A highway safety warning system based on vehicle-road cooperation

Abstract

The present invention discloses a road safety warning system based on vehicle-road cooperation. The system includes: a data acquisition module for forming a standardized driving environment data set; a multi-dimensional environment feature construction module for extracting a set of driving environment features that have a key impact on driving safety through a feature selection method; a quantum genetic variation optimization module for generating an optimized set of driving environment features; a multi-objective driving warning optimization module for solving a multi-objective driving warning optimization model to obtain an optimal driving warning strategy; and a warning information transmission module for transmitting the optimal driving warning strategy to a vehicle control system or a driver terminal in a multi-node and multi-path data transmission manner in real time, so as to realize road safety warning of vehicle-road cooperation. The present invention can perform real-time optimization and adjustment according to traffic flow states and driving behaviors, ensuring that the warning system can not only quickly respond to emergencies but also avoid causing excessive interference to drivers.

Description

Technical Field

[0001] The present invention relates to the field of radar technology, and in particular to a highway safety warning system based on vehicle-road cooperation. Background Art

[0002] In recent years, with the development of intelligent transportation technology, vehicle-road cooperation technology has become an important means to improve highway safety. By enabling information interaction between vehicles, road infrastructure, and cloud platforms, vehicle-road cooperation technology can effectively improve the efficiency of traffic flow and reduce the accident rate, especially in low visibility environments. However, the current highway safety warning system still faces many technical challenges in low visibility environments.

[0003] Currently, traditional highway safety warnings mainly rely on the following methods: First, environmental perception based on fixed monitoring devices can provide road conditions and weather information, but due to the limited monitoring range, it is difficult to dynamically reflect the overall driving environment in real time; Second, early warning systems based on vehicle intelligence can sense surrounding obstacles and driving environments, but are limited by the vehicle's own sensing capabilities, and the detection accuracy and reliability decrease significantly in low visibility environments; Third, centralized warning schemes based on traffic control centers rely on data collected by monitoring centers for global analysis and transmit information to drivers through broadcasts and dynamic display screens, but there are problems such as long data transmission delays and lack of personalization, making it difficult to adapt to sudden changes in the driving environment in real time.

[0004] Traditional technologies have significant defects in practical applications: First, information lag: Due to the centralized processing mode of traffic control centers, there are time delays in the generation and transmission of warning signals, which cannot meet the requirements of rapid response; Second, poor environmental adaptability: The warning capabilities of vehicle intelligence are limited by the sensing range and environmental adaptability of in-vehicle sensors, and the performance drops significantly in bad weather; Third, insufficient global optimization ability: Existing warning systems based on fixed monitoring devices or vehicle intelligence are difficult to comprehensively consider the overall road traffic flow, vehicle status, and environmental changes, resulting in a single warning strategy and lack of optimization and adjustment capabilities. In addition, existing technologies often lack intelligent feature extraction and optimization strategies, making it difficult to improve the accuracy and reliability of warning strategies.

[0005] Therefore, existing technologies have significant limitations in driving safety warnings in low visibility environments, and there is an urgent need for a technical solution that combines multi-source environmental data, optimizes driving warning strategies, and can transmit warning information in real time to improve highway driving safety. Summary of the Invention

[0006] An object of the present invention is to propose a road safety warning system based on vehicle-road cooperation. The present invention can perform real-time optimization and adjustment according to traffic flow states and driving behaviors, ensuring that the warning system can not only quickly respond to emergencies but also avoid causing excessive interference to drivers.

[0007] A road safety warning system based on vehicle-road cooperation according to an embodiment of the present invention includes the following modules:

[0008] A data acquisition module, configured to acquire a driving environment data set in a low visibility environment, and perform time synchronization, noise filtering, and normalization processing on the driving environment data set to form a normalized driving environment data set;

[0009] A multi-dimensional environment feature construction module, configured to construct a multi-dimensional environment feature model based on the normalized driving environment data set, and extract a set of driving environment features that have a key impact on driving safety through a feature selection method;

[0010] A quantum genetic variation optimization module, configured to perform global search and optimization on the set of driving environment features by using a quantum genetic variation algorithm to generate an optimized set of driving environment features;

[0011] A multi-objective driving warning optimization module, configured to construct a multi-objective driving warning optimization model based on the optimized set of driving environment features, solve the multi-objective driving warning optimization model, and obtain an optimal driving warning strategy;

[0012] A warning information transmission module, configured to transmit the optimal driving warning strategy to a vehicle control system or a driver terminal in real time in a multi-node and multi-path data transmission manner to achieve road safety warning of vehicle-road cooperation.

