please see Figure 1 As shown, the present invention provides a method for monitoring and early warning of sudden accidents on highways based on road condition characteristics, including: S1, real-time acquisition of rainfall conditions and road shape data of the target vehicle's current location and the next road segment of its travel route, the road shape data including the curvature amplitude value of the current location and the next road segment, the curve continuity indicator and the number of tunnels.

[0020] The above rainfall conditions include regional rainfall intensity and duration, and the steps for obtaining rainfall conditions are as follows: the current location information of the target vehicle and the road boundary coordinates of the next road segment are defined as the data collection area.

[0021] Based on the highway number, lane, and direction of travel of the target vehicle, and combined with the highway stored in the electronic map database, the extended road segment of the current travel segment is matched as the next road segment.

[0022] Rainfall data at each monitoring point is collected in real time through multiple meteorological monitoring stations deployed along the data collection area.

[0023] Spatial interpolation is performed on discrete rainfall data collected from various meteorological monitoring stations to generate regionalized rainfall intensity distribution data covering the continuous space of the next road segment.

[0024] It should be noted that if the rainfall intensity of a single or a few discrete points is used to represent the rainfall situation of the entire road section, there will be a bias of substituting a point for the whole area. For example, if a monitoring station is located in an area with low rainfall, but the data is used to calculate the humidity of the entire road section, the calculated humidity value of the area with high actual rainfall will be lower. Spatial interpolation is used to reduce the probability of bias.

[0025] The continuous monitoring of rainfall status in the data collection area, from the start of rainfall to the current moment, is recorded as the duration of rainfall.

[0026] If a rainfall interruption is detected, and the interruption duration does not exceed the set interruption duration, the sum of the rainfall duration before the interruption and the rainfall duration after the resumption will be recorded as the rainfall duration. If the interruption duration exceeds the set interruption duration, the rainfall duration will be reset, and the timer will be updated when the rainfall resumes.

[0027] It should be noted that the set interruption duration is based on historical monitoring data of rainfall interruption and road surface moisture changes in different seasons and initial rainfall intensities on the next section of the target highway. Combined with the physical law that road surface moisture decays to a threshold that has no substantial impact on road risk correction, the statistical average value is used as the base value. If a high-risk warning is still issued after a long period of rainfall interruption when the road surface has dried, the base value is appropriately shortened and recorded as the set interruption duration.

[0028] When the duration of rainfall interruption does not exceed the set value, although there is no new rainfall to replenish the road surface temporarily, the road surface moisture formed by the previous rainfall has not yet significantly decreased. At this time, the rainfall duration before and after the interruption is accumulated and calculated to truly reflect the continuous effect of rainfall on the road surface, avoiding the misjudgment of the short-term interruption as the end of rainfall, which would cause a sharp drop in the calculated moisture value and a disconnect from the actual road surface condition.

[0029] When the duration of rainfall interruption exceeds the set value, the road surface humidity will decrease significantly due to factors such as natural evaporation and road surface drainage. At this time, the cumulative effect of rainfall has become invalid. Resetting the duration of rainfall and re-timing when rainfall resumes can avoid including invalid historical rainfall duration in the new humidity calculation and prevent the humidity calculation value from being too high due to invalid historical data.

[0030] The steps for obtaining the above-mentioned curve curvature amplitude value are as follows: extract the continuous geometric coordinate sequence of the centerline of the next road segment from the electronic map database.

[0031] Based on the geometric coordinate sequence, the start and end points of all planar curves in the road segment are identified, and the road units between adjacent start and end points are marked as an independent curve.

[0032] Based on the geometric coordinate sequence of an independent curve arranged in the direction of travel, three adjacent coordinate sequences are selected in sequence. When the three points are not collinear, their circumcircle is determined based on the three points, and the radius of the circle is used as the instantaneous radius of curvature of the intermediate sampling point. When the three points are collinear, the radius of curvature of the point is determined to be infinite and marked as a straight line segment.

[0033] Calculate the arithmetic mean of all instantaneous radii of curvature on an independent curve to obtain the average radius of curvature of the curve, and use this value as the curvature amplitude of the curve.

