Wind-induced driving risk driven traffic control method, system and device

Abstract

The application discloses a wind-induced driving risk driven traffic control method, system and equipment, relates to the field of traffic control, and the method comprises the following steps: based on a pre-constructed aerodynamic reduced-order model, determining aerodynamic six-component forces according to measured vehicle motion states and measured wind field data, loading the aerodynamic six-component forces into a vehicle multi-body dynamics reference model to obtain vehicle dynamics and kinematics responses, and updating the vehicle dynamics model; calculating a comprehensive instability risk index based on the updated vehicle dynamics model and the peak adhesion coefficient of the current road surface; constructing a dynamic risk field based on map information, obstacle information and the comprehensive instability risk index, then determining an optimal reference path and a maximum allowed safe speed curve in a future time domain to form a dynamic safety envelope; and determining a vehicle traffic control strategy for each lane and each vehicle type based on the dynamic safety envelope. The application realizes real-time and accurate quantitative evaluation and control of vehicle driving stability under the condition of multi-physical field coupling.

Description

Technical Field

[0001] This application relates to the field of traffic management, and in particular to traffic management methods, systems and equipment driven by wind-induced driving risks. Background Technology

[0002] On complex road sections such as cross-sea bridges, mountain passes, and tunnel entrances and exits, crosswind stability is a key factor affecting driving safety. However, current vehicle crosswind control and traffic management technologies still have significant technical bottlenecks in terms of environmental perception accuracy, vehicle-side response mechanisms, and the precision of control strategies, making it difficult to cope with extremely complex driving environments.

[0003] In terms of environmental wind field monitoring, existing traffic meteorological monitoring systems suffer from significant deficiencies in spatiotemporal resolution. Current wind speed data acquisition primarily relies on regional meteorological stations or sparsely distributed meteorological monitoring equipment along roads. These devices typically provide macroscopic meteorological data, such as average wind speed and prevailing wind direction over a 10-minute period. While such long-term statistical averages can reflect large-scale weather trends, they suffer from severe spatiotemporal lag, failing to capture the "instantaneous gusts" that pose the greatest threat to high-speed vehicles. More seriously, existing monitoring methods neglect abrupt changes in local flow fields within the road's micro-environment. In actual driving, vehicles encounter extremely strong "sudden wind changes" when passing bridge towers, sound barrier breaks, or tunnel exits; simultaneously, large vehicles generate strong aerodynamic wake interference during overtaking or passing. These local micro-wind field changes caused by infrastructure singularities (such as the bridge tower obstruction effect) and traffic flow dynamics (such as vehicle interactions) are major contributing factors to vehicle instability, but the current macro-meteorological monitoring system is almost entirely blind to these, unable to provide accurate local measured wind field data.

[0004] At the vehicle stability control level, existing onboard systems lack the ability to actively perceive and feedforward control of environmental loads. Current mass-produced vehicle electronic stability programs (ESP / Electronic Stability Control, ESC) primarily employ a passive, feedback-based control logic. The system relies on onboard gyroscopes and accelerometers to monitor changes in the vehicle's motion (such as yaw rate deviation or lateral acceleration), intervening only after the vehicle shows signs of physical instability. This reactive control mode often results in delayed vehicle attitude correction due to lag when facing strong crosswinds. Currently, vehicles generally lack sensor configurations capable of real-time perception of external aerodynamic loads (wind) and road surface adhesion (road), and it is difficult to perform high-precision fluid dynamics calculations to predict aerodynamic forces on onboard chips with limited computing power. Therefore, vehicles cannot achieve feedforward control based on environmental perception, meaning they cannot preemptively adjust the chassis state to counteract interference in the event of a gust of wind.

[0005] In terms of traffic control strategies, there is a lack of refined handling methods for the coupled conditions of "strong crosswinds + high vehicle speeds + low lateral adhesion." Existing traffic control on crosswind sections typically adopts a single-dimensional "threshold triggering" model. That is, a "one-size-fits-all" strategy of implementing uniform speed limits or road closures along the entire route based solely on the monitored natural wind speed. This extensive management model has two major flaws: First, it ignores the coupling effects of multiple physical fields. The risk of vehicle instability is not determined solely by wind speed, but is the result of multiple conditions acting together. On low-adhesion, slippery roads, even if the wind speed does not reach typhoon levels, high-speed vehicles are highly susceptible to skidding or overturning due to insufficient lateral adhesion. Existing single wind speed warnings cannot cover this high-risk, complex condition. Second, there is a lack of differentiated control capabilities based on lane and vehicle type. Different types of vehicles (such as high-center-of-gravity vans and low-center-of-gravity cars) have vastly different sensitivities to crosswinds, and the actual wind load on different lanes (windward and leeward sides) is also significantly different. Existing control measures cannot distinguish these differences, often resulting in efficiency losses such as "the entire road being closed due to the risk of a few high-risk vehicles," or ignoring the hidden instability risks of specific lanes or vehicle types on slippery roads under seemingly safe wind speeds. Summary of the Invention

[0006] The purpose of this application is to provide a traffic management method, system, and device driven by wind-induced driving risks, which enables real-time and accurate quantitative assessment and control of vehicle driving stability under multi-physics coupling conditions.

[0007] To achieve the above objectives, this application provides the following solution: Firstly, this application provides a traffic management method driven by wind-induced driving risks, applied to a vehicle-road cooperative edge computing architecture, wherein the vehicle-road cooperative edge computing architecture is deployed on a pre-defined high-risk road section, including: Collect measured vehicle motion status, measured wind field data, peak adhesion coefficient of the current road surface, map information, and obstacle information; Based on a pre-constructed aerodynamic reduced-order model, a high-dimensional fluid dynamics simulation task is performed according to the measured vehicle motion state and the measured wind field data to determine the six aerodynamic components; the six aerodynamic components are then loaded into the vehicle multibody dynamics reference model, and an aerodynamic-dynamic coupling solution task is performed to obtain the vehicle dynamics and kinematic response; the vehicle dynamics model is updated based on the vehicle dynamics and kinematic response; wherein, the aerodynamic reduced-order model is constructed by combining a deep neural network and an intrinsic orthogonal decomposition algorithm; Based on the updated vehicle dynamics model and the peak adhesion coefficient of the current road surface, a comprehensive instability risk index covering sideslip risk, rollover risk and lateral deviation risk is calculated. A dynamic risk field is constructed based on the map information, the obstacle information, and the comprehensive instability risk index. Based on the principle of minimizing potential energy, the optimal reference path and the corresponding maximum permissible safe speed curve in the future time domain are determined in the dynamic risk field to form a dynamic safety envelope. Based on the dynamic safety envelope, traffic control strategies for different lanes and vehicle types are determined.

[0008] Secondly, this application provides a traffic management system driven by wind-induced driving risks, applied to a vehicle-road cooperative edge computing architecture, wherein the vehicle-road cooperative edge computing architecture is deployed on a pre-defined high-risk road section, including: The data acquisition module is used to collect data on the actual vehicle motion status, actual wind field data, peak adhesion coefficient of the current road surface, map information, and obstacle information. The dynamics update module is used to perform a high-dimensional fluid dynamics simulation task based on the pre-built aerodynamic reduced-order model, according to the measured vehicle motion state and the measured wind field data, to determine the six aerodynamic components; load the six aerodynamic components into the vehicle multibody dynamics reference model, and perform an aerodynamic-dynamic coupling solution task to obtain the vehicle dynamics and kinematic response; update the vehicle dynamics model based on the vehicle dynamics and kinematic response; wherein, the aerodynamic reduced-order model is constructed by combining a deep neural network and an intrinsic orthogonal decomposition algorithm; The instability risk calculation module is used to calculate a comprehensive instability risk index covering sideslip risk, rollover risk and lateral deviation risk based on the updated vehicle dynamics model and the peak adhesion coefficient of the current road surface. The dynamic risk field construction module is used to construct a dynamic risk field based on the map information, the obstacle information, and the comprehensive instability risk index. The safety calculation module is used to determine the optimal reference path and the corresponding maximum permissible safe speed curve in the future time domain in the dynamic risk field based on the principle of minimizing potential energy, so as to form a dynamic safety envelope. The strategy generation module is used to determine vehicle traffic control strategies for different lanes and vehicle types based on the dynamic safety envelope.