[0013] The present invention also discloses a road safety warning method based on vehicle-road cooperation, which is applied to a road safety warning system based on vehicle-road cooperation, and includes the following steps:

[0014] S1. Acquire a driving environment data set in a low visibility environment, perform time synchronization, noise filtering, and normalization processing on the driving environment data set to form a normalized driving environment data set;

[0015] S2. Based on the normalized driving environment data set, construct a multi-dimensional environment feature model, and extract a set of driving environment features that have a key impact on driving safety through a feature selection method;

[0016] S3. Use a quantum genetic variation algorithm to perform global search and optimization on the set of driving environment features to obtain an optimized set of driving environment features;

[0017] S4. Based on the optimized set of driving environment characteristics, construct a multi-objective driving warning optimization model;

[0018] S5. Solve the multi-objective driving warning optimization model to obtain the optimal driving warning strategy;

[0019] S6. Transmit the optimal driving warning strategy to the vehicle control system or the driver terminal in real time in a multi-node and multi-path data transmission manner to achieve road-vehicle collaborative highway safety warning.

[0020] Optionally, the S1 includes the following steps:

[0021] S11. Collect the driving environment data set in low visibility environment : Wherein, represents the th driving environment data, is the total number of collected driving environment data, is the time stamp corresponding to the driving environment data, is the numerical information of the driving environment data, is the device identifier of the driving environment data source, is the category label of the driving environment data, including vehicle dynamic data collected by in-vehicle sensors, road condition and environment data collected by road infrastructure, meteorological data collected by meteorological equipment, and traffic flow data collected by traffic monitoring systems;

[0022] S12. Perform time synchronization processing on the driving environment data set, construct a unified time reference, and correct the time deviation of each data source through interpolation and time alignment methods to obtain the time-synchronized driving environment data set;

[0023] S13. According to the category labels of the driving environment data in the driving environment data set, perform noise filtering processing on the vehicle dynamic data, road condition and environment data, meteorological data, and traffic flow data respectively, and use the adaptive filtering method to remove outliers to obtain the noise-filtered driving environment data set;

[0024] S14. Use the standardization method to normalize the noise-filtered driving environment data set, unify the numerical scales of various driving environment data, and obtain the standardized driving environment data set ;

[0025] S15. Remove redundant features from the standardized driving environment data set based on the correlation analysis of the driving environment data, calculate the correlation matrix of the driving environment data to quantify the correlation between the driving environment data, and obtain the finally standardized driving environment data set :

[0026] Among them, represents the th piece of driving environment data after final standardization, represents the adjusted standard timestamp, represents the numerical information of the driving environment data after removing redundant features.

[0027] Optionally, the S2 includes the following steps:

[0028] S21. Extract features from each piece of driving environment data in the finally standardized driving environment data set to construct a multi-dimensional environmental feature vector : ; among them, represents the meteorological feature extracted from the th piece of driving environment data, represents the road condition and environmental feature extracted from the th piece of driving environment data, represents the traffic flow feature extracted from the th piece of driving environment data, represents the vehicle dynamic feature extracted from the th piece of driving environment data;

[0029] S22. Construct a multi-dimensional environmental feature model ;

[0030]

[0031] S23. Use a feature selection method to process each multi-dimensional environmental feature vector in the multi-dimensional environmental feature model to extract a set of driving environment features that have a key impact on driving safety: Among them, is the kth key feature selected from the multi-dimensional environmental feature model, represents the key feature index set.

[0032] Optionally, the S3 includes the following steps:

[0033] S31. Perform quantum superposition state encoding based on the set of driving environment features to construct a quantum environmental feature population , and each quantum individual in the quantum environmental feature population represents a combination of driving environment features: ; among them, is the quantum population size, represents the A quantum individual, including the combination of driving environment characteristics in a low visibility environment, and are both the quantum amplitude coefficients corresponding to the th key feature in the th quantum individual;

[0034] S32. Perform quantum measurement on the quantum environment feature population to obtain the set of classical driving environment feature individuals wherein, is the th driving environment feature individual obtained after measurement, is the set of feature indices selected after measurement;