[0034] See Figure 2 As shown, the steps for obtaining the curve continuity markers are as follows: Select two adjacent independent curves along the driving direction as the preceding curve and the following curve, respectively.

[0035] The length of the straight section between the two curves is obtained by calculating the difference between the end point of the preceding curve and the beginning point of the following curve.

[0036] If the straight segment meets the condition of three points being collinear and there are no other road structures separating the two curves, then the preceding curve and the following curve are determined to be in a continuous and related state.

[0037] Repeat the above operation for all adjacent independent curves and aggregate curves with consecutive associated states into a curve group.

[0038] Count the number of curve groups and the number of independent curves in each group, and record these two sets of data as curve continuity indicators.

[0039] The number of tunnels mentioned above is based on statistics from the highway electronic map database.

[0040] S2. Obtain the initial dryness data of the road surface of the next road segment, and calculate the real-time road surface humidity level of the corresponding road segment in combination with the rainfall conditions.

[0041] See Figure 3 As shown, the calculation steps for the road surface humidity data are as follows: obtain the initial dry state data of the next road segment, and read the regionalized rainfall intensity and duration of rainfall in the data collection area in real time.

[0042] If the regionalized rainfall intensity is greater than zero, it is determined to be a rainfall state; otherwise, it is determined to be a no-rainfall state.

[0043] Under rainfall conditions, the cumulative effect value of rainfall is calculated based on the product relationship between regional rainfall intensity and rainfall duration. Combined with the initial dry state data, real-time road surface humidity data is obtained by humidity superposition calculation.

[0044] The formula for calculating the above real-time road surface humidity data is as follows: .

[0045] in, This represents the real-time road surface humidity value, with a value range of 0 to 1. Indicates the saturation humidity of the road surface. The initial dry state data is represented by R, the regionalized rainfall intensity is represented by T, the duration of rainfall is represented by k, the humidity absorption coefficient is represented by e, and the natural constant is represented by e.

[0046] The specific explanation of the above formula is as follows: This indicates the potential range of humidity change for the road surface from its initial dry state to reaching saturation humidity, which is the maximum theoretical space in which the road surface can still absorb moisture.

[0047] This demonstrates that as rainfall intensity, duration, and road surface moisture absorption coefficient all play a role, the process of road surface moisture approaching saturation is not linear, but rather exhibits an exponential trend. The negative sign in the exponential part indicates that as time and other factors increase, the remaining space for approaching saturation decreases exponentially.

[0048] This indicates the remaining amount of moisture on the road surface that has not yet reached saturation after considering the absorption characteristics of the road surface material, rainfall intensity, and duration. In other words, it represents how much moisture is still away from saturation after rainfall, starting from the initial dry state.

[0049] Real-time road surface humidity data is obtained by calculating humidity decay based on historical humidity data before the rainfall interruption and the duration of the interruption, under conditions of no rainfall.

[0050] The formula for calculating the above real-time road surface humidity data is as follows: .

[0051] in, This represents real-time road surface humidity data. This indicates the historical value of road surface moisture before the rainfall stopped. Represents the evaporation attenuation coefficient. Indicates the duration of the absence of rainfall.

[0052] The specific explanation of the above formula is as follows: This represents the nonlinear process by which the road surface moisture decreases from a wet state when rainfall ceases to a dry state when there is no rainfall.

[0053] If rainfall resumes after an interruption, the duration of rainfall will be reset and the humidity accumulation calculation will be restarted.

[0054] It should be noted that after rainfall stops, the moisture on the road surface will evaporate. If the interruption is long, the cumulative effect of the previous rainfall will be greatly reduced or even disappear. Resetting the timer is to avoid overestimating the risk of slippery road surfaces.

[0055] S3. Determine the risk level based on road shape data, generate a basic road risk index, and adjust the basic road risk index based on road surface moisture level to generate a comprehensive road risk value.

[0056] The specific steps for generating the basic road risk index are as follows: If the curvature amplitude value of the curve is greater than its historical average value, and the number of curves in the curve continuity identification group is less than the preset number of groups, it is judged as a low risk level; otherwise, it is judged as a medium risk level and recorded as the first judgment value.

[0057] It should be noted that the greater the curvature amplitude, the sharper the individual curve, and theoretically the higher the risk of vehicle skidding and loss of control. If there are fewer consecutive curves, the driver has sufficient time to complete the continuous operation of deceleration, steering, and straightening, and the adjacent curves form a smooth transition due to the collinearity of the three points, reducing the risk of sudden changes in operation.

[0058] The preset number of groups is obtained by combining the minimum number of warnings for continuous curves in the road design specifications, the statistical conclusions of sudden changes in risk of continuous curves in the historical accident data of the next road segment, and then dynamically calibrating according to the actual characteristics of the road segment.

[0059] If any group of curves in the curve continuity indicator has a number of curves exceeding the preset number, and the number of tunnels is higher than its average, it is judged as a medium-risk level; otherwise, it is judged as a low-risk level and recorded as the second judgment value.

[0060] It should be noted that when a road section has both a dense cluster of curves and a high density of tunnels, the two risk factors can have a synergistic amplifying effect. For example, if a driver suddenly enters a tunnel after a series of curves, or if a tunnel exit is immediately followed by a curve, this combination will significantly increase operational complexity and environmental uncertainty, thereby raising the overall risk level.

[0061] In risk assessment models, a single factor may be classified as low risk, but when two factors occur simultaneously, the probability of an accident increases significantly, thus requiring an upgrade to medium risk.

[0062] If the curvature amplitude of a curve is greater than its historical average, and there is a curve group with more curves than the preset number of curve groups, and the number of tunnels is higher than its average, it is judged as a high-risk level; otherwise, it is judged as a medium-risk level and recorded as the third judgment value.

[0063] It should be noted that the greater the curvature, the greater the centrifugal force required for the vehicle to pass through, making it extremely easy for sideslip or rollover to occur, and requiring extremely high control over vehicle speed and steering wheel.

[0064] It deprives drivers of the time and space to recover and adjust between curves, leading to continuous mental stress and making them prone to operational errors or fatigue.

[0065] Sudden changes in light inside and outside tunnels affect visual adaptation, the space feels enclosed, and there are often crosswinds and changes in road surface materials, making them potential accident hotspots.

[0066] The first, second, and third judgment values ​​are assigned preset weight coefficients, and a basic road risk index is generated through weighted fusion calculation.

[0067] The process of allocating the preset weight coefficients is as follows: First, collect historical data of the next section of the target highway, including records of each risk scenario corresponding to the first judgment value, the second judgment value, and the third judgment value, as well as accident incidence rate and accident severity data for each scenario over the past 3-5 years, to ensure that the data covers road conditions in different seasons and during different traffic flow periods.

[0068] Next, the correlation coefficients between the risk scenarios corresponding to each judgment value and the accident incidence rate are calculated. For example, the Pearson correlation coefficient is used to analyze the correlation strength between the medium-risk scenario of the first judgment value, the medium-risk scenario of the second judgment value, and the high-risk scenario of the third judgment value and the accident incidence rate. The larger the absolute value of the correlation coefficient, the more significant the impact of the risk scenario corresponding to that judgment value on the occurrence of accidents. Subsequently, multiple linear regression analysis is used, with the accident incidence rate as the dependent variable and the quantified values ​​of each judgment value as independent variables to construct a regression model. The contribution of the first, second, and third judgment values ​​to road accident risk is determined by the coefficients of each independent variable output by the model. The larger the coefficient, the higher the risk contribution of the corresponding judgment value. Finally, the risk contribution of each judgment value is normalized by dividing the contribution of each judgment value by the sum of the contributions of the three judgment values ​​to obtain the corresponding weight coefficient, and the sum of the three weight coefficients is 1.

[0069] The specific steps for generating the comprehensive road risk value are as follows: Identify the humidity data distribution characteristics of the curve area, divide the curve and its adjacent road sections into the curve humidity influence zone, identify the humidity data distribution characteristics of the tunnel entrance and exit area, divide the tunnel and its connecting road sections into the tunnel humidity influence zone, and divide the straight road sections into the general road section humidity zone.

[0070] It should be noted that insufficient lateral friction leading to water film slippage requires vehicles to provide centripetal force in curves, placing extremely high demands on the tires' lateral grip. When the road surface is wet or even has a water film, lateral friction drops sharply, making skidding or fishtailing very likely. By dividing curves and adjacent road sections into separate zones, curvature-based correction coefficients can be applied to precisely amplify the humidity risk in the curve area.

[0071] Tunnels often suffer from poor ventilation and insufficient sunlight, leading to slow evaporation of moisture from the road surface and making it more prone to persistent slippery sections. Drastic changes in light at tunnel entrances and exits further amplify the risk of slippery surfaces. Designating tunnels and their connecting sections as independent areas allows for a separate assessment of the combined risks posed by these local microclimate and visual challenges.

[0072] On straight road sections, humidity risk is relatively singular, mainly manifesting as rear-end collision risk. This baseline risk area avoids averaging the high risks associated with curves and tunnels across the entire road section, thus reducing the degree of danger at critical points.

[0073] For the humidity-affected area of ​​curves, a graded correction coefficient is set according to the curvature amplitude value and continuity indicator. For the humidity-affected area of ​​tunnels, a graded correction coefficient is set according to the number of tunnels. For the humidity-affected area of ​​general road sections, a baseline correction coefficient is set.

[0074] The steps for setting the graded correction coefficients are as follows: If there is no curve group in the next road segment, the correction coefficient is set as the first correction coefficient.

[0075] It should be noted that if there is no curve group in the next road segment, it indicates that the curve distribution of the road segment is sparse or it is a straight road segment. The risk impact of the curve is relatively constant. Therefore, the correction coefficient for such areas is set as a basic first correction coefficient.

[0076] If there is a group of curves in the next road segment, the curvature amplitude values ​​of each curve in the group are arranged in order along the driving direction to form a curvature sequence. The standard deviation of the curvature sequence is calculated, and a second correction coefficient is set based on the first correction coefficient and the standard deviation of the curvature sequence.

[0077] It should be noted that the larger the standard deviation of the curvature sequence, the more frequently the driver needs to adapt to curves that are sometimes sharp and sometimes gentle. The higher the operating load and risk, the larger the standard deviation of the curvature sequence, and the greater the increase of the second correction factor compared to the first correction factor.

[0078] If the number of tunnels in the next road segment is lower than the preset tunnel density benchmark for the area, the third correction factor is set; otherwise, the fourth correction factor is set based on the third correction factor and the length of the connecting road segment.

[0079] It should be noted that the preset area tunnel density benchmark is a critical value set based on historical road condition data, accident statistics, and road design specifications of the target road section.

[0080] If the number of tunnels in the next section is lower than the preset tunnel density benchmark, then the risks brought by the tunnels, such as changes in visibility and spatial oppression, are considered to be at a normal level. Therefore, the correction factor for such tunnel areas is set as the third correction factor of a benchmark.

[0081] If the number of tunnels exceeds the preset benchmark, the road section is considered to have a high density of tunnels, resulting in a significant risk superposition effect. In this case, the correction coefficient needs to be increased based on the third correction coefficient, with the increase depending on the length of the road section connecting the tunnel entrance and exit. This is because frequent entry and exit from tunnels within a short distance will increase the visual adaptation burden and the complexity of driving operations.

[0082] Therefore, based on the ratio of the total length of connecting road sections in densely tunneled road sections to the total length of the road section, the third correction coefficient is adjusted upward. The higher the proportion of connecting road sections, the more obvious the risk amplification effect.

[0083] The basic road risk index is multiplied by the correction coefficient for each region to obtain the preliminary corrected risk value for each region.

[0084] Calculate the coverage length ratio of each humidity-affected zone in the next road segment, and use this ratio as a weight to weight and fuse the preliminary corrected risk values ​​of each zone to generate the comprehensive road risk value.

[0085] S4. Real-time collection of the vehicle's speed and lane change frequency.

[0086] The aforementioned driving speed is obtained in real time by the Beidou positioning system on the vehicle, which acquires the vehicle's continuous geographical coordinates. The positioning coordinates of two adjacent moments are recorded within a preset short time interval. The straight-line distance traveled by the vehicle within that time interval is calculated using the difference in latitude and longitude, and then the speed is calculated using the speed formula.

[0087] The above lane change frequency acquisition steps are as follows: real-time acquisition of the lateral displacement data and lane line recognition information of the target vehicle; determination of the lane number of the vehicle's current lane based on the lane line recognition information; and recording of the lane number sequence.

[0088] The system continuously monitors changes in the lane number sequence. When a lane number change is detected, it is determined as a lane change event, and the lane change timestamp is recorded.

[0089] Set a time interval for lane change statistics, and count the cumulative number of lane change events that occur within the time interval.

[0090] The ratio of the cumulative number of lane change events to the duration of the time interval is the lane change frequency value per unit time.

[0091] S5. Perform correlation analysis between the comprehensive road risk value and driving speed and lane change frequency to calculate the accident risk probability of the target vehicle in the next road segment, issue warnings based on the accident risk probability classification, and push them to the target vehicle.

[0092] The steps for calculating the above-mentioned specific risk probability are as follows: Calculate the speed deviation rate between the target vehicle's speed and the speed limit standard of the next road segment.

[0093] Calculate the speed risk amplification factor and the lane-changing risk amplification factor respectively according to the comprehensive road risk value, the absolute value of the speed deviation rate, and the lane-changing frequency.

[0094] Calculate and perform normalization processing based on the speed risk amplification factor, the lane-changing risk amplification factor, and the reference value to obtain the accident risk probability.

[0095] The above speed risk amplification factor , where A represents the comprehensive road risk value, represents the absolute value of the speed deviation rate, e represents the natural constant. When the comprehensive road risk or speed deviation is large, the speed risk amplification factor will increase sharply, emphasizing the amplification effect of speed deviation under high-risk road conditions.

[0096] The above lane-changing risk amplification factor , where f represents the lane-changing frequency, e represents the natural constant, amplifying the influence of the lane-changing frequency on the road risk value, and reflecting the contribution of unstable driving behavior to the accident probability.

[0097] The above accident risk probability , where B represents the reference value, which is a preset constant.

[0098] When S + L > B, (S + L - B) > 0, indicating that the risk is higher than the normal level; when S + L < B, (S + L - B) < 0, indicating that the risk is lower than the normal level. (S + L - B) represents integrating the three key information of road environment risk, vehicle speed risk, and driving behavior risk into a comprehensive risk score . This score is a real number without upper or lower limits, and the larger the value, the higher the risk.

[0099] Finally, map an arbitrary real number (S + L - B) to a value between 0 and 1.

[0100] The accident risk probability calculation formula is a logistic regression function, whose function is to map the comprehensive risk score (S + L) to a probability value P between 0 and 1. The role of the reference value B is to define the critical threshold for high and low risks.

[0101] The value of the reference value B is not fixed, but is obtained through statistical learning based on the historical accident data of the target highway section. In this embodiment, it can be obtained through the following method: collect a large number of historical operation data samples on this section during a specific period. Each sample should include the comprehensive risk score (S + L) calculated at a specific time point and specific location, and mark whether an accident occurred subsequently for this sample; use statistical methods (such as finding the inflection point of the (S + L) distribution of accident and non-accident samples in historical data) to determine the optimal reference value B, that is, this value has the highest prediction accuracy for historical data.

[0102] For example, analysis of historical data from a mountainous highway section revealed that a statistically significant threshold of approximately 3 for (S+L) indicates a high risk of accidents. Therefore, the initial value of the baseline B can be set to 3. In practical applications, the value of B can be dynamically optimized and calibrated periodically based on new data accumulated after the system's operation.

[0103] Therefore, (S+LB) represents the deviation of the current comprehensive risk score relative to the historical benchmark level. This deviation is converted into an intuitive accident risk probability through a logical function, which facilitates graded early warning.

[0104] The parameters involved in the above formula are all dimensionless and calculated numerically. The formula is a formula obtained from the most recent real situation by collecting a large amount of data and simulating it with software. The preset parameters in the formula are set by those skilled in the art according to the actual situation.

[0105] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0106] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0107] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0108] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0109] Finally, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for monitoring and early warning of sudden accidents on highways based on road condition characteristics, characterized in that: include: S1. Real-time acquisition of rainfall conditions and road shape data for the next segment of the target vehicle's current location and driving route. The road shape data includes the curvature amplitude of curves at the current location and the next segment, curve continuity indicators, and the number of tunnels. S2. Obtain the initial dryness data of the road surface of the next road segment, and calculate the real-time road surface humidity level of the corresponding road segment in combination with the rainfall conditions; S3. Determine the risk level based on road shape data, generate a basic road risk index, and correct the basic road risk index by combining the road surface moisture level to generate a comprehensive road risk value. S4. Real-time collection of the vehicle's current speed and lane change frequency; S5. Perform correlation analysis between the comprehensive road risk value and driving speed and lane change frequency, calculate the accident risk probability of the target vehicle in the next road segment, issue warnings according to the accident risk probability level, and push them to the target vehicle. The steps for obtaining the curve continuity identifier are as follows: Select two adjacent independent curves sequentially along the driving direction as the preceding curve and the following curve, respectively; calculate the difference between the end point of the preceding curve and the starting point of the following curve to obtain the length of the straight segment between the two curves; if the straight segment does not meet the condition of three points being collinear and there is no other road structure separating the two curves, then the preceding curve and the following curve are determined to be in a continuous associated state; repeat the above operation for all adjacent independent curves and aggregate the curves in the continuous associated state into curve groups; count the number of curve groups and the number of independent curves in each group, and record the two sets of data as curve continuity identifiers. The specific steps for generating the basic road risk index are as follows: If the curvature amplitude value of a curve is greater than its historical average, and the number of curves in the curve continuity marker is less than the preset number of groups, it is judged as a low-risk level; otherwise, it is judged as a medium-risk level and recorded as the first judgment value. If the curve continuity marker has a curve group with more curves than the preset number of groups and the number of tunnels is higher than its average, it is judged as a medium-risk level; otherwise, it is judged as a low-risk level and recorded as the second judgment value. If the curvature amplitude value of a curve is greater than its historical average, and the number of curves in the curve group exceeds the preset number of groups and the number of tunnels is higher than its average, it is judged as a high-risk level; otherwise, it is judged as a medium-risk level and recorded as the third judgment value. The first judgment value, the second judgment value, and the third judgment value are respectively assigned preset weight coefficients, and the basic road risk index is generated by weighted fusion calculation. The specific steps for generating the comprehensive road risk value are as follows: Identify the humidity data distribution characteristics of curve areas, classify curves and adjacent road sections as curve humidity influence zones, identify the humidity data distribution characteristics of tunnel entrance and exit areas, classify tunnels and connecting road sections as tunnel humidity influence zones, and classify straight road sections as general road humidity zones; for curve humidity influence zones, set graded correction coefficients based on curve curvature amplitude and curve continuity indicators; for tunnel humidity influence zones, set graded correction coefficients based on the number of tunnels; for general road humidity zones, set a baseline correction coefficient; multiply the basic road risk index by the correction coefficients for each region to obtain the preliminary corrected risk value for each region. Calculate the coverage length ratio of each humidity-affected zone in the next road segment, and use this ratio as a weight to weight and fuse the preliminary corrected risk values ​​of each zone to generate the comprehensive road risk value.

2. The method for monitoring and early warning of sudden accidents on highways based on road condition characteristics according to claim 1, characterized in that: The rainfall conditions include regional rainfall intensity and duration, and the steps for obtaining the rainfall conditions are as follows: The current location information of the target vehicle and the road boundary coordinates of the next road segment are defined as the data collection area; Rainfall data at each monitoring point is collected in real time through multiple meteorological monitoring stations deployed along the data collection area. Spatial interpolation is performed on discrete rainfall data collected from various meteorological monitoring stations to generate regionalized rainfall intensity distribution data covering the continuous space of the next road segment; The continuous time period from the start of rainfall to the current moment is recorded as the duration of rainfall, based on the continuous monitoring of the rainfall status in the data collection area. If a rainfall interruption is detected, and the interruption duration does not exceed the set interruption duration, the sum of the rainfall duration before the interruption and the rainfall duration after the resumption will be recorded as the rainfall duration. If the interruption duration exceeds the set interruption duration, the rainfall duration will be reset, and the timer will be updated when the rainfall resumes.

3. The method for monitoring and early warning of sudden accidents on highways based on road condition characteristics according to claim 1, characterized in that: The steps for obtaining the curvature amplitude value of the curve are as follows: Extract the continuous geometric coordinate sequence of the centerline of the next road segment from the electronic map database; Based on the geometric coordinate sequence, the start and end points of all planar curves in the road segment are identified, and the road units between adjacent start and end points are marked as an independent curve. Based on the geometric coordinate sequence of an independent curve arranged in the direction of travel, three adjacent coordinate sequences are selected in sequence. When the three points are not collinear, their circumcircle is determined based on the three points, and the radius of the circle is used as the instantaneous radius of curvature of the intermediate sampling point. When the three points are collinear, the radius of curvature of the three points is determined to be infinite and marked as a straight line segment. Calculate the arithmetic mean of all instantaneous radii of curvature on an independent curve to obtain the average radius of curvature of the curve, and use this value as the curvature amplitude of the curve.

4. The method for monitoring and early warning of sudden accidents on highways based on road condition characteristics according to claim 1, characterized in that: The calculation process for the road surface moisture level is as follows: Acquire the initial dryness data of the road surface of the next section, and read the regional rainfall intensity and duration of rainfall in the data collection area in real time; If the regionalized rainfall intensity is greater than zero, it is determined to be a rainfall state; otherwise, it is determined to be a no-rainfall state. Under rainfall conditions, the cumulative effect value of rainfall is calculated based on the product relationship between regional rainfall intensity and rainfall duration. Combined with the initial dry state data, the real-time road surface humidity level is obtained by humidity superposition calculation. In the absence of rainfall, the real-time road surface humidity level is obtained by calculating humidity decay based on historical humidity data before the rainfall interruption and the duration of the interruption. If rainfall resumes after an interruption, the duration of rainfall will be reset and the humidity accumulation calculation will be restarted.

5. The method for monitoring and early warning of sudden accidents on highways based on road condition characteristics according to claim 4, characterized in that: The steps for setting the graded correction coefficients are as follows: If there is no curve group in the next road segment, the correction factor is set to the first correction factor; If there is a group of curves in the next road segment, the curvature amplitude values ​​of each curve in the group are arranged in order along the driving direction to form a curvature sequence. The standard deviation of the curvature sequence is calculated, and a second correction coefficient is set based on the first correction coefficient and the standard deviation of the curvature sequence. If the number of tunnels in the next road segment is lower than the preset tunnel density benchmark for the area, the third correction factor is set; otherwise, the fourth correction factor is set based on the third correction factor and the length of the connecting road segment.

6. The method for monitoring and early warning of sudden accidents on highways based on road condition characteristics according to claim 1, characterized in that: The frequency acquisition steps for the channel change are as follows: Real-time acquisition of lateral displacement data and lane line recognition information of target vehicles; determination of the lane number of the vehicle's current lane based on the lane line recognition information; and recording of the lane number sequence. Continuously monitor changes in the lane number sequence. When a lane number change is detected, it is determined as a lane change event, and the lane change timestamp is recorded. Set a time interval for lane change statistics, and count the cumulative number of lane change events that occur within the time interval; The ratio of the cumulative number of lane change events to the duration of the time interval is the lane change frequency value per unit time.

7. The method for monitoring and early warning of sudden accidents on highways based on road condition characteristics according to claim 1, characterized in that: The probability analysis of the accident risk is as follows: Calculate the speed deviation rate between the target vehicle's speed and the speed limit standard of the next road segment; Based on the comprehensive road risk value, the absolute value of the speed deviation rate, and the lane change frequency, calculate the speed risk amplification factor and the lane change risk amplification factor respectively. The probability of accident risk is calculated and normalized based on the speed risk amplification factor, lane change risk amplification factor, and benchmark value.

Images