[0009] Thirdly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement a traffic control method driven by wind-induced driving risks.

[0010] According to the specific embodiments provided in this application, the following technical effects are disclosed: This application is applied to a vehicle-road cooperative edge computing architecture, which is deployed on preset high-risk road sections, thereby combining roadside data with vehicle data. It collects measured wind field data to capture the instantaneous gust impacts that are often overlooked in existing technologies. This allows for the identification of "local strong crosswind risk points" at specific bridge towers or tunnel exits through subsequent processing, providing a reliable input source for subsequent precise control and fundamentally eliminating blind spots in environmental perception. This application, through a vehicle-road cooperative edge computing architecture and a POD aerodynamic reduction model, improves the spatiotemporal resolution of environmental perception to the millisecond and meter levels, ensuring the real-time performance and accuracy of the data.

[0011] This application introduces the peak adhesion coefficient of the current road surface and combines it with the vehicle dynamics model to calculate a comprehensive instability risk index covering sideslip risk, rollover risk, and lateral deviation risk, making the environmental risk assessment of this application environmentally adaptable. Especially in low adhesion conditions such as rain and snow, the peak adhesion coefficient can be used to pre-determine the vehicle's ability to withstand crosswinds, thereby intervening in advance before the vehicle sideslips, rolls over, or lateral deviations occur, achieving a leap from remediation to prevention, and is particularly suitable for multi-physics coupled conditions.

[0012] This application also constructs a dynamic risk field and determines the optimal reference path and corresponding maximum permissible safe speed curve in the future time domain to form a dynamic safety envelope. This helps maintain vehicle stability under various wind speeds and road conditions, avoiding overconfident driving behavior. Finally, based on the dynamic safety envelope, this application determines lane-specific and vehicle-type-specific traffic control strategies, avoiding a one-size-fits-all approach to road closures. Through refined proactive control based on lanes and vehicle types, it maximizes road capacity while ensuring zero accidents in extreme weather conditions. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 This is a diagram illustrating the overall logical architecture of a traffic control method driven by wind-induced driving risks in one embodiment of this application.

[0015] Figure 2 This is a schematic diagram of the vehicle-road cooperative edge computing architecture in this application.

[0016] Figure 3 This is a flowchart illustrating a traffic control method driven by wind-induced driving risks in one embodiment of this application.

[0017] Figure 4 This is a schematic diagram illustrating the calculation and application of the peak adhesion coefficient of the current road surface in one embodiment of this application.

[0018] Figure 5 This is a diagram of the real-time digital twin architecture of the wind turbine in this application.

[0019] Figure 6 This is a diagram of the multi-dimensional wind-driven vehicle stability architecture for this application.

[0020] Figure 7 This is a diagram of the spatiotemporal dynamic risk field architecture of this application.

[0021] Figure 8 This is a traffic control architecture diagram based on a risk field in this application.

[0022] Figure 9 This is a schematic diagram of the three-dimensional risk assessment surface of this application.

[0023] Figure 10 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0024] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0025] This application constructs a vehicle-road cooperative edge computing architecture, transferring high-load unsteady hydrodynamic calculations to the roadside. Utilizing roadside-measured microscopic wind field data and real-time road surface adhesion coefficients fed back from the vehicle, a real-time digital twin model integrating all elements of "wind-vehicle-road" is established in the digital space. Based on this, a unified spatiotemporal dynamic risk field and a three-dimensional risk assessment surface are constructed to accurately calculate the critical point of vehicle physical instability under the coupled effects of high wind speed (crosswind interference), high vehicle speed (dynamic limit approximation), and low adhesion (slippery road surface). This breaks the traditional "one-size-fits-all" control model, enabling differentiated proactive traffic control based on lane and vehicle type, maximizing road traffic efficiency while ensuring driving safety under extreme weather conditions.

[0026] In summary, this application provides a digital twin and refined control method for vehicle crosswind stability based on vehicle-road cooperative edge computing, which can perform real-time and accurate quantitative assessment and control of vehicle driving stability under multi-physics field coupling conditions of high wind speed, high vehicle speed, and low adhesion.

[0027] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0028] In one exemplary embodiment, such as Figure 1 As shown, a traffic management method driven by wind-induced driving risks is provided. The overall logical architecture of this method includes the following processing: construction based on vehicle-road cooperative edge computing architecture, real-time aerodynamic-dynamic digital twin inference at the roadside end, closed-loop parameter correction based on vehicle feedback, multi-dimensional comprehensive evaluation of wind-induced driving stability, construction of spatiotemporal dynamic risk field and safety envelope calculation, and refined traffic management based on three-dimensional risk surface.

[0029] Of the various processing methods mentioned above, the first one pertains to device-related processing, while the remaining five pertain to algorithm-related processing. For example... Figure 2 As shown, this application presents a vehicle-road cooperative edge computing architecture for vehicle crosswind stability control, which transfers high-load computing tasks from the vehicle to the roadside.

[0030] The Vehicle-to-Infrastructure (V2X-MEC) architecture is a distributed computing system framework. This architecture utilizes roadside edge computing units (MECs) to handle high-dimensional real-time fluid dynamics simulation tasks that would otherwise be difficult for onboard terminals to manage. Leveraging the low-latency 5G / C-V2X network, it enables real-time interaction between roadside-collected local measured wind field data and vehicle motion status uploaded from the vehicle, achieving a task offloading mode of "heavy computation on the roadside, light execution on the vehicle," thus ensuring real-time response to sudden crosswinds in high-speed driving scenarios.

[0031] The vehicle-road cooperative edge computing architecture is deployed on a pre-defined high-risk road section, and the setup process of the vehicle-road cooperative edge computing architecture includes: (1) When the target is a cross-sea bridge, the gradient wind speed in different return periods of the bridge site area is inverted based on the historical wind data of the bridge site area (which can be collected by the meteorological station in the bridge site area), and the wind speed at the bridge site area at the preset height is calculated in combination with the surface type. The preset height can be ten meters.

[0032] (2) Establish a calculation model of the topographic wind environment of the bridge site area or conduct a topographic wind tunnel experiment. Based on the wind speed and wind profile characteristics of the bridge site area at a preset height, conduct a study on the wind environment of the bridge site area and calculate the wind field characteristics simultaneously. Among them, the wind field characteristics include the bridge site location, wind profile coefficient, wind angle of attack and turbulence intensity, etc.

[0033] (3) Select the sections to be used based on the bridge type layout diagram and cross-section diagram of the entire route of the cross-sea bridge; among them, the sections to be used include sections with high clearance height and other representative sections, such as typical cross slope sections, sections with different cross-sectional heights, etc.

[0034] (4) Calculate the wind speed at the section to be used based on the wind field characteristics; wherein, the wind field characteristics are specifically the gradient wind speed at the bridge site.

[0035] (5) Conduct aerodynamic characteristic analysis of bridge traffic under different cross sections under calculated wind speed, combine dynamic analysis to determine the dangerous locations of bridges along the entire route, and mark them as the layout locations of the vehicle-road cooperative edge computing architecture.

[0036] See Figure 2 The vehicle-road cooperative edge computing architecture includes an on-board unit (i.e., the on-board terminal layer), a roadside unit (i.e., the roadside infrastructure layer), and a communication unit (i.e., the communication link layer). The roadside unit includes a roadside perception layer and a roadside computing layer. The roadside unit can be deployed at the bridge towers of cross-sea bridges, tunnel entrances / exits, and mountain passes of expressways, etc., in areas with high crosswind incidence.