[0035] For each driving environment feature individual calculate the fitness value : wherein, is the early warning response time calculated based on the driving environment feature individual , is the shortest early warning response time, is the early warning accuracy rate of the driving environment feature individual , is the calculation result of the road environment complexity of the driving environment feature individual , is a dynamic weight parameter, adjusted according to the current low visibility environment characteristics to improve the early warning optimization effect under different environmental conditions;

[0036] S33. Based on driving dynamic perception, construct a quantum rotation update rule, and combine with the road information flow provided by the vehicle-road collaborative platform to dynamically adjust the search direction of quantum individuals and update quantum individuals: ; wherein, represents the th quantum individual in the (t + 1)th generation, represents the th quantum individual in the tth generation, is an adaptive quantum rotation gate based on vehicle-road collaborative environmental factors, is a step size control parameter, is a random perturbation factor of Lévy distribution, is the difference in quantum states between the current quantum individual and the population optimal individual, used to enhance the search ability, and the rotation angle is calculated as follows: wherein, is a quantum rotation step size control parameter, represents the low visibility environment impact factor, is the maximum low visibility environmental impact factor, is the road environment complexity, is the maximum road complexity, is the intensity of road monitoring information provided by the vehicle-road collaborative platform, is the optimal fitness value in the current population;

[0037] S34. Judge whether the quantum population converges according to the optimization convergence criterion. If the convergence condition is met, output the finally optimized set of driving environment characteristics ; otherwise, return to step S32 to continue iterative optimization.

[0038] Optionally, the S34 includes the following steps:

[0039] S341. Calculate the average fitness of the current population based on the quantum environmental characteristic population and the fitness value and the average fitness of the previous generation , and calculate the fitness change rate : wherein, is the quantum population size, is the th individual of the driving environment characteristics at the th generation;

[0040] S342. Define the quantum population diversity metric function , and calculate the population mean square error based on the amplitude vectors of the current generation of quantum individuals: wherein, is the characteristic dimension of the quantum individual encoding, and are both the amplitude parameters of the th quantum individual in the th characteristic dimension, and are both the average amplitudes of all quantum individuals in this characteristic dimension, reflects the diversity of the quantum population, and the smaller the value, the more the population tends to converge;

[0041] S343. Calculate the convergence determination parameter of the quantum population according to the fitness change rate and the quantum population diversity metric function: wherein, and are the convergence determination weight parameters;

[0042] S344. Define the optimization termination threshold , and when the following conditions are met, it is determined that the quantum population reaches the optimized convergence state: Otherwise, return to step S341 to continue the optimization iteration;

[0043] S345. After meeting the optimization convergence condition, select the individual with the highest fitness from the current quantum population as the finally optimized driving environment feature set: ; where in the t-th optimization iteration, the q-th driving environment feature individual in the population, represents the complete driving environment feature individual population in the t-th optimization iteration, and argmax is to find the fitness function.

[0044] Optionally, the step S4 includes the following steps:

[0045] S41. Based on the driving environment feature set , define a multi-objective driving warning optimization model , the optimization objectives of the multi-objective driving warning optimization model include minimizing the warning time, maximizing the warning accuracy rate, and optimizing the driving comfort. The objective function of the multi-objective driving warning optimization model is defined as follows:

[0046] where is the driving warning strategy to be optimized, is the warning time of the driving warning strategy, is the warning accuracy rate of the driving warning strategy, is the degree of interference of the driving warning strategy to the driver, is the multi-objective weight parameter;

[0047] S42. Define the objective of minimizing the warning time. The warning time of the driving warning strategy is affected by the driving environment feature set and the low visibility environment impact factor: where represents the detectable distance in the current low visibility environment, represents the transmission speed of the driving warning signal, is the processing time of the vehicle-mounted control system;

[0048] S43. Define the objective of maximizing the warning accuracy rate. The warning accuracy rate is affected by the driving environment feature set and the road monitoring information provided by the vehicle-road collaborative platform: where represents the number of false alarms or missed alarms in the current driving environment, is the total number of warnings;

[0049] S44. Define the objective of optimizing the driving comfort. The degree of interference to the driving comfort Affected by the warning frequency and the driver's warning response time: Among them, represents the number of warnings triggered per unit time, represents the average response time of the driver to the warning signal;

[0050] S45. Construct an adaptive weight adjustment mechanism and define a dynamic adjustment weight parameter Make the dynamic adjustment weight parameter adapt to different low visibility environments: Among them, is the optimized weight parameter of the (t + 1)-th generation, is the weight parameter of the t-th generation, is the learning rate, is the low visibility environment impact factor, is the maximum low visibility environment impact factor, is the road monitoring information intensity provided by the vehicle-road collaborative platform, is the maximum road monitoring information intensity.

[0051] Optionally, the S3 includes the following steps:

[0052] S31. Based on the set of driving environment characteristics Perform quantum superposition state encoding to construct a quantum driving environment characteristic population , and each quantum individual in the quantum driving environment characteristic population represents a combination of driving environment characteristics: ; among them, is the quantum population size, represents the -th quantum individual, including the combination of driving environment characteristics in low visibility environment, and are both the quantum amplitude coefficients corresponding to the -th key feature in the -th quantum individual;

[0053] S32. Perform quantum measurement on the quantum driving environment characteristic population to obtain a set of classical driving environment characteristic individuals Among them, is the -th driving environment characteristic individual obtained after measurement, is the set of selected feature indexes after measurement;

[0054] For each driving environment characteristic individual calculate the fitness value : Among them, is based on the driving environment characteristic individual The calculated warning response time is the shortest warning response time is the warning accuracy rate of the individual of the driving environment characteristics ; is the calculation result of the road environment complexity of the individual of the driving environment characteristics ; is a dynamic weight parameter, which is adjusted according to the current low visibility environment characteristics to improve the warning optimization effect under different environmental conditions;

[0055] S33. Construct a quantum rotation update rule based on driving dynamic perception, combine the road information flow provided by the vehicle-road collaborative platform, dynamically adjust the search direction of the quantum individual, and update the quantum individual: ; where represents the th quantum individual in the (t + 1)th generation represents the th quantum individual in the tth generation is an adaptive quantum rotation gate based on vehicle-road collaborative environmental factors is a step size control parameter is a random perturbation factor of the Lévy distribution is the difference in the quantum states between the current quantum individual and the optimal individual in the population, which is used to enhance the search ability, and the rotation angle is calculated as follows: where is a quantum rotation step size control parameter represents the low visibility environment impact factor is the maximum low visibility environment impact factor is the road environment complexity is the maximum road complexity is the intensity of the road monitoring information provided by the vehicle-road collaborative platform is the optimal fitness value in the current population;

[0056] S34. Judge whether the quantum population converges according to the optimization convergence criterion. If the convergence condition is met, output the finally optimized set of driving environment characteristics , otherwise return to step S32 to continue iterative optimization.

[0057] Optionally, the S34 includes the following steps:

[0058] S341. Calculate the fitness mean value of the current population and the fitness mean value of the previous generation based on the quantum environment characteristic population and the fitness value, and calculate the fitness change rate : where is the quantum population size, is the th individual of driving environment characteristics at the th generation;

[0059] S342. Define the quantum population diversity metric function , and calculate the population mean square deviation based on the amplitude vectors of the current generation of quantum individuals: where is the characteristic dimension of the quantum individual encoding, and are both the amplitude parameters of the th characteristic dimension in the th quantum individual, and are both the mean amplitudes of all quantum individuals in this characteristic dimension, reflects the diversity of the quantum population, and the smaller the value, the more the population tends to converge;

[0060] S343. Calculate the convergence determination parameter of the quantum population according to the fitness change rate and the quantum population diversity metric function: where and are the convergence determination weight parameters;

[0061] S344. Define the optimization termination threshold , and when the following conditions are met, it is determined that the quantum population has reached the optimized convergence state: Otherwise, return to step S341 to continue the optimization iteration;

[0062] S345. After meeting the optimization convergence condition, select the individual with the highest fitness from the current quantum population as the finally optimized set of driving environment characteristics: ; where is the th individual of driving environment characteristics in the population during the t-th optimization iteration,

[0063] Optionally, the said S4 includes the following steps:

[0064] S41. Based on the set of driving environment characteristics , define the multi-objective driving warning optimization model , the optimization objectives of the multi-objective driving warning optimization model include minimizing the warning time, maximizing the warning accuracy rate, and optimizing the driving comfort. The objective function of the multi-objective driving warning optimization model is defined as follows:

[0065] Among them, is the driving warning strategy to be optimized, is the warning time of the driving warning strategy, is the warning accuracy rate of the driving warning strategy, is the degree of interference of the driving warning strategy to the driver, is the multi-objective weight parameter;

[0066] S42. Define the objective of minimizing the warning time. The warning time of the driving warning strategy is affected by the set of driving environment characteristics and the low visibility environment impact factor: Among them, represents the detectable distance in the current low visibility environment, represents the transmission speed of the driving warning signal, is the processing time of the vehicle-mounted control system;

[0067] S43. Define the objective of maximizing the warning accuracy rate. The warning accuracy rate is affected by the set of driving environment characteristics and the road monitoring information provided by the vehicle-road collaborative platform: Among them, represents the number of false alarms or missed alarms in the current driving environment, is the total number of warnings;

[0068] S44. Define the objective of optimizing the driving comfort. The degree of interference to the driving comfort is affected by the warning frequency and the driver's warning response time: Among them, represents the number of warnings triggered per unit time, represents the average reaction time of the driver to the warning signal;

[0069] S45. Construct an adaptive weight adjustment mechanism and define the dynamic adjustment weight parameter to make the dynamic adjustment weight parameter adapt to different low visibility environments: Among them, is the optimized weight parameter of the (t + 1)-th generation, is the weight parameter of the t-th generation, is the learning rate, is the low visibility environment impact factor, is the maximum low visibility environment impact factor, is the intensity of the road monitoring information provided by the vehicle-road collaborative platform, is the maximum road monitoring information intensity.

[0070] The beneficial effects of the present invention are as follows:

[0071] The present invention uses a quantum genetic mutation algorithm to globally search and optimize the driving environment feature set. Through quantum superposition state encoding and quantum rotation update rules, it can efficiently explore the optimal combination of driving environment features in a multi-dimensional feature space. By introducing the probability search mechanism of quantum computing, quantum individuals can explore multiple feature combinations simultaneously and dynamically adjust the search direction, thereby ensuring the global optimality of the optimization result.

[0072] The present invention proposes a multi-objective driving warning optimization model. The multi-objective driving warning optimization model comprehensively considers three core indicators: minimizing the warning time, maximizing the warning accuracy rate, and optimizing the driving comfort. Through an adaptive weight adjustment mechanism, it realizes dynamic balance. By dynamically adjusting the multi-objective weight parameters, in a low visibility environment, it can perform real-time optimization and adjustment according to the traffic flow state and driving behavior, ensuring that the warning system can not only quickly respond to emergencies but also avoid causing excessive interference to the driver. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] The drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation to the present invention. In the drawings:

[0074] Figure 1 is a flowchart of a road safety warning system based on vehicle-road cooperation proposed by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0075] Now, the present invention will be further described in detail with reference to the drawings. These drawings are all simplified schematic diagrams, only showing the basic structure of the present invention in a schematic way, so they only show the components related to the present invention.

[0076] Refer to Figure 1 , a road safety warning system based on vehicle-road cooperation, includes the following modules:

[0077] A data acquisition module, which is used to collect a driving environment data set in a low visibility environment, and perform time synchronization, noise filtering, and standardization processing on the driving environment data set to form a standardized driving environment data set;

[0078] A multi-dimensional environment feature construction module, which is used to construct a multi-dimensional environment feature model based on the standardized driving environment data set, and extract a set of driving environment features that have a key impact on driving safety through a feature selection method;

[0079] A quantum genetic variation optimization module, which is used to globally search and optimize the driving environment feature set by using the quantum genetic variation algorithm, and generate an optimized driving environment feature set;

[0080] A multi-objective driving warning optimization module, which is used to construct a multi-objective driving warning optimization model based on the optimized driving environment feature set, solve the multi-objective driving warning optimization model, and obtain an optimal driving warning strategy;

[0081] A warning information transmission module, which is used to transmit the optimal driving warning strategy to the vehicle control system or the driver terminal in a multi-node and multi-path data transmission manner in real time, so as to realize road-vehicle collaborative highway safety warning.

[0082] A highway safety warning method based on road-vehicle collaboration, which is applied to a highway safety warning system based on road-vehicle collaboration, and includes the following steps:

[0083] S1. Collect a driving environment data set in a low visibility environment, perform time synchronization, noise filtering and standardization processing on the driving environment data set to form a standardized driving environment data set;

[0084] S2. Based on the standardized driving environment data set, construct a multi-dimensional environment feature model, and extract a driving environment feature set that has a key impact on driving safety through a feature selection method;

[0085] S3. Use the quantum genetic variation algorithm to globally search and optimize the driving environment feature set to obtain an optimized driving environment feature set;

[0086] S4. Based on the optimized driving environment feature set, construct a multi-objective driving warning optimization model;

[0087] S5. The system constructs a multi-objective driving warning optimization model based on the optimized driving environment feature set, and its core goal is to minimize the warning time , maximize the warning accuracy rate , and at the same time reduce driver interference , the system adopts an adaptive weight adjustment mechanism to dynamically adjust the optimization target weights according to real-time traffic flow information and environmental conditions , perform a global search in the multi-objective optimization space through a heuristic solution algorithm to generate an optimal warning strategy , this process ensures that under different visibility conditions and traffic flow densities, the warning plan can not only respond quickly, but also avoid over-alarming, improving the stability and adaptability of the overall warning system;

[0088] S6. The system will use the optimal warning strategy obtained by solving Through a multi-node and multi-path data transmission strategy, it is distributed in real time to the vehicle control system and the driver terminal. The system uses V2X vehicle networking communication to ensure that data can reach the target vehicle quickly and reliably, and triggers a warning instruction within 0.1 seconds. If the vehicle is in the autonomous driving mode, the system directly issues a speed limit, lane change, or braking signal to the vehicle controller; if it is in the manual driving mode, the driver terminal will remind the driver to take corresponding actions in the form of voice alarm + HUD head-up display. This strategy ensures low latency and high reliability, effectively reducing accidents in low visibility environments.

[0089] In this embodiment, S1 includes the following steps:

[0090] S11. Collect the driving environment data set in low visibility environment :

[0091]

[0092] Among them, represents the th driving environment data, is the total number of collected driving environment data, is the time stamp corresponding to the driving environment data, is the numerical information of the driving environment data, is the device identifier of the source of the driving environment data, is the category label of the driving environment data, including vehicle dynamic data collected by on-vehicle sensors, road condition and environment data collected by road infrastructure, meteorological data collected by meteorological equipment, and traffic flow data collected by traffic monitoring systems;

[0093] S12. Perform time synchronization processing on the driving environment data set, construct a unified time reference, and correct the time deviation of each data source through interpolation and time alignment methods to obtain the time-synchronized driving environment data set;

[0094] S13. According to the category labels of the driving environment data in the driving environment data set, perform noise filtering processing on the vehicle dynamic data, road condition and environment data, meteorological data, and traffic flow data respectively, and use an adaptive filtering method to remove outliers to obtain the noise-filtered driving environment data set;

[0095] S14. Use the standardization method to normalize the noise-filtered driving environment data set, unify the numerical scales of various driving environment data, and obtain the standardized driving environment data set ;

[0096] S15. Remove redundant features from the standardized driving environment data set based on the correlation analysis of driving environment data, calculate the correlation matrix of driving environment data to quantify the correlation between driving environment data, and obtain the finally standardized driving environment data set :

[0097] Among them, represents the th driving environment data after final standardization, represents the adjusted standard timestamp, represents the numerical information of the driving environment data after removing redundant features.

[0098] In this embodiment, step S2 includes the following steps:

[0099] S21. Extract features from each piece of driving environment data in the finally standardized driving environment data set to construct a multi-dimensional environmental feature vector :

[0100]

[0101] Among them, represents the meteorological features extracted from the th driving environment data, represents the road condition and environmental features extracted from the th driving environment data, represents the traffic flow features extracted from the th driving environment data, represents the vehicle dynamic features extracted from the th driving environment data;

[0102] S22. Construct a multi-dimensional environmental feature model

[0103]

[0104] S23. Use a feature selection method to process each multi-dimensional environmental feature vector in the multi-dimensional environmental feature model to extract a set of driving environment features that have a key impact on driving safety:

[0105]

[0106] Among them, is the kth key feature selected from the multi-dimensional environmental feature model, represents the set of key feature indexes.

[0107] In this embodiment, step S3 includes the following steps:

[0108] S31. Based on the set of driving environment characteristics perform quantum superposition state encoding to construct a quantum environment characteristic population , where each quantum individual in the quantum environment characteristic population represents a combination of driving environment characteristics: ; where is the quantum population size, represents the th quantum individual, including the combination of driving environment characteristics in low visibility environment, and are both the quantum amplitude coefficients corresponding to the th key feature in the th quantum individual;

[0109] S32. Perform quantum measurement on the quantum environment characteristic population to obtain a set of classical driving environment characteristic individuals where is the th driving environment characteristic individual obtained after measurement, is the set of selected feature indices after measurement;

[0110] Calculate the fitness value for each driving environment characteristic individual : where is the early warning response time calculated based on the driving environment characteristic individual , is the shortest early warning response time, is the early warning accuracy rate of the driving environment characteristic individual , is the calculation result of the road environment complexity of the driving environment characteristic individual , is a dynamic weight parameter, adjusted according to the current low visibility environment characteristics to improve the early warning optimization effect under different environmental conditions;

[0111] S33. Based on the driving dynamic perception, construct a quantum rotation update rule, and combine with the road information flow provided by the vehicle-road collaborative platform to dynamically adjust the search direction of quantum individuals and update the quantum individuals: ; where represents the th quantum individual in the (t + 1)th generation, represents the th quantum individual in the tth generation, It is an adaptive quantum rotation gate based on vehicle-road collaborative environmental factors. is the step size control parameter. is the random perturbation factor of Lévy distribution. is the difference in quantum states between the current quantum individual and the population-optimal individual, which is used to enhance the search ability, and the rotation angle is calculated as follows: Among them, is the quantum rotation step size control parameter. represents the low visibility environment impact factor. is the maximum low visibility environment impact factor. is the road environment complexity. is the maximum road complexity. is the road monitoring information intensity provided by the vehicle-road collaborative platform. is the optimal fitness value in the current population;

[0112] S34. Judge whether the quantum population converges according to the optimization convergence criterion. If the convergence condition is met, output the finally optimized set of driving environment characteristics , otherwise return to step S32 to continue iterative optimization.

[0113] In this embodiment, step S34 includes the following steps:

[0114] S341. Calculate the average fitness of the current population based on the quantum environment characteristic population and the fitness value and the average fitness of the previous generation , and calculate the fitness change rate : Among them, is the quantum population size. is the th driving environment characteristic individual in the th generation of fitness value;

[0115] Practical significance:

[0116] The average fitness calculates the overall fitness of the current quantum population, measuring the overall performance of the population in this round of optimization. The higher the fitness, the better the quality of the optimized early warning strategy.

[0117] The fitness change rate reflects the improvement degree of the current round of optimization compared with the previous round of optimization:

[0118] If is still large, it means that the optimization is still in progress, the algorithm has not converged, and continue iterative optimization.

[0119] If is close to 0, it indicates that the optimization tends to be stable and the optimization can be terminated.

[0120] Actual application scenario:

[0121] During a certain optimization process, the initial average fitness is , and after 3 rounds of optimization, it is improved to , , indicating that the optimization is still effective and the algorithm continues to execute.

[0122] When the optimization reaches the 10th round, is only 0.5%, indicating that the optimization is basically completed, the system enters the convergence state, and finally the optimal driving warning strategy is selected.

[0123] S342. Define the quantum population diversity metric function , and calculate the population mean square deviation based on the amplitude vectors of the current generation of quantum individuals: where, is the characteristic dimension of the quantum individual encoding, and are both the amplitude parameters of the th characteristic dimension in the th quantum individual, and are both the average amplitudes of all quantum individuals in this characteristic dimension, reflects the diversity of the quantum population, and the smaller the value, the more the population tends to converge;

[0124] Practical significance:

[0125] Quantum population diversity reflects the distribution of individuals in the current population:

[0126] If is large, it indicates that the individuals in the population are still relatively dispersed and the optimization is still in the exploration stage, and the optimization search can continue.

[0127] If approaches 0, it indicates that the population has converged near an optimal solution, and at this time the optimization can be terminated.

[0128] Actual application scenario:

[0129] During the warning optimization process, the initial (the population characteristics vary greatly), indicating that the search space is still wide;

[0130] After multiple rounds of optimization, , indicating that the optimization strategy of the population has been stable, the algorithm terminates, and the optimal warning strategy is output.

[0131] S343. Calculate the convergence determination parameter of the quantum population according to the fitness change rate and the quantum population diversity metric function: Wherein, and are the convergence determination weight parameters;

[0132] S344. Define the optimization termination threshold , and when the following conditions are met, it is determined that the quantum population reaches the optimized convergence state: Otherwise, return to step S341 to continue the optimization iteration;

[0133] S345. After meeting the optimization convergence condition, select the individual with the highest fitness from the current quantum population as the finally optimized driving environment feature set: ; wherein, In the t-th optimization iteration, the q-th driving environment feature individual in the population, represents the complete driving environment feature individual population in the t-th optimization iteration, and argmax is to find the fitness function.

[0134] In this embodiment, step S4 includes the following steps:

[0135] S41. Based on the driving environment feature set , define the multi-objective driving warning optimization model . The optimization objectives of the multi-objective driving warning optimization model include minimizing the warning time, maximizing the warning accuracy rate, and optimizing the driving comfort. The objective function of the multi-objective driving warning optimization model is defined as follows:

[0136] Wherein, is the driving warning strategy to be optimized, is the warning time of the driving warning strategy, is the warning accuracy rate of the driving warning strategy, is the degree of interference of the driving warning strategy on the driver, is the multi-objective weight parameter;

[0137] Practical significance:

[0138] This optimization model comprehensively considers three objectives: : Warning time (objective: minimize), : Warning accuracy rate (objective: maximize), : Driving interference degree (objective: minimize)

[0139] Dynamic weight adjustment mechanism Enable the system to adjust and optimize the direction according to the current road environment:

[0140] If the current haze is severe (low visibility), the system will increase the weight, making the system pay more attention to the warning accuracy rate;

[0141] If the current traffic condition is relatively stable, the system will reduce , reducing the interference to the driver.

[0142] Actual application scenario:

[0143] In a certain warning optimization, the traditional method only focuses on the warning time, which may lead to overly frequent alarms and make the driver fatigued. However, through multi-objective optimization, the present invention shortens the warning time, improves the accuracy rate, and reduces driving interference at the same time, thereby enhancing the overall driving experience.

[0144] S42. Define the goal of minimizing the warning time, and the warning time of the driving warning strategy is affected by the set of driving environment characteristics and the low visibility environment impact factor: Among them, represents the detectable distance in the current low visibility environment, represents the transmission speed of the driving warning signal, is the processing time of the vehicle-mounted control system;

[0145] S43. Define the goal of maximizing the warning accuracy rate, and the warning accuracy rate is affected by the set of driving environment characteristics and the road monitoring information provided by the vehicle-road collaborative platform: Among them, represents the number of false alarms or missed alarms in the current driving environment, is the total number of warnings;

[0146] S44. Define the goal of optimizing driving comfort, and the degree of driving comfort interference is affected by the warning frequency and the driver's warning response time: Among them, represents the number of warnings triggered per unit time, represents the average reaction time of the driver to the warning signal;

[0147] S45. Construct an adaptive weight adjustment mechanism and define dynamic adjustment weight parameters Make the dynamic adjustment weight parameters adapt to different low visibility environments: Among them, is the optimized weight parameter of the (t + 1)-th generation, is the weight parameter of the t-th generation, is the learning rate, is the low visibility environment impact factor, is the maximum low visibility environmental impact factor, The intensity of road monitoring information provided to the vehicle-road collaboration platform, It is the maximum road monitoring information intensity.

[0148] Practical significance:

[0149] Automatically adjust weights according to the environment: When low visibility conditions are severe ( ),but , the weight is reduced, and the system pays more attention to dynamic adjustment capabilities; when the road monitoring information is insufficient ( ),but At this time, the system enhances the optimization weight and improves the adaptive ability.

[0150] This mechanism enables the system to proactively increase its focus on safety warnings and improve the accuracy of alarms in bad weather. When monitoring data is more complete, it reduces unnecessary adjustments, reduces false alarms, and improves the driving experience.

[0151] Practical application scenarios:

[0152] On a certain section of mountain highway, dense fog caused low visibility ( ), while the road monitoring capability is low ( ), the system adjusts the weights through this formula, so that the early warning system focuses more on high accuracy and stability, reduces false alarms, and effectively reduces the risk of accidents.

[0153] Embodiment 1:

[0154] At 4 a.m. on XX / XX / XXXX, visibility on a mountainous section of the G72 National Highway dropped sharply to below 30 meters due to sudden fog, causing many heavy trucks to slow down. The section was on a long downhill section. During the winter of the past three years, there had been six serious chain rear-end collisions there, involving more than 50 vehicles, causing significant property losses and casualties. The local traffic management department had installed traditional fixed monitoring equipment and high-speed cameras on the section, but the high-speed cameras were severely affected by the fog and had limited monitoring capabilities, so traditional methods were unable to provide timely and effective safety warnings.

[0155] At 4:12 a.m., the G72 Expressway Vehicle-Road Collaborative Highway Safety Warning System was activated. The system collects real-time driving environment data in low-visibility environments through vehicle sensors, road infrastructure sensors, and traffic flow monitoring systems, including:

[0156] Meteorological information: fog concentