Risk-based dynamic control method and control device for highway lane markings

Abstract

This invention discloses a dynamic control method and device for highway lane markings based on risk assessment. It acquires historical traffic flow information and vehicle trajectory data from roadside equipment, extracts various feature parameters to construct historical data samples, and trains a vehicle conflict risk prediction model suitable for dynamically controlled lane sections based on these samples. The model's input is the feature parameters from the data samples, and its output is the ETTC (Electronic Time to Collision) at a later time. It then extracts current feature parameters and uses the vehicle conflict risk prediction model to predict the ETTC at the target prediction time. Based on the ETTC of all vehicles in the full sample at the target prediction time, it calculates the comprehensive safety assessment score of the dynamically controlled lane section at the target prediction time, and classifies the risk level of the dynamically controlled lane section according to the score. Finally, it implements corresponding control measures for the variable lane markings in the dynamically controlled lane section. This invention offers high accuracy and efficiency in highway control, and is safe and reliable.

Description

Technical Field

[0001] This invention relates to the field of vehicle safety control technology, and in particular to a method and device for dynamic control of highway lane markings based on risk assessment. Background Technology

[0002] With the continuous increase in the number of motor vehicles in my country, the traffic safety situation has become increasingly severe. Highways, due to factors such as large differences in vehicle speed, susceptibility to weather conditions, dangerous driving behaviors such as frequent lane changes, and complex road environments, are prone to traffic accidents and have high fatality rates. Highway entrances and exits are particularly high-risk areas for accidents. To ensure the normal operation of highways, improve operational efficiency, and reduce the accident rate, real-time monitoring and control measures must be implemented in high-accident areas of highways.

[0003] Current highway traffic control strategies primarily consider variable speed limit control, emergency lane management, and ramp control. However, highway lane markings have a single function and cannot be dynamically adjusted according to real-time traffic conditions. Although scholars both domestically and internationally have conducted research on the design of lane guidance markings, such as laying LED light strips around zebra crossings, lane dividers, and directional arrows at intersections, and synchronizing their colors with traffic lights, most of these studies on dynamic lane guidance markings focus on intersections. There is currently no research on applying dynamic lane markings to highways to achieve real-time dynamic vehicle control, which plays a crucial role in reducing traffic conflicts, improving traffic efficiency, and enhancing highway traffic safety. Furthermore, current vehicle monitoring and data collection are mostly based on single video surveillance or radar systems, which suffer from drawbacks such as low data accuracy and susceptibility to weather conditions. Therefore, while adding new control measures, optimizing data collection is also critical. Summary of the Invention

[0004] This invention provides a method and device for dynamic control of highway lane markings based on risk assessment, which achieves dynamic traffic safety management by predicting vehicle conflicts in real time.

[0005] To achieve the above technical objectives, the present invention adopts the following technical solution:

[0006] A risk assessment-based dynamic control method for highway lane markings includes:

[0007] S01. For sections with dynamic lane control, historical traffic flow information and vehicle trajectory data are obtained from roadside equipment, and environmental feature parameters, traffic flow feature parameters, and full sample vehicle feature parameters are extracted at each historical moment to construct historical data samples.

[0008] S02, Based on the historical data samples obtained in step S01, a vehicle conflict risk prediction model suitable for dynamic lane control road sections is trained. Its input data includes environmental feature parameters, traffic flow feature parameters and vehicle feature parameters at earlier times, and the output data is the vehicle collision time ETTC at later times.

[0009] S03, following the method of step S01, extract the environmental feature parameters, traffic flow feature parameters, and full sample vehicle feature parameters from several previous moments, and use the vehicle conflict risk prediction model obtained in step S03 to predict the vehicle collision time ETTC of the full sample vehicles at the target prediction time.

[0010] S04. Based on the vehicle collision time ETTC of all sample vehicles at the target prediction time obtained in step S03, calculate the comprehensive safety assessment index score of the dynamic lane control section at the target prediction time, and classify the risk level of the dynamic lane control section according to the comprehensive safety assessment index score.

[0011] S05. Based on the risk level determined in step S04, control the working mode of the variable lane markings in the dynamic lane control section.

[0012] Furthermore, the variable lane markings use illuminated smart road studs arranged longitudinally, with the longitudinal spacing between adjacent smart road studs adjusted according to the road speed limit. The working modes of the smart road studs are: green constant light, red constant light, and yellow flashing.

[0013] If the dynamic lane control section is a key control section, then strip-shaped light strips are further set at intervals between adjacent illuminated smart road studs. The working modes of the strip-shaped light strips are standby off and red constant on. When the working mode of the smart road stud is red constant on, the working mode of the strip-shaped light strips is also set to red constant on. Otherwise, the working mode of the strip-shaped light strips is standby off.

[0014] Furthermore, the extracted environmental feature parameters include: distance to the vehicle ahead, speed of the vehicle ahead, number of lanes; and / or,

[0015] The extracted traffic flow characteristic parameters include: traffic volume, average vehicle speed / standard deviation, proportion of different vehicle types; and / or,

[0016] The extracted vehicle feature parameters for each sample include: vehicle position coordinates, vehicle instantaneous speed, acceleration, vehicle type, and lane.

[0017] Furthermore, the method for calculating the vehicle collision time ETTC used as output data in the historical data sample is as follows:

[0018]

[0019] In the formula, L l O l V l Let L be the length of the preceding vehicle l, the center point of the preceding vehicle l at the time when the ETTC is to be calculated, and its velocity. f O f V f Let f be the length of the following vehicle f, and let f be the center point and speed of the following vehicle f at the ETTC time to be calculated.

[0020] Furthermore, in step S02, when training the vehicle conflict risk prediction model, three machine learning models based on random forest, support vector machine and artificial neural network are first used to fit the vehicle conflict risk prediction model. Then, the prediction capabilities of the three models are compared based on the model evaluation index, and the model with the best prediction capability is selected as the final vehicle conflict risk prediction model.

[0021] Furthermore, in step S04, a comprehensive safety assessment index is calculated by constructing an evaluation index system;

[0022] The evaluation index system is comprised of the following indicators:

[0023] (1) Number of high-risk traffic conflicts in a region (TC): the number of high-risk traffic conflicts that occur within a preset time period; where, if the collision time ETTC of a vehicle is less than the preset value, the vehicle is considered to have committed a high-risk traffic conflict.

[0024] (2) Regional Conflict Rate (TCR): The proportion of high-risk regional conflicts (TC) to the inbound traffic volume of dynamically controlled lane sections.

[0025] (3) Regional traffic conflict mean ETTC avg The average ETTC value of all vehicles involved in high-risk traffic conflicts within a dynamically controlled lane section;

[0026] (4) Regional Traffic Conflict Extreme Values ​​(ETTC) min The minimum ETTC value for all vehicles involved in a high-risk traffic conflict within a dynamically controlled lane section;

[0027] (5) Number of vehicles decelerating rapidly (DV): The number of vehicles whose real-time deceleration is less than the preset value in the dynamic lane control section.

[0028] Then, the entropy weight method is used to assign weights to the above indicators, and the TOPSIS method is used to calculate the comprehensive safety assessment index score of the dynamic lane control section.

[0029] Furthermore, in step S04, the risk level of the dynamic lane control section is divided according to the comprehensive safety assessment index. Specifically, if the comprehensive index score is in [0, 0.2], it is classified as "high risk"; if the score is in (0.2-0.5], it is classified as "medium risk"; if the score is in (0.5-0.8], it is classified as "low risk"; and when the score is in (0.8-1) or no conflict occurs, it is classified as "safe".

[0030] Furthermore, step S05 controls the operating mode of the variable lane markings according to the risk level, specifically as follows:

[0031] (1) When the risk level rises from “safe” to “low risk”, “medium risk” or “high risk”, the working mode of the variable lane markings in the dynamic lane control section will be set to yellow flashing for a certain period of time and then turn to solid red, prohibiting vehicles from changing lanes;

[0032] (2) When the risk level is reduced from other levels to "safe", the working mode of the variable lane markings in the dynamic lane control section will be set to yellow flashing for a certain period of time and then turn into constant red, allowing vehicles to change lanes.

[0033] Furthermore, when the risk level is "high risk", all variable lane markings on the entire road section will be activated to flashing yellow or red lights;

[0034] When the risk level is "medium risk", the continuous section of the variable lane marking with a length of 0.75L will be activated with flashing yellow and red lights, while the remaining markings will be activated with green lights; where L is the total length of the dynamic lane control section.

[0035] When the risk level is "low risk", the continuous 0.5L section of the variable lane markings will be activated with flashing yellow and red lights, while the remaining markings will be activated with green lights.

[0036] When the risk level is "safe", all reversible lane markings on the entire road section will be activated to green.

[0037] A control device based on the risk assessment-based dynamic control method for highway lane markings as described above, comprising:

[0038] Data sample acquisition module: acquires historical data samples and real-time test data samples from roadside equipment; the method for roadside equipment to construct historical data samples and real-time test data samples is as follows: first, traffic flow information and vehicle trajectory data are acquired, and then environmental feature parameters, traffic flow feature parameters and full sample vehicle feature parameters are extracted at each time.

[0039] Vehicle Conflict Risk Prediction Model: Based on historical data samples, it is suitable for predicting conflict risks on dynamically controlled road sections. Its input data includes environmental characteristic parameters, traffic flow characteristic parameters, and vehicle characteristic parameters at earlier times, and the output data is the vehicle collision time ETTC at later times.

[0040] The risk level classification module is used to: calculate the comprehensive safety assessment index score of the dynamic lane control section at the target prediction time based on the predicted vehicle collision time ETTC of all sample vehicles at the target prediction time, and classify the risk level of the dynamic lane control section based on the comprehensive safety assessment index score.

[0041] The control measures execution module is used to implement corresponding control measures on the variable lane markings of the dynamic lane control section based on the risk level determined.

[0042] Beneficial effects

[0043] Compared with the prior art, the advantages of the present invention are as follows:

[0044] 1. This invention predicts the risk of vehicle conflicts in the controlled road segment in the future based on real-time monitoring and extraction of traffic flow, traffic environment, and vehicle parameters. It then assesses the overall safety of the controlled road segment and classifies its safety risk level based on the overall vehicle conflict risk. Based on the regional risk classification, it implements tiered dynamic lane marking control, achieving coordinated control with traffic flow safety and reducing safety hazards during driving. When the regional risk level reaches "low risk" or above, dynamic lane marking control measures are immediately implemented. Smart road studs within the dynamic control area flash yellow for 3-7 seconds to warn vehicles that lane changes are not permitted, then turn red, prohibiting lane changes. Simultaneously, the display screen at the front of the dynamic lane markings shows the changes in road markings ahead, giving drivers sufficient reaction time, improving traffic safety, reducing vehicle weaving behavior, and lowering the risk of conflict. When the regional risk level drops to "safe," the smart road studs flash yellow for 3-7 seconds, then turn green, allowing vehicles to change lanes, thus realizing intelligent dynamic changes in highway lane markings.

[0045] 2. In the dynamic safety management of lane markings, the smart road studs and strip luminous strips are fully and reasonably utilized, which increases the function of highway lane markings, maximizes their effectiveness, has low construction costs, can be deployed on a large scale according to highway markings, and is highly practical.

[0046] 3. The roadside equipment of the present invention adopts a radar-visual integrated machine, which is used in conjunction with smart road studs for vehicle monitoring. Compared with a single video surveillance system or radar system, the radar-visual integrated machine and smart road studs have a wide detection range, high data accuracy, all-weather operation, rich detection target types, and can support multi-functional applications. It provides excellent data support for traffic flow safety monitoring and highway safety management, and meets the requirements for real-time assessment of traffic safety.

[0047] 4. The entropy weight-TOPSIS method is used to construct a comprehensive safety evaluation index for the controlled road section area. Compared with TTC, PET, and ETTC, it is a more comprehensive and complete index. Among them, the entropy weight method has the characteristics of handling the uncertainty of weights and the strong correlation of criteria, while the TOPSIS method can better analyze the situation of multiple indicators. Therefore, the combination of the entropy weight and TOPSIS methods in this invention not only has high objectivity and scientificity, but also can more accurately assess the safety status of the controlled road section area. Attached Figure Description

[0048] Figure 1 This is a detailed flowchart illustrating the method of an embodiment of the present invention;

[0049] Figure 2 This is the technical route of the method in the embodiments of the present invention;

[0050] Figure 3 This is a schematic diagram of the deployment of the intelligent road studs and strip-shaped luminous strips in the merging area of ​​the dynamic lane markings in an embodiment of the present invention; Figure 3 (a) is a schematic diagram of the deployment of the intelligent road studs and strip luminous strips of the present invention when the traffic volume in the merging area is large. Figure 3 (b) is a schematic diagram of the deployment of the intelligent road studs and strip luminous strips of the present invention in the merging zone when the traffic volume is moderate. Figure 3 (c) is a schematic diagram of the deployment of the intelligent road studs and strip luminous strips of the present invention when the traffic volume in the merging zone is small.

[0051] Figure 4 This is a schematic diagram of the dynamic display board according to an embodiment of the present invention;

[0052] Figure 5 This is a schematic diagram of the dynamic lane marking sign described in an embodiment of the present invention. Detailed Implementation

[0053] The embodiments of the present invention will be described in detail below. These embodiments are based on the technical solutions of the present invention and provide detailed implementation methods and specific operation processes to further explain the technical solutions of the present invention.

[0054] Example 1

[0055] This embodiment provides a dynamic control method for highway lane markings based on risk assessment, referencing... Figure 1 As shown, it includes:

[0056] S01. For sections with dynamic lane control, historical traffic flow information and vehicle trajectory data are obtained from roadside equipment, and environmental feature parameters, traffic flow feature parameters, and full sample vehicle feature parameters are extracted at each historical moment to construct historical data samples.

[0057] (1) Dynamic lane control sections

[0058] The dynamic lane control sections are selected based on road design, accident data, or control experience, using the frequent occurrence of vehicle weaving, vehicle accidents, or traffic congestion as criteria. These sections typically include highway on / off ramps, sections with changing lane numbers, and sections entering and exiting toll stations.

[0059] In sections with dynamic lane control, the lane markings that need to be set up for dynamic control are determined based on the number of lanes, the geometric characteristics of the section, and the driving characteristics of the vehicles.

[0060] Furthermore, smart road studs are deployed along the lane dividers of selected highway dynamic lane control sections. Based on existing industry standards for road studs (raised pavement markers), the longitudinal spacing of the smart road studs in this invention can be adjusted according to road speed limits, with the total deployment length equal to the length of the controlled section. Each smart road stud includes a light-emitting module in green, yellow, and red colors. It features two flashing modes: constant illumination and strobe mode. Green and red studs use the constant illumination mode, while yellow uses the strobe mode. When no warning event occurs, the smart road stud normally illuminates in green, allowing vehicles to change lanes. When the conflict prediction value reaches the risk threshold, a safety warning and dynamic lane control are activated. The smart road stud flashes yellow for 3-7 seconds, warning vehicles that lane changing is imminent; subsequently, the color changes to red, prohibiting lane changing.

[0061] Furthermore, for key controlled road sections, long strips of light are installed between the smart road studs, emitting red light. When the smart road studs turn red, the light strips also light up, visually forming a constantly lit red lane marking, increasing the control effect, and vehicles are prohibited from changing lanes at this time.

[0062] Furthermore, the deployment of smart road studs and strip-shaped illuminated belts in controlled road sections can be adjusted appropriately according to traffic volume. Taking ramps as an example, when traffic volume is relatively low, below 1000 pcu / h, only road studs are deployed in the controlled road section, with a longitudinal spacing of 3m between upstream and downstream road studs, and a longitudinal spacing of 1.5m between road studs in weaving areas. Figure 3(c) As shown; when the traffic volume is moderate between 1000-2000 pcu / h, road studs are installed at a spacing of 3m upstream and downstream of the controlled road section, and at a spacing of 1.5m in the weaving area. Simultaneously, strip-shaped luminous strips are installed between the road studs in the weaving area to enhance the control effect, such as... Figure 3 (b) As shown; when the traffic volume is high, exceeding 2000 pcu / h, road studs are installed on the controlled road sections, with strip-shaped luminous strips between each stud. The spacing between upstream and downstream studs is 3m, and the spacing between studs in weaving areas is 1.5m. Figure 3 As shown in (a).

[0063] Furthermore, roadside dynamic electronic displays are installed before dynamic lane markings to provide reminders about dynamic lane control. The placement location is determined based on the road speed limit to ensure drivers have sufficient reaction time (20 seconds) after seeing the information; electronic displays are installed 700 meters before sections with a speed limit of 120 km / h; and 500 meters before sections with a speed limit of 80 km / h.

[0064] Furthermore, the electronic display screen displays the message "Dynamic lane markings ahead, please pay attention to the marking color," to remind the driver to notice the dynamic changes in road markings ahead and thus make the correct lane change. Figure 4 As shown.

[0065] Furthermore, static signs will be placed every 500 meters on both sides of the controlled road section, with the following information: solid green road stud: lane changing is permitted; flashing yellow road stud: lane changing will soon be prohibited; solid red road stud: lane changing is prohibited. Figure 5 As shown.

[0066] (2) Obtain traffic flow information and vehicle trajectory data

[0067] Roadside equipment includes sensing devices and edge computing units. Sensing devices include underground loop detectors, cameras, and integrated radar / visual detectors, capable of real-time sensing of traffic elements such as vehicle position, speed, and traffic flow at sub-second levels, with a sensing interval of 0.1 seconds. The edge computing unit preprocesses the data obtained from the sensing devices, extracting vehicle trajectories with an accuracy exceeding 90%, and collects traffic flow data for the controlled road segment every 5 seconds.

[0068] (3) Extract feature parameters

[0069] The edge computing unit in the roadside equipment calculates characteristic parameters related to traffic conflicts, including key parameters such as traffic volume, individual vehicle speed, acceleration, steering angle, headway, and traffic conflict ETTC between vehicles.

[0070] The data samples constructed using various extracted feature parameters include vehicle feature parameters (vehicle position coordinates, instantaneous vehicle speed, acceleration, vehicle type, lane, etc.), environmental feature parameters (distance to the vehicle ahead, speed of the vehicle ahead, number of lanes, etc.), and traffic flow feature parameters (traffic volume, average vehicle speed / standard deviation, proportion of different vehicle types, etc.). When constructing the historical data samples, the original data duration included one hour each of morning and evening peak hours and off-peak hours, totaling five hours of data.

[0071] In addition, when constructing historical data samples, it is also necessary to calculate the vehicle collision time used for output data to determine whether a collision occurs at a later time. The calculation formula is as follows:

[0072]

[0073] In the formula, L l O l V l Let L be the length of the preceding vehicle l, the center point of the preceding vehicle l at the time when the ETTC is to be calculated, and its velocity. f O f V f Here, represents the length of the following vehicle f, its center point, and its speed at the time when ETTC is to be calculated. In this embodiment, ETTC is used as a proxy index to estimate the collision, calculating whether a collision will occur if the vehicle maintains the given relative speed and position at this location; if the collision occurs, it will occur within the threshold.

[0074] S02, based on the historical data samples obtained in step S01, train a vehicle conflict risk prediction model suitable for dynamic lane control road sections. Its input data includes environmental feature parameters, traffic flow feature parameters and vehicle feature parameters at earlier times, and the output data is the vehicle collision time ETTC at later times.

[0075] In this embodiment, the ETTC value of each vehicle at time t is used as the prediction quantity, and the vehicle characteristic parameters, environmental characteristic parameters, and traffic flow characteristic parameters at time t-5 seconds or t-10 seconds are used as input independent variables. Three machine learning models based on random forest (RF), support vector machine (SVM), and artificial neural network (ANN) are used to fit the real-time vehicle risk prediction model. To prevent the model from overfitting, confusion matrix and AUC curve are used to measure the accuracy of the prediction model. Based on multiple evaluation indicators such as accuracy, recall, true positive rate, sensitivity, and AUC value, the overall prediction ability of the three models is compared, and the optimal model is selected as the final vehicle conflict risk prediction model.

[0076] S03, following the method in step S01, extract the environmental feature parameters, traffic flow feature parameters, and full sample vehicle feature parameters from several previous moments, and use the vehicle conflict risk prediction model obtained in step S02 to predict the vehicle collision time ETTC of the full sample vehicles at the target prediction time.

[0077] Based on real-time data perception, the required input parameters in the vehicle conflict risk prediction model are obtained, and the model prediction value is calculated at time T, that is, the ETTC value of each vehicle at T+5 seconds or T+10 seconds.

[0078] S04. Based on the vehicle collision time ETTC of all sample vehicles at the target prediction time obtained in step S03, calculate the comprehensive safety assessment index score of the dynamic lane control section at the target prediction time, and classify the risk level of the dynamic lane control section according to the comprehensive safety assessment index score.

[0079] (1) Selecting safety assessment indicators

[0080] In this embodiment, five regional safety assessment indicators are selected to construct an evaluation indicator system. The time interval for determining vehicle status is 0.1 seconds, and the indicators are extracted and aggregated every 5 seconds. The indicator definitions and calculation methods are as follows:

[0081] ① Regional High-Risk Conflict Count (TC): This refers to the number of high-risk traffic conflicts that occur within a preset time period. Specifically, if the collision time-to-traffic-time (ETTC) of a vehicle is less than a preset value, the vehicle is considered to have engaged in a high-risk traffic conflict. In this invention, the criterion for determining whether a traffic conflict has occurred is whether the ETTC value is less than 4 seconds within the time period. If it is less than 4 seconds, it is considered a high-risk traffic conflict.

[0082] ② Regional Conflict Rate (TCR): The calculation formula is TCR = TC / MPCU, where MPCU represents the mixed traffic volume at the entrance. MPCU refers to the traffic volume equivalent to vehicular traffic volume when multiple traffic participants share the same single lane of road, calculated based on their impact on road capacity and traffic safety. This invention classifies vehicle types into cars, medium-sized vehicles, large vehicles, and trailers. The conversion factor is 1 for cars and small trucks, 1.5 for medium-sized trucks, and 3 for trailers. A higher regional traffic conflict rate indicates a higher incidence of traffic conflicts, meaning a higher driving risk on the road segment, and vice versa.

[0083] ③ Average Regional Traffic Conflicts (ETTC) avg The average ETTC value (in this example, the ETTC value within 0-4s) of all vehicles involved in high-risk traffic conflicts within the dynamically controlled lane section.

[0084] ④ Regional Traffic Conflict Extreme Values ​​(ETTC) minThe minimum ETTC value for all vehicles involved in a high-risk traffic conflict within a dynamically controlled lane section.

[0085] ⑤ Number of Vehicles Decelerating Rapidly (DV): The number of vehicles in a dynamically controlled lane section whose real-time deceleration is less than a preset value. Vehicles often engage in dangerous driving behaviors such as rapid deceleration due to special conditions ahead or driver error, which significantly impact traffic safety. Therefore, a fixed threshold method is used to identify vehicles decelerating rapidly, with a threshold of -2.78 m / s². 2 When the vehicle's real-time deceleration is less than this threshold, the vehicle is determined to be in a state of rapid deceleration driving behavior.

[0086] (2) Constructing a comprehensive security assessment index

[0087] By definition, the larger TC, TCR, and DV are, the more dangerous it is, while ETTC avg ETTC min The smaller the value, the more dangerous it is; therefore, TC, TCR, and DV are defined as negative indicators. ETTC avg ETTC min Defined as a positive indicator, the five evaluation indicators are weighted using the entropy weight method after the safety assessment indicator data is normalized.

[0088] The normalization formula is calculated as follows:

[0089]

[0090] In the formula, x ij That is, the value of the j-th index in the i-th sample, 0≤P ij ≤1.

[0091] Information entropy E of each indicator j Calculate using the following formula:

[0092]

[0093] Coefficient of difference G for each indicator j Calculate using the following formula:

[0094]

[0095] Determine the weights W of each indicator j :

[0096]

[0097] Finally, the TOPSIS comprehensive evaluation model with weighted adjustments is used to calculate the comprehensive safety assessment index score of the control area and quantify the degree of conflict risk. The risk level of the area is divided according to the index score. The comprehensive safety assessment index score ranges from 0 to 1 and is divided into 4 levels. The closer the comprehensive index score is to 0, the smaller the index indicates the greater the danger, and vice versa.

[0098] (3) Classification

[0099] The risk level of a region is classified according to the comprehensive safety assessment index score. A comprehensive index score in the range of [0, 0.2] is classified as "high risk", a score in the range of (0.2-0.5) is classified as "medium risk", a score in the range of (0.5-0.8) is classified as "low risk", and a score in the range of (0.8-1) or when no conflict occurs (in which case the comprehensive index value cannot be calculated) is classified as "safe".

[0100] Before calculating the actual scores and classifying the levels in step S04, steps (1), (2), and (3) above involve constructing an evaluation index system and formulating level classification rules using historical data samples. Afterward, the comprehensive safety assessment index score of the dynamically controlled road segment at the target prediction time can be calculated based on this evaluation index system, and the risk level of the dynamically controlled road segment can be classified according to the comprehensive safety assessment index score and the level classification rules.

[0101] S05. Based on the risk level determined in step S04, implement corresponding control measures for the variable lane markings in the dynamic lane control section.

[0102] Based on the regional risk classification assessment results, a tiered dynamic lane control system will be implemented:

[0103] (1) When the risk level rises from "safe" to "low risk", "medium risk" or "high risk", the variable lane markings in the dynamic lane control section will be set to flash yellow for 3-7 seconds and then turn red to prohibit vehicles from changing lanes. At the same time, the electronic display screen at the front of the dynamic lane markings will show the change of the road markings ahead, reminding drivers that changing lanes is prohibited in the controlled section;

[0104] (2) When the risk level is reduced from other levels to "safe", the variable lane markings of the dynamic lane control section will be set to flash yellow for 3-7 seconds and then turn on green, allowing vehicles to change lanes.

[0105] Furthermore, the length of the smart road stud section implementing dynamic control within the controlled road segment varies with different risk levels. L represents the total length of all nodes in the controlled road segment. Under different risk levels, the length of the road stud section implementing color-changing is specified as follows:

[0106] Comprehensive index score [0,0.2] (0.2-0.5] (0.5,0.8] (0.8,1] Risk level High risk Medium risk Low risk Safety Length of road sections subject to dynamic control (m) L 0.75L 0.5L 0

[0107] Furthermore, to ensure the effectiveness of dynamic control of road sections, electronic monitoring equipment is deployed on the controlled road sections to capture vehicles violating traffic rules and report the information to the traffic management department's information system for strict enforcement.

[0108] In a further embodiment, sample videos and data are re-collected at fixed intervals to update the database, dynamically revise the comprehensive safety assessment indicators and vehicle conflict risk prediction models for the controlled road sections, revise the model code, and implement a new round of dynamic lane marking control and early warning.

[0109] Example 2

[0110] This embodiment provides a control device based on the risk assessment-based dynamic control method for highway lane markings described in Embodiment 1, with reference to... Figure 2 As shown, it includes:

[0111] Data sample acquisition module: acquires historical data samples and real-time test data samples from roadside equipment; the method for roadside equipment to construct historical data samples and real-time test data samples is as follows: first, traffic flow information and vehicle trajectory data are acquired, and then environmental feature parameters, traffic flow feature parameters and full sample vehicle feature parameters are extracted at each time.

[0112] Vehicle Conflict Risk Prediction Model: Based on historical data samples, it is suitable for predicting conflict risks on dynamically controlled road sections. Its input data includes environmental characteristic parameters, traffic flow characteristic parameters, and vehicle characteristic parameters at earlier times, and the output data is the vehicle collision time ETTC at later times.

[0113] The risk level classification module is used to: calculate the comprehensive safety assessment index score of the dynamic lane control section at the target prediction time based on the predicted vehicle collision time ETTC of all sample vehicles at the target prediction time, and classify the risk level of the dynamic lane control section based on the comprehensive safety assessment index score.

[0114] The control measures execution module is used to implement corresponding control measures on the variable lane markings of the dynamic lane control section based on the risk level determined.

[0115] The above embodiments are preferred embodiments of this application. Those skilled in the art can make various changes or improvements based on them. Without departing from the overall concept of this application, these changes or improvements should fall within the scope of protection claimed in this application.

Claims

1. A dynamic control method for highway lane markings based on risk assessment, characterized in that, include: S01. For sections with dynamic lane control, historical traffic flow information and vehicle trajectory data are obtained from roadside equipment, and environmental feature parameters, traffic flow feature parameters, and full sample vehicle feature parameters are extracted at each historical moment to construct historical data samples. S02, Based on the historical data samples obtained in step S01, a vehicle conflict risk prediction model suitable for dynamic lane control road sections is trained. Its input data includes environmental feature parameters, traffic flow feature parameters and vehicle feature parameters at earlier times, and the output data is the vehicle collision time ETTC at later times. S03, following the method of step S01, extract the environmental feature parameters, traffic flow feature parameters, and full sample vehicle feature parameters from several previous moments, and use the vehicle conflict risk prediction model obtained in step S02 to predict the vehicle collision time ETTC of the full sample vehicles at the target prediction time. S04. Based on the vehicle collision time ETTC of all sample vehicles at the target prediction time obtained in step S03, calculate the comprehensive safety assessment index score of the dynamic lane control section at the target prediction time, and classify the risk level of the dynamic lane control section according to the comprehensive safety assessment index score. In step S04, a comprehensive safety assessment index is calculated by constructing an evaluation index system; The evaluation index system is comprised of the following indicators: (1) Number of high-risk traffic conflicts in a region (TC): the number of high-risk traffic conflicts that occur within a preset time period; where, if the collision time ETTC of a vehicle is less than the preset value, the vehicle is considered to have committed a high-risk traffic conflict. (2) Regional Conflict Rate (TCR): the proportion of high-risk regional conflicts (TC) to the inbound traffic volume of dynamically controlled lane sections; (3) Average regional traffic conflict The average ETTC value of all vehicles involved in high-risk traffic conflicts within a dynamically controlled lane section; (4) Regional traffic conflict extreme values The minimum ETTC value for all vehicles involved in a high-risk traffic conflict within a dynamically controlled lane section; (5) Number of vehicles decelerating rapidly (DV): The number of vehicles in the dynamic lane control section whose real-time deceleration is less than the preset value. Then, the entropy weight method is used to assign weights to the above indicators, and the TOPSIS method is used to calculate the comprehensive safety assessment index score of the dynamic lane control section. S05. Based on the risk level determined in step S04, control the working mode of the variable lane markings in the dynamic lane control section.

2. The method for dynamic control of highway lane markings based on risk assessment according to claim 1, characterized in that, The variable lane markings use illuminated smart road studs arranged longitudinally. The longitudinal spacing between adjacent smart road studs is adjusted according to the road speed limit. The working modes of the smart road studs are green constant light, red constant light, and yellow flashing. If the dynamic lane control section is a key control section, then a strip of light is further set between adjacent illuminated smart road studs. The working mode of the strip of light is divided into standby off and red constant light. When the working mode of the smart road stud is red constant light, the working mode of the strip of light is also set to red constant light. Otherwise, the working mode of the strip of light is standby off.

3. The method for dynamic control of highway lane markings based on risk assessment according to claim 1, characterized in that, The extracted environmental feature parameters include: distance to the vehicle ahead, speed of the vehicle ahead, number of lanes; and / or, The extracted traffic flow characteristic parameters include: traffic volume, average vehicle speed / standard deviation, proportion of different vehicle types; and / or, The extracted vehicle feature parameters for each sample include: vehicle position coordinates, vehicle instantaneous speed, acceleration, vehicle type, and lane.

4. The method for dynamic control of highway lane markings based on risk assessment according to claim 1, characterized in that, The method for calculating the vehicle collision time ETTC used as output data in the historical data sample is as follows: ； In the formula, , , The front car l The length of the vehicle and the vehicle preceding the time slot to be calculated for ETTC. l The center point and velocity, The following car f The length of the vehicle and the vehicle after the time to be calculated for ETTC. f The center point and velocity.

5. The method for dynamic control of highway lane markings based on risk assessment according to claim 1, characterized in that, In step S02, when training the vehicle conflict risk prediction model, three machine learning models—random forest, support vector machine, and artificial neural network—are first used to fit the vehicle conflict risk prediction model. Then, the prediction capabilities of the three models are compared based on the model evaluation index, and the model with the best prediction capability is selected as the final vehicle conflict risk prediction model.

6. The method for dynamic control of highway lane markings based on risk assessment according to claim 1, characterized in that, In step S04, the risk level of the dynamic lane control section is divided according to the comprehensive safety assessment index. Specifically, if the comprehensive index score is in [0, 0.2], it is classified as "high risk"; if the score is in (0.2-0.5], it is classified as "medium risk"; if the score is in (0.5-0.8], it is classified as "low risk"; and when the score is in (0.8-1] or no conflict occurs, it is classified as "safe".

7. The method for dynamic control of highway lane markings based on risk assessment according to claim 2, characterized in that, Step S05 controls the operating mode of variable lane markings according to the risk level, specifically as follows: (1) When the risk level rises from "safe" to "low risk", "medium risk" or "high risk", the working mode of the variable lane markings in the dynamic lane control section will be set to yellow flashing for a certain period of time and then turn to solid red, prohibiting vehicles from changing lanes; (2) When the risk level is reduced from other levels to "safe", the working mode of the variable lane markings in the dynamic lane control section is set to yellow flashing for a certain period of time and then turned to solid red, allowing vehicles to change lanes.

8. The dynamic control method for highway lane markings based on risk assessment according to claim 7, characterized in that, When the risk level is "high risk", all reversible lane markings on the entire road section will be activated to flashing yellow or red lights; When the risk level is "medium", the continuous section of the variable lane marking with a length of 0.75L will be activated with flashing yellow and red lights, while the remaining markings will be activated with green lights; where L is the total length of the dynamic lane control section. When the risk level is "low risk", the continuous 0.5L section of the variable lane markings will be activated with flashing yellow and red lights, while the remaining markings will be activated with green lights. When the risk level is "safe", all reversible lane markings on the entire road section will be activated to green.

9. A control device based on the risk assessment-based dynamic control method for highway lane markings according to any one of claims 1-8, characterized in that, include: Data sample acquisition module: acquires historical data samples and real-time test data samples from roadside equipment; the method for roadside equipment to construct historical data samples and real-time test data samples is as follows: first, traffic flow information and vehicle trajectory data are acquired, and then environmental feature parameters, traffic flow feature parameters and full sample vehicle feature parameters are extracted at each time. Vehicle Conflict Risk Prediction Model: Based on historical data samples, it is suitable for predicting conflict risks on dynamically controlled road sections. Its input data includes environmental characteristic parameters, traffic flow characteristic parameters, and vehicle characteristic parameters at earlier times, and the output data is the vehicle collision time ETTC at later times. The risk level classification module is used to: calculate the comprehensive safety assessment index score of the dynamic lane control section at the target prediction time based on the predicted vehicle collision time ETTC of all sample vehicles at the target prediction time, and classify the risk level of the dynamic lane control section based on the comprehensive safety assessment index score. The control measures execution module is used to implement corresponding control measures on the variable lane markings of the dynamic lane control section based on the risk level determined.

Images