[0037] The on-board unit is used to sense and collect the measured vehicle motion state, the peak adhesion coefficient of the current road surface, the vehicle side map information, and the vehicle side obstacle information in real time, and then upload them to the communication unit; wherein, the peak adhesion coefficient of the current road surface is obtained in real time by using the unscented Kalman filter of the chassis domain controller in the on-board unit combined with the Burckhardt tire model.

[0038] The roadside sensing layer is used to sense and collect measured wind field data, roadside map information, and roadside obstacle information in real time, and then upload them to the communication unit. Specifically, the roadside sensing layer includes a high-frequency ultrasonic anemometer array or Doppler wind lidar and traffic monitoring cameras arranged linearly along the road to acquire data such as absolute wind speed, wind direction, road geometry, and obstacles.

[0039] The communication unit is used to establish a bidirectional data transmission link between the roadside unit and the vehicle-mounted unit using a 5G or C-V2X communication module.

[0040] The roadside computing unit is deployed on an edge computing server (MEC) within a roadside cabinet. It integrates an unsteady fluid simulation engine for vehicle aerodynamics and a multibody dynamics solver. The roadside computing unit is used to: determine map information by combining vehicle-side map information with roadside map information; determine obstacle information by combining vehicle-side obstacle information with roadside obstacle information; and perform high-dimensional fluid dynamics simulation tasks, aerodynamic-dynamic coupling solution tasks for vehicle aerodynamics, and risk extrapolation and strategy generation tasks.

[0041] In practical applications, the vehicle traffic control strategies generated by the roadside computing unit are fed back to the vehicle-mounted unit. Therefore, the vehicle-mounted unit, as a mobile terminal, is also responsible for executing the received commands or instructions.

[0042] Through the above architecture, computationally intensive unsteady aerodynamic calculations and environmental risk simulations are transferred from the vehicle to the roadside.

[0043] like Figure 3 As shown, this application provides a traffic control method driven by wind-induced driving risks. This method is executed by computer equipment, specifically by a terminal or server alone, or by both a terminal and a server. When applied to the vehicle-road cooperative edge computing architecture described above, the method of this application is implemented by executing a computer program through the roadside computing layer, namely, steps 101 to 106.

[0044] Step 101: Collect the measured vehicle motion status, measured wind field data, peak adhesion coefficient of the current road surface, map information, and obstacle information.

[0045] Among them, the measured vehicle motion status, measured wind field data, map information, and obstacle information can all be directly collected by the on-board unit and roadside unit in the architecture described above. And as... Figure 4 As shown, the process for determining the peak adhesion coefficient of the current road surface is as follows: The vehicle's chassis domain controller utilizes locally mounted wheel speed sensors, accelerometers, and steering wheel angle signals to run a Unscented Kalman Filter (UKF) state observer. The UKF state observer uses longitudinal vehicle speed, lateral vehicle speed, and yaw rate as state variables, and combines them with the Burckhardt tire model to identify the road adhesion coefficient in real time. In the Burckhardt tire model, the composite adhesion coefficient... With synthetic slip ratio The relationship is expressed as follows: .

[0046] in, , , For the pavement characteristic parameters to be identified, initial values ​​can be set based on a standard pavement spectrum, and then adjusted according to the pavement adhesion coefficient. The identification process is adaptively corrected.

[0047] The unscented Kalman filter state observer minimizes the difference between the vehicle dynamics model predictions and the onboard sensor measurements (such as lateral acceleration). The residuals between (etc.) are used to iteratively update the current (etc.) value.

[0048] The identification obtained through the vehicle unit The value is transmitted back to the roadside unit in real time via the V2X network, thus achieving the acquisition of the current peak adhesion coefficient of the road surface. After receiving the data, the roadside unit will use it in subsequent steps. Value update of tire force constraint boundary of vehicle dynamics model ( This ensures that subsequent safety assessments and risk field construction are based on real road conditions. The aerodynamic disturbance forces calculated from the roadside can also be sent to the vehicle for feedforward control compensation.

[0049] Step 102, as follows Figure 5 As shown, based on the pre-constructed aerodynamic reduced-order model, a high-dimensional fluid dynamics simulation task is performed according to the measured vehicle motion state and the measured wind field data to determine the six aerodynamic components; the six aerodynamic components are loaded into the vehicle multibody dynamics reference model, and an aerodynamic-dynamic coupling solution task is performed to obtain the vehicle dynamics and kinematic response; the vehicle dynamics model is updated according to the vehicle dynamics and kinematic response.

[0050] The aerodynamic order reduction model is constructed by combining deep neural networks and intrinsic orthogonal decomposition (POD) algorithms. Specifically, the POD-based aerodynamic reduced order model refers to a mathematical model used to quickly obtain vehicle aerodynamic loads. Addressing the issue of long simulation times in traditional CFD simulations, this model utilizes POD technology to reduce the dimensionality of high-dimensional flow field data. Through an "offline high-precision training-online fast mapping" mechanism, it establishes the mapping relationship between vehicle motion states and the six nonlinear aerodynamic components within milliseconds. This model is the core algorithmic foundation for achieving digital twin simulation under high vehicle speed and strong crosswind conditions in this application.

[0051] In a practical application, the roadside server receives real-time status data uploaded by vehicles via an uplink communication link, i.e., the measured vehicle motion status, including: longitudinal vehicle speed.v x Vehicle transient attitude angles and vehicle ID; vehicle transient attitude angles include heading angle. lateral angle yaw rate Side slip angle The measured wind field data includes absolute wind speed, absolute wind direction, ambient lateral wind speed, and wind deflection angle. The ambient lateral wind speed refers to the component of the absolute wind speed perpendicular to the vehicle's direction of travel, and is synthesized from the absolute wind speed, absolute wind direction, and road direction.

[0052] like Figure 5 As shown, the process of constructing the aerodynamic reduced-order model includes: (21) Obtain a set of measured sample data; the measured sample data includes the measured sample vehicle motion state and the measured sample wind field data.

[0053] (22) Based on the vehicle ID in the measured sample vehicle motion state, call the vehicle geometric model and dynamic parameter set.

[0054] (23) Based on the measured sample vehicle motion state and the measured sample wind field data, calculate the composite airflow velocity and composite aerodynamic deflection angle.

[0055] Specifically, based on measured data from Doppler wind-measuring lidar or high-frequency ultrasonic anemometer arrays deployed in the roadside sensing layer, the effective local wind speed at the vehicle's current location is obtained through spatial interpolation or by accessing a pre-stored infrastructure local wind field correction database. Effective local wind direction and turbulence intensity The effective local wind speed, effective local wind direction, and vehicle motion vector are combined to calculate the current composite airflow velocity of the vehicle. Combined aerodynamic deflection angle The calculation formula is as follows: .

[0056] in, This is the time step for communication and computation. This step establishes the relative boundary conditions of the vehicle in the flow field.

[0057] (24) Based on the vehicle geometric model and dynamic parameter set, a high-dimensional aerodynamic sample space covering the entire operating envelope of the vehicle is constructed, and the synthetic airflow velocity and the synthetic aerodynamic deflection angle are used as boundary conditions.

[0058] In the real-time aerodynamic load calculation stage, in order to solve the physical bottleneck that traditional computational fluid dynamics (CFD) simulation is too time-consuming to meet the millisecond-level control response of vehicles, this application adopts an aerodynamic reduced-order model based on a combination of deep neural network and intrinsic orthogonal decomposition. The use of this model specifically includes two stages: offline data construction and model training.

[0059] In the offline phase, a high-dimensional aerodynamic sample space covering the entire vehicle operating envelope is constructed. The longitudinal vehicle speed, ambient lateral wind speed, synthetic aerodynamic deflection angle, and transient vehicle attitude angles (including body roll angle) are selected. Pitch angle and lateral angle ) as an independent design variable.

[0060] (25) In the high-dimensional aerodynamic sample space, batch CFD numerical simulations are performed using the unsteady Reynolds-averaged Navier-Stokes equations to obtain flow field snapshots and corresponding nonlinear aerodynamic six components under different working conditions.

[0061] Specifically, by employing Latin hypercube sampling or orthogonal experimental design methods, batch CFD numerical simulations are performed in a high-dimensional aerodynamic sample space using unsteady Reynolds-averaged Navier-Stokes (URANS) equations to obtain high-fidelity flow field snapshots and corresponding nonlinear aerodynamic six-component force data under different operating conditions, thereby establishing a high-precision original aerodynamic database.

[0062] (26) Introducing intrinsic orthogonal decomposition technology to reduce the dimension of the flow field snapshot and extract key physical features; the key physical features include the synthetic airflow velocity, the synthetic aerodynamic deflection angle and the vehicle transient attitude angle.

[0063] (27) Input the key physical features and the corresponding aerodynamic six components into a preset deep neural network model for feature learning and mapping training to obtain an aerodynamic reduced-order model.

[0064] During the training phase, machine learning algorithms are used to perform feature learning and model training on the data in the original aerodynamic database. A multi-dimensional state vector consisting of synthetic airflow velocity, synthetic aerodynamic deflection angle, turbulence intensity, wind attack angle, vehicle roll angle, pitch angle, yaw angle, and the longitudinal distance of the vehicle relative to the nearest infrastructure is used as input. The six aerodynamic forces (drag, lateral force, lift, roll moment, pitch moment, and yaw moment) are used as output. Intrinsic orthogonal decomposition (IOD) is introduced to perform modal decomposition on the flow field snapshots (pressure field, velocity field, etc.), extracting the main energetic modes to reduce data dimensionality, resulting in the input mentioned above. Then, a nonlinear regression mapping model is constructed by combining a backpropagation (BP) neural network or a radial basis function (RBF) network. By defining a mean squared error (MSE) loss function and using gradient descent for iterative training, the model can accurately capture the strongly coupled nonlinear relationship between vehicle attitude changes (especially unsteady effects caused by roll and pitch) and aerodynamic loads until the model's prediction accuracy on the validation set meets the preset convergence criterion.

[0065] The trained aerodynamic reduced-order model is encapsulated into a ROM model for real-time engineering applications. That is, during the online operation phase of the roadside unit, it receives motion state data uploaded by the vehicle and wind field data acquired by the sensing layer in real time, dynamically assembles the current five-dimensional state vector, and calls the pre-trained aerodynamic reduced-order model to perform forward inference. It completes the accurate mapping from "motion state" to "aerodynamic six components" in microseconds, thereby realizing a real-time digital twin of the vehicle's aerodynamic load under complex flow field conditions.

[0066] In the dynamic response derivation and state update stage, the real-time nonlinear aerodynamic six-component force obtained from the previous steps is used as a time-varying external excitation load and applied to the vehicle multibody dynamics reference model running in the roadside computing unit. The vehicle multibody dynamics reference model deeply integrates road geometric feature parameters (including longitudinal slope, horizontal curve curvature, and road surface superelevation) and uses the Newton-Euler dynamic equations for high-frequency solution to obtain the complete kinematic response of the vehicle under the combined action of current wind load and road surface excitation. The kinematic response specifically covers key state variables such as yaw rate, center of mass sideslip angle, vehicle roll angle, vehicle pitch angle, longitudinal / lateral acceleration, and lateral displacement. The kinematic and dynamic responses obtained above are fed back to the relative flow field synthesis module of the next simulation time step to update the vehicle's windward angle and attitude boundary conditions relative to the wind field.

[0067] Through the above processing, this application constructs a time-domain recursive closed-loop mechanism of "wind-vehicle" in the digital space, realizing high-fidelity interactive evolution of vehicle aerodynamic characteristics and multibody system dynamic characteristics, and ensuring that the simulation results can truly reflect the nonlinear aerodynamic stability of the vehicle in the dynamic process.

[0068] Step 103: Based on the updated vehicle dynamics model and the peak adhesion coefficient of the current road surface, calculate the comprehensive instability risk index covering sideslip risk, rollover risk and lateral deviation risk.

[0069] In this step, this application introduces a peak adhesion coefficient to construct a closed-loop correction mechanism for the road adhesion coefficient, which is a cross-domain parameter fusion mechanism. Addressing the limitation that roadside sensors cannot directly perceive the degree of road slippage, this mechanism utilizes the unscented Kalman filter of the vehicle chassis domain controller and the Burckhardt tire model to identify the peak road adhesion coefficient in real time and feed it back to the roadside unit. This mechanism is used to dynamically update the tire force constraint boundary of the dynamic model, ensuring that subsequent risk assessments accurately reflect the vehicle instability critical point under low adhesion conditions (such as rainy or snowy roads).

[0070] Based on the above mechanism, a comprehensive evaluation system for vehicle safety is constructed through step 103. Combining vehicle geometry and mass distribution parameters, a two-degree-of-freedom linear vehicle model is introduced as a reference benchmark to calculate the vehicle's steady-state response under current speed and steering input. For example... Figure 6 As shown, it includes the following steps: (31) Based on the updated vehicle dynamics model and the peak adhesion coefficient of the current road surface, calculate the sideslip index, the lateral deviation index and the rollover index; wherein the sideslip index and the rollover index are derived from the two-degree-of-freedom vehicle model.

[0071] The sideslip index includes dynamic yaw rate. and centroid side slip angle The function formula is as follows: .

[0072] in, This is the upper limit of the centroid sideslip angle. This is the actual sideslip angle of the centroid. For the ideal yaw rate, The wheel rotation angle is represented by sgn(), which is the sign function. Where is the peak adhesion coefficient of the current road surface, and g is the acceleration due to gravity. Let be the longitudinal vehicle speed, and min{} be the minimum value function; The lateral slip index includes the maximum lateral displacement at the vehicle body boundary point. The function formula is as follows: .

[0073] Among them, e y This refers to the lateral position error of the vehicle's center of gravity. Let be the angle between the centroid o, the vehicle boundary point point, and the vehicle's longitudinal axis. For vehicle width, For road width, The distance from the vehicle's center of mass to the vehicle's boundary point. The distance from the center of mass to the front and rear axles. Maximum permissible lateral displacement at vehicle boundary points; The rollover index includes lateral load transfer rate. The function formula is as follows: .

[0074] Where B1 is the wheel track, The height from the center of mass to the center of roll. The height of the center of mass from the ground. The roll angle is... This is the roll acceleration. For the sprung mass, For vehicle quality, Let be the moment of inertia about the x-axis. This refers to the vehicle's lateral acceleration.

[0075] (32) Normalize the sideslip index, the lateral deviation index and the rollover index respectively to obtain the sideslip risk coefficient, the lateral deviation risk coefficient and the rollover risk coefficient.

[0076] To address the mathematical challenge of inconsistent physical dimensions and the difficulty in directly superimposing the indicators of sideslip, rollover, and lateral slip, this application introduces a normalized mapping mechanism to construct a dimensionless comprehensive evaluation system for vehicle instability risk. Specifically, a set of safety thresholds determined by vehicle dynamics limits is preset and defined as: the center-of-gravity sideslip angle limit threshold. Yaw angular velocity limit threshold Lateral load transfer rate limit threshold and the maximum permissible lateral deviation of the lane Then, based on the above thresholds, the real-time calculated dynamic response is dimensionless, and the risk coefficients of each component are calculated: Sideslip risk factor The weighted normalized form of the sideslip angle and yaw rate is used, and the specific calculation is performed using the following formula: .

[0077] Side rollover risk factor The calculation is based directly on the lateral load transfer rate, using the following formula: .

[0078] Trajectory deviation risk factor The following formula is used to calculate the maximum lateral displacement of the vehicle body: .

[0079] in, For the ideal yaw rate, , Assign coefficients to the weights, satisfying It can be calibrated according to specific vehicle models.

[0080] (33) Based on the DS evidence theory, the sideslip risk coefficient, the lateral deviation risk coefficient and the rollover risk coefficient are fused to obtain the comprehensive instability risk index. .

[0081] Specifically, define the identification framework These represent the "stable" and "unstable" states, respectively. A basic probability allocation function (BPA) is constructed based on the risk coefficients of each component: .

[0082] The Dempster combination rule is applied to fuse the three sources of evidence. For two sources of evidence... , The fusion rules are as follows: .

[0083] Among them, the conflict coefficient .

[0084] By sequentially fusing the three sources of evidence, the basic probability distribution after fusion is obtained. ,satisfy , For credibility, It is an empty set. As the first source of evidence The focal length; As the second source of evidence Jiao Yuan, To identify a subset of the framework, the comprehensive instability risk index Defined as the support for the "instability" state after fusion: .

[0085] in, The value ranges from [0,1], with a larger value indicating a higher confidence level that the vehicle is in an unstable state. This index serves as a core state variable and is used for subsequent dynamic risk field construction and traffic control strategy generation.

[0086] Step 104: Construct a dynamic risk field based on the map information, the obstacle information, and the comprehensive instability risk index.

[0087] The Unified Spatiotemporal Dynamic Risk Field (URFLT) is a virtual potential energy field constructed in digital space to quantify driving risks. This virtual potential energy field is composed of a road boundary repulsion field, an obstacle repulsion field, and a wind-induced stability potential energy field based on the vehicle instability index. It explicitly transforms the invisible aerodynamic impact of crosswinds and the degree of road slippage into high-risk potential energy. By calculating the vehicle's potential energy intensity in this field, the physical instability risk of the vehicle under combined crosswind and low-adhesion conditions is comprehensively determined.

[0088] like Figure 7 As shown, step 104 utilizes roadside computing power to construct a unified potential energy field that includes environmental and dynamic constraints, and can generate a safe driving envelope based on this field, including the following steps: (41) Based on the map information, a repulsive field (to prevent vehicles from leaving the road surface) is constructed using an exponential function to obtain the potential energy field at the road boundary. The calculation formula is: .

[0089] Where y is the vehicle's current x-coordinate on the map, y center k represents the coordinates of the road boundary on the map. road Let be the attenuation coefficient of the potential energy field. This is the road potential energy amplitude coefficient.

[0090] The map information used above is lane line information from a high-precision map. It is determined through a combination of vehicle-to-road interaction, vehicle sensor sensing, and import of external maps (built into the vehicle), thereby improving the accuracy of vehicle positioning and lane line information.

[0091] (42) Based on the obstacle information, construct the obstacle potential energy field using a Gaussian distribution function. The calculation formula is: .

[0092] in, For the first The maximum repulsive potential energy of an obstacle , The standard deviation represents the range of influence of the obstacle; (X,Y) represents the real-time coordinates of the vehicle in the global coordinate system, where X is the standard deviation of the obstacle's influence range. obs,i Y is the x-coordinate of the geometric center of the i-th obstacle. obs,i Let be the ordinate of the geometric center of the i-th obstacle. This represents the number of obstacles.

[0093] The aforementioned obstacle information is constructed based on the perception data from roadside and vehicle-mounted sensors, and is based on a Gaussian distribution function. The geometric center of the obstacle is taken as the peak potential energy point, forming a repulsive field that diffuses outward.

[0094] (43) Based on the aforementioned comprehensive instability risk index Considering the saddle-shaped bifurcation characteristics of vehicle instability, a wind-induced stability potential energy field is constructed. Comprehensive Instability Risk Index The calculation has deeply integrated the real-time identified road adhesion coefficient in the previous steps. The defined dynamic boundary, based on which the potential energy field is derived, is defined at the longitudinal vehicle speed. and effective lateral wind speed In the state space with basis, the calculation formula is: .

[0095] in, The reference potential energy amplitude coefficient; This is the pavement condition sensitivity gain function; it is negatively correlated with the pavement adhesion coefficient. The critical instability speed is determined by... ,in This is the threshold for the risk of instability. v wind_lat This refers to the lateral wind speed. This is the comprehensive instability risk index calculated for the preceding steps. This definition ensures that the potential energy field... A stable potential well is formed when Approaching The potential well depth approaches zero, accurately characterizing the nonlinear properties of the vehicle approaching the physical instability boundary.

[0096] (44) The potential energy field of the road boundary, the potential energy field of the obstacle, and the potential energy field of the wind-induced stability are superimposed to generate a dynamic risk field with unified time and space. This ensures that the safety envelope generated subsequently can cover the extreme combined conditions of "strong crosswinds + low adhesion".

[0097] Step 105: Based on the principle of minimizing potential energy, the optimal reference path in the future time domain is determined using the gradient descent method within the dynamic risk field. and the corresponding maximum permissible safe speed curve This forms a dynamic safety envelope, which can then be transmitted to the vehicle unit via V2X.

[0098] In a specific application, such as Figure 8As shown, this application abandons the single wind speed threshold judgment logic and constructs a visualized three-dimensional risk assessment surface by reverse deducing the nonlinear mapping relationship between the risk potential energy function and the vehicle driving state. Based on this, differentiated strategies are formulated for different lanes and vehicle types, thereby implementing spatiotemporally synchronized active control. Step 105 includes: (51) Define the ratio of the potential energy field strength of the dynamic risk field to the critical potential energy threshold for vehicle physical instability as the vehicle instability risk value, and calculate it using the following function: .

[0099] in, It is a superposition field function that includes the potential energy of the road boundary, the potential energy of the obstacle, and the potential energy of wind-induced stability. The current road surface adhesion coefficient The critical potential energy threshold when a vehicle skids or overturns.

[0100] (52) Based on the dynamic risk field, a three-dimensional risk assessment surface is established using lateral wind speed, longitudinal vehicle speed, and vehicle instability risk values ​​as coordinate axes. For example... Figure 9 As shown, the 3D Risk Assessment Surface is a visualized traffic control decision-making method. This surface is constructed using lateral wind speed and longitudinal vehicle speed as independent variables on the plane, and vehicle instability risk value as the height axis. This surface can intuitively describe the nonlinear trend of different vehicle types (such as high-center-of-gravity trucks and low-center-of-gravity cars) approaching the instability boundary as vehicle speed increases. It can be used to inversely calculate the maximum permissible speed that meets the safety threshold, thereby achieving refined traffic control by lane and vehicle type.

[0101] Based on the potential energy field function During the reverse calculation, the environmental lateral wind speed is taken into account. and road surface adhesion coefficient It is an objectively existing time-varying environmental input, while the longitudinal vehicle speed It is an adjustable active variable, therefore it is constructed based on lateral wind speed. X-axis and longitudinal vehicle speed Y-axis, vehicle instability risk value Let Z be the three-dimensional feature space.

[0102] In low adhesion (such as wet and slippery conditions caused by rain or snow), Under these conditions, the maximum lateral friction force provided by the road surface is significantly reduced. The significant drop in values ​​leads to high-risk areas in the three-dimensional surface (i.e., The steep rise in wind speed (the region of rising wind speed) expands significantly into the low wind speed and low vehicle speed coordinate domain. This means that on slippery roads, even if the wind speed has not yet reached typhoon level, as the vehicle speed increases, the vehicle will pass the nonlinear critical point earlier and thus fall into the unstable high potential energy region.

[0103] By traversing and calculating the global variables, a three-dimensional map of "wind speed-vehicle speed-risk" for the road segment at the current moment is generated, which quantitatively describes the dynamic trend of vehicles approaching the instability boundary as vehicle speed increases under specific wind field and road surface conditions.

[0104] (53) The three-dimensional risk assessment surface is sliced ​​and reverse-calculated, and the maximum permissible safe speed and optimal reference path of each lane / vehicle type are calculated in a collaborative manner to form a dynamic safety envelope.

[0105] Step 106: Based on the dynamic safety envelope, determine the vehicle traffic control strategy by lane and vehicle type.

[0106] In a specific application embodiment, this application further divides traffic conditions into multiple control levels based on the maximum permissible safe speed curve, and determines corresponding vehicle traffic control strategies for each control level, categorized by lane and vehicle type; the vehicle traffic control strategies are transmitted to the on-board unit and the roadside unit respectively, so as to provide information prompts on the vehicle side and the roadside respectively.

[0107] Take a typical combined scenario of "strong crosswind + high vehicle speed + low adhesion" as an example: Assuming the measured lateral gust speed on the current cross-sea bridge section reaches 22 m / s (Force 9 strong wind), and the road surface adhesion coefficient is reduced due to heavy rain... The value drops to 0.4. At this point, the roadside calculation unit performs real-time slice analysis on the three-dimensional risk surface: Regarding vehicle type differences: For high-center-of-gravity SUVs or vans, due to their high aerodynamic center and large lateral wind exposure area, the risk of rollover and skidding is extremely high on low-friction surfaces. Three-dimensional surface calculations show that when the vehicle speed exceeds 60km / h, The value exceeds the safety threshold; however, for cars with a low center of gravity, thanks to their better aerodynamic shape, the risk value only approaches the critical point when the vehicle speed reaches 90km / h under the same operating conditions.

[0108] Regarding the differences between lanes: the first lane (windward side) is directly impacted by the airflow, resulting in high turbulence and a steep risk curve; while the third lane (leeward side) is affected by the shielding effect of bridge railings or vehicles on the outside, resulting in a lower local equivalent wind speed and a relatively gentle risk curve.

[0109] Based on the calculated real-time risk value Traffic conditions are divided into four control levels, and corresponding response logic is established: Region I (Free flow, 0 < Value ≤ a): Extremely low risk, no restrictions, only routine monitoring is maintained.

[0110] Region II (early warning flow, a< (Value ≤ b): There is a potential aerodynamic disturbance. A "recommended speed" is generated, and a "crosswind warning" is displayed on the information board.

[0111] Area III (flow restriction, b<) Value ≤ c): The vehicle is approaching the edge of instability, and a "speed limit" is forcibly generated (i.e., the value calculated above). This triggers the lane-specific speed limit instruction.

[0112] Area IV (No Entry) Value > c): Even at extremely low speeds, the risk remains uncontrollable (such as during typhoon conditions), triggering "lane closure" or "restriction of certain vehicle types" commands.

[0113] After determining the control level and response logic, the above control strategies are executed through dual-modal control in both physical and digital spaces, and are reflected in real time on the road through the following two paths, forming a closed-loop interaction of "people-vehicle-road": Roadside Physical Layer (Visual Signage): The roadside computing unit sends control commands via fiber optic network to variable message signs (CMS), gantry-mounted lane-specific variable speed limit signs (VSL), and LED guide light strips deployed on the roadside. For example, the speed limit sign above the first lane of a bridge displays "60" in real time, while the second lane displays "80". If a lane is deemed too risky, a red "X"-shaped no-entry light above that lane is illuminated, physically guiding drivers to avoid high-risk wind field areas through visual signals.

[0114] Vehicle-to-Everything (V2X) Digital Layer: Relying on the 5G / C-V2X communication link of the communication layer, the roadside computing unit broadcasts structured data packets (extended frames of the MAP / SPAT message set) containing "lane-specific speed limits," "crosswind risk levels," and "recommended driving trajectories" to all vehicles equipped with OBUs within range. After receiving this data, autonomous vehicles or intelligent connected vehicles can directly use it as a constraint for path planning, automatically reducing cruising speed or initiating lane-change requests, thereby achieving active safety control that does not rely on driver reaction.

[0115] In summary, this application utilizes the computing power of roadside units to obtain precise aerodynamic loads on vehicles based on measured wind fields and vehicle conditions, through a time-domain recursive closed-loop mechanism of aerodynamics and multibody dynamics. By employing a vehicle-road cooperative edge computing architecture and a POD aerodynamic reduction model, the spatiotemporal resolution of environmental perception is improved to the millisecond and meter levels. By reconstructing the unsteady flow field around vehicles in real time at the edge using roadside measured data, it can accurately capture instantaneous gust impacts that are often overlooked by existing technologies. This means that in high-wind-speed weather, this application can identify "local strong crosswind risk points" at specific bridge towers or tunnel exits, providing a reliable input source for subsequent precise control and fundamentally eliminating blind spots in environmental perception.

[0116] This application leverages the vehicle's perception of road conditions to correct the boundary conditions of the roadside model, forming a data closed loop. While the roadside system can calculate aerodynamic loads, it cannot directly perceive the degree of road slippage. Therefore, a road adhesion coefficient closed-loop correction mechanism based on vehicle-side UKF feedback is introduced. By allowing the vehicle to inform the roadside system in real time how slippery the road surface is, the safety boundary (i.e., the friction circle radius) of the dynamic model can be dynamically reduced. This gives the risk assessment model "environmental adaptability": in low-adhesion conditions such as rain and snow, it pre-determines that the vehicle's ability to withstand crosswinds is significantly reduced, thus allowing for early intervention (such as lowering the speed limit threshold) before the vehicle skids, achieving a leap from remediation to prevention.

[0117] This application constructs a three-dimensional risk assessment surface, unifying three core variables—high wind speed (environment), high vehicle speed (object), and low adhesion (medium)—into a quantitative model. This allows for the reverse calculation of the maximum permissible speeds for different vehicle types (e.g., cars vs. trucks) to maintain stability under current wind and road conditions. In terms of safety, under extreme conditions of wet and slippery roads with strong winds, the application precisely calculates the necessary reduction in high speed to offset the decrease in road adhesion, preventing reckless driving. In terms of efficiency, when wind speeds are high but the road surface is dry, or when vehicles are in the leeward lane, higher speeds are permitted, avoiding unnecessary road closures due to the risk posed by a few high-risk vehicles. Ultimately, this application achieves refined proactive control by lane and vehicle type, maximizing road capacity while ensuring zero accidents in extreme weather conditions.

[0118] Based on the same inventive concept, this application also provides a system. The solution provided by this system is similar to the solution described in the above method. Therefore, the specific limitations of one or more system embodiments provided below can be found in the limitations of the method above, and will not be repeated here.

[0119] In one exemplary embodiment, a wind-driven traffic management system is provided, which is applied to a vehicle-road cooperative edge computing architecture deployed on a pre-defined high-risk road section. The system includes: The data acquisition module is used to collect data on the actual vehicle motion status, actual wind field data, peak adhesion coefficient of the current road surface, map information, and obstacle information.

[0120] The dynamics update module is used to perform a high-dimensional fluid dynamics simulation task based on the pre-built aerodynamic reduced-order model, according to the measured vehicle motion state and the measured wind field data, to determine the six aerodynamic components; to load the six aerodynamic components into the vehicle multibody dynamics reference model, and to perform an aerodynamic-dynamic coupling solution task to obtain the vehicle dynamics and kinematic response; and to update the vehicle dynamics model based on the vehicle dynamics and kinematic response; wherein, the aerodynamic reduced-order model is constructed by combining a deep neural network and an intrinsic orthogonal decomposition algorithm.

[0121] The instability risk calculation module is used to calculate a comprehensive instability risk index covering sideslip risk, rollover risk and lateral deviation risk based on the updated vehicle dynamics model and the peak adhesion coefficient of the current road surface.

[0122] The dynamic risk field construction module is used to construct a dynamic risk field based on the map information, the obstacle information, and the comprehensive instability risk index.

[0123] The safety calculation module is used to determine the optimal reference path and the corresponding maximum permissible safe speed curve in the future time domain in the dynamic risk field based on the principle of minimizing potential energy, so as to form a dynamic safety envelope.

[0124] The strategy generation module is used to determine vehicle traffic control strategies for different lanes and vehicle types based on the dynamic safety envelope.

[0125] Compared with the prior art, this application has the following advantages: (1) A real-time aerodynamic digital twin method based on a vehicle-road cooperative distributed architecture: This application adopts a distributed processing architecture that migrates high-load fluid dynamics calculation tasks from the vehicle end to the roadside edge computing unit. This application also adopts an "offline high-precision training-online fast mapping" mechanism, the core of which is to construct an aerodynamic reduced-order model based on intrinsic orthogonal decomposition and deep neural network, so that it can calculate the nonlinear aerodynamic six-component force containing unsteady effects in milliseconds based on the measured wind field on the roadside and the real-time attitude of the vehicle. This provides the necessary real-time computation for dealing with instantaneous aerodynamic changes at high vehicle speeds and solves the technical bias that traditional CFD simulation cannot be used for real-time control.

[0126] (2) Dynamic boundary correction mechanism based on vehicle-side road surface adhesion coefficient feedback: This application adopts a cross-domain parameter fusion and model boundary correction method to specifically solve the perception blind spot of roadside equipment that "can see the wind but cannot see the slippage". Taking advantage of the perception of the vehicle chassis domain controller, the peak adhesion coefficient of the current road surface is identified in real time by combining unscented Kalman filtering with the Burckhardt tire model, and then transmitted back to the roadside through the V2X network.

[0127] The roadside unit uses this feedback data to update the tire friction circle boundary in the multibody dynamics reference model in real time. This mechanism ensures that the roadside risk assessment model can perceive low adhesion conditions such as "wet and slippery conditions," avoiding the safety hazards caused by assessing the road surface according to dry road surface standards when it is wet and slippery, and realizing real-time synchronization of boundary conditions between physical space and digital space.

[0128] (3) A unified spatiotemporal dynamic risk field construction method integrating road surface state sensitivity: This application constructs a virtual potential energy field that can simultaneously characterize road geometry, obstacles, and wind-induced risks. Its core lies in constructing a wind-induced stability potential energy function modulated by the road surface adhesion coefficient. In this function, the road surface adhesion coefficient is designed as a negative correlation parameter of the potential energy gain. When the system detects low adhesion (such as...) When the value decreases, even if the wind speed and vehicle speed remain unchanged, the potential well depth of the wind-induced potential energy field will automatically increase exponentially. This successfully quantifies the physical law that "the slipperier the road surface, the lower the tolerance to wind speed and vehicle speed" at the mathematical level, providing a unified scalar basis for subsequent automated control.

[0129] (4) Lane- and Vehicle-Specific Refined Control Strategy Based on Three-Dimensional Risk Assessment Surface: This application adopts a traffic control decision model based on multi-dimensional variable inverse calculation to replace the traditional single wind speed threshold judgment. This application constructs a three-dimensional risk assessment surface with lateral wind speed, longitudinal vehicle speed, and comprehensive instability risk index as coordinate axes. By slicing and solving this surface, under given real-time wind speed and road adhesion conditions, the maximum permissible safe speed required for different vehicle types (such as high center of gravity trucks vs. low center of gravity sedans) to maintain stability is calculated inversely. This achieves a leap from "qualitative road closure" to "quantitative speed control". In high-wind weather, the system can reduce high vehicle speeds to gain stability (i.e., find low-risk operating points on the three-dimensional surface) and implement differentiated speed limits for different lanes (windward / leeward), maximizing traffic efficiency while ensuring safety.

[0130] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 10As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and databases. The internal memory provides the environment for the operating system and computer programs stored in the non-volatile storage media to run. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. When the computer program is executed by the processor, it implements a traffic management method driven by wind-induced driving risks.

[0131] Those skilled in the art will understand that Figure 10 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0132] In one exemplary embodiment, a computer device is also provided, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps in the above-described method embodiments.

[0133] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0134] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0135] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0136] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0137] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0138] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0139] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A traffic control method driven by wind-induced driving risks, characterized in that, The method, applied to a vehicle-road cooperative edge computing architecture deployed on a pre-defined high-risk road segment, includes: Collect measured vehicle motion status, measured wind field data, peak adhesion coefficient of the current road surface, map information, and obstacle information; Based on a pre-constructed aerodynamic reduced-order model, a high-dimensional fluid dynamics simulation task is performed according to the measured vehicle motion state and the measured wind field data to determine the six aerodynamic components; the six aerodynamic components are then loaded into the vehicle multibody dynamics reference model, and an aerodynamic-dynamic coupling solution task is performed to obtain the vehicle dynamics and kinematic response; the vehicle dynamics model is updated based on the vehicle dynamics and kinematic response; wherein, the aerodynamic reduced-order model is constructed by combining a deep neural network and an intrinsic orthogonal decomposition algorithm; Based on the updated vehicle dynamics model and the peak adhesion coefficient of the current road surface, a comprehensive instability risk index covering sideslip risk, rollover risk and lateral deviation risk is calculated. A dynamic risk field is constructed based on the map information, the obstacle information, and the comprehensive instability risk index. Based on the principle of minimizing potential energy, the optimal reference path and the corresponding maximum permissible safe speed curve in the future time domain are determined in the dynamic risk field to form a dynamic safety envelope. Based on the dynamic safety envelope, traffic control strategies for different lanes and vehicle types are determined.

2. The traffic control method driven by wind-induced driving risks according to claim 1, characterized in that, The setup process for the vehicle-road cooperative edge computing architecture includes: When the target is a cross-sea bridge, the gradient wind speed in different return periods of the bridge site area is inverted based on the historical wind data of the bridge site area, and the wind speed at the bridge site area at the preset height is calculated in combination with the surface type. Establish a calculation model of the topographic wind environment of the bridge site area or conduct a topographic wind tunnel experiment. Based on the wind speed and wind profile characteristics of the bridge site area at a preset height, conduct a study on the wind environment of the bridge site area and calculate the wind field characteristics simultaneously. Based on the overall bridge layout and cross-sectional diagram of the cross-sea bridge, select the section to be used; Based on the wind field characteristics, calculate the calculated wind speed at the section to be used; We conducted aerodynamic characteristic analysis of bridge traffic under calculated wind speeds for different cross sections, combined with dynamic analysis to determine the dangerous locations of bridges along the entire route, and marked them as the locations for the vehicle-road cooperative edge computing architecture.

3. The traffic control method driven by wind-induced driving risks according to claim 1, characterized in that, The vehicle-road cooperative edge computing architecture includes an on-board unit, a roadside unit, and a communication unit; the roadside unit includes a roadside perception layer and a roadside computing layer. The on-board unit is used to sense and collect the measured vehicle motion state, the peak adhesion coefficient of the current road surface, the vehicle side map information, and the vehicle side obstacle information in real time, and then upload them to the communication unit; wherein, the peak adhesion coefficient of the current road surface is obtained in real time by using the unscented Kalman filter of the chassis domain controller in the on-board unit combined with the Burckhardt tire model. The roadside sensing layer is used to sense and collect measured wind field data, roadside map information and roadside obstacle information in real time, and then upload them to the communication unit. The communication unit is used to establish a bidirectional data transmission link between the roadside unit and the vehicle-mounted unit. The roadside computing unit is used to: determine map information by combining the vehicle-side map information and the roadside map information; determine obstacle information by combining the vehicle-side obstacle information and the roadside obstacle information; and perform high-dimensional fluid dynamics simulation tasks, unsteady aerodynamic calculation tasks for vehicle aerodynamics, and risk inference and strategy generation tasks.

4. The traffic control method driven by wind-induced driving risks according to claim 1, characterized in that, The measured vehicle motion states include: longitudinal vehicle speed, vehicle transient attitude angle, and vehicle ID; the measured wind field data includes the ambient lateral wind speed and wind deflection angle. The process of constructing the aerodynamic reduced-order model includes: Acquire a set of measured sample data; the measured sample data includes the measured sample vehicle motion state and the measured sample wind field data; Based on the vehicle ID in the measured sample vehicle motion state, the vehicle geometric model and dynamic parameter set are called; Based on the measured sample vehicle motion state and the measured sample wind field data, the synthetic airflow velocity and synthetic aerodynamic deflection angle are calculated. Based on the vehicle geometric model and dynamic parameter set, a high-dimensional aerodynamic sample space covering the entire operating envelope of the vehicle is constructed, and the synthetic airflow velocity and the synthetic aerodynamic deflection angle are used as boundary conditions. Within the high-dimensional aerodynamic sample space, batch CFD numerical simulations are performed using the unsteady Reynolds-averaged Navier-Stokes equations to obtain flow field snapshots and corresponding nonlinear aerodynamic six-component forces under different working conditions. Intrinsic orthogonal decomposition technology is introduced to reduce the dimension of the flow field snapshot and extract key physical features; the key physical features include the synthetic airflow velocity, the synthetic aerodynamic deflection angle, and the vehicle transient attitude angle; The key physical features and their corresponding aerodynamic six components are input into a preset deep neural network model for feature learning and mapping training to obtain a reduced-order aerodynamic model.

5. The traffic control method driven by wind-induced driving risks according to claim 1, characterized in that, Based on the updated vehicle dynamics model and the peak adhesion coefficient of the current road surface, a comprehensive instability risk index covering sideslip risk, rollover risk, and lateral deviation risk is calculated, including: Based on the updated vehicle dynamics model and the peak adhesion coefficient of the current road surface, the sideslip index, lateral deviation index and rollover index are calculated. The sideslip index, the lateral deviation index, and the rollover index are normalized to obtain the sideslip risk coefficient, the lateral deviation risk coefficient, and the rollover risk coefficient, respectively. Based on the DS evidence theory, the sideslip risk coefficient, the lateral deviation risk coefficient, and the rollover risk coefficient are fused to obtain a comprehensive instability risk index.

6. The traffic control method driven by wind-induced driving risks according to claim 5, characterized in that, The sideslip index includes dynamic yaw rate. and centroid side slip angle The function formula is as follows: ； in, This is the upper limit of the centroid sideslip angle. This is the actual sideslip angle of the centroid. For the ideal yaw rate, The wheel rotation angle is represented by sgn(), which is the sign function. Where is the peak adhesion coefficient of the current road surface, and g is the acceleration due to gravity. Let be the longitudinal vehicle speed, and min{} be the minimum value function; The lateral slip index includes the maximum lateral displacement at the vehicle body boundary point. The function formula is as follows: ； Among them, e y This refers to the lateral position error of the vehicle's center of gravity. Let be the angle between the centroid o, the vehicle boundary point point, and the vehicle's longitudinal axis. For vehicle width, For road width, The distance from the vehicle's center of mass to the vehicle's boundary point. The roll angle is... The distance from the center of mass to the front and rear axles. Maximum permissible lateral displacement at vehicle boundary points; The rollover index includes lateral load transfer rate. The function formula is as follows: ； Where B1 is the wheel track, The height from the center of mass to the center of roll. The height of the center of mass from the ground. This is the roll acceleration. For the sprung mass, For vehicle quality, Let be the moment of inertia about the x-axis. This refers to the vehicle's lateral acceleration. The sideslip risk coefficient is calculated using the following formula. : ； The lateral risk coefficient is calculated using the following formula. : ； The rollover risk coefficient is calculated using the following formula. : ； in, For the ideal yaw rate, , Assign coefficients to the weights, satisfying ; The threshold value for the centroid sideslip angle. The yaw rate limit threshold. This represents the lateral load transfer rate limit threshold. This represents the maximum permissible lateral deviation of the lane. The comprehensive instability risk index The calculation process includes: The basic probability assignment function is constructed using the following formula to obtain three sources of evidence: ； in, For credibility, S refers to a stable state and U refers to an unstable state; For any two of the three sources of evidence , The following formula is used for fusion, and the basic probability allocation after fusion is obtained. : ； Among them, the conflict coefficient ; Identification framework ; satisfy ; ； in, To comprehensively assess the risk of instability, For the overall credibility of risk, It is an empty set. As the first source of evidence The focal length; As the second source of evidence Jiao Yuan, This is a subset of the identification frame.

7. The traffic control method driven by wind-induced driving risks according to claim 1, characterized in that, Based on the map information, the obstacle information, and the comprehensive instability risk index, a dynamic risk field is constructed, including: Based on the map information, a repulsive field is constructed using an exponential function to obtain the potential energy field at the road boundary. The calculation formula is: ； Where y is the vehicle's current x-coordinate on the map, y center k represents the coordinates of the road boundary on the map. road Let be the attenuation coefficient of the potential energy field. This refers to the road potential energy amplitude coefficient. Based on the obstacle information, a Gaussian distribution function is used to construct the obstacle potential energy field. The calculation formula is: ； in, Let be the maximum repulsive potential energy of the i-th obstacle. , The standard deviation represents the range of influence of the obstacle; (X,Y) are the real-time coordinates of the vehicle, X... obs,i Y is the x-coordinate of the geometric center of the i-th obstacle. obs,i Let be the ordinate of the geometric center of the i-th obstacle. The number of obstacles; Based on the aforementioned comprehensive instability risk index Considering the saddle-shaped bifurcation characteristics of vehicle instability, a wind-induced stability potential energy field is constructed. The calculation formula is: ； in, The reference potential energy amplitude coefficient; This is the pavement condition sensitivity gain function, which is negatively correlated with the pavement adhesion coefficient. The critical speed at which the vehicle becomes unstable. , This is the threshold for the risk of instability. v wind_lat This refers to the lateral wind speed; The potential energy field of the road boundary, the potential energy field of the obstacle, and the potential energy field of the wind-induced stability are superimposed to generate a dynamic risk field in a unified time and space.

8. The traffic control method driven by wind-induced driving risks according to claim 1, characterized in that, Based on the principle of potential energy minimization, the optimal reference path and the corresponding maximum permissible safe speed curve in the future time domain are determined in the dynamic risk field to form a dynamic safety envelope, including: The ratio of the potential energy field strength of the dynamic risk field to the critical potential energy threshold for vehicle physical instability is defined as the vehicle instability risk value. Based on the dynamic risk field, a three-dimensional risk assessment surface is established with lateral wind speed, longitudinal vehicle speed, and vehicle instability risk value as coordinate axes. The three-dimensional risk assessment surface is sliced ​​and reverse-calculated to obtain the maximum permissible safe speed and optimal reference path for each lane / vehicle type, thus forming a dynamic safety envelope.

9. A traffic control system driven by wind-induced driving risks, characterized in that, An edge computing architecture for vehicle-road cooperation is deployed on pre-defined high-risk road sections. The system includes: The data acquisition module is used to collect data on the actual vehicle motion status, actual wind field data, peak adhesion coefficient of the current road surface, map information, and obstacle information. The dynamics update module is used to perform a high-dimensional fluid dynamics simulation task based on the pre-built aerodynamic reduced-order model, according to the measured vehicle motion state and the measured wind field data, to determine the six aerodynamic components; load the six aerodynamic components into the vehicle multibody dynamics reference model, and perform an aerodynamic-dynamic coupling solution task to obtain the vehicle dynamics and kinematic response; update the vehicle dynamics model based on the vehicle dynamics and kinematic response; wherein, the aerodynamic reduced-order model is constructed by combining a deep neural network and an intrinsic orthogonal decomposition algorithm; The instability risk calculation module is used to calculate a comprehensive instability risk index covering sideslip risk, rollover risk and lateral deviation risk based on the updated vehicle dynamics model and the peak adhesion coefficient of the current road surface. The dynamic risk field construction module is used to construct a dynamic risk field based on the map information, the obstacle information, and the comprehensive instability risk index. The safety calculation module is used to determine the optimal reference path and the corresponding maximum permissible safe speed curve in the future time domain in the dynamic risk field based on the principle of minimizing potential energy, so as to form a dynamic safety envelope. The strategy generation module is used to determine vehicle traffic control strategies for different lanes and vehicle types based on the dynamic safety envelope.

10. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the traffic control method driven by wind-induced driving risks according to any one of claims 1-8.

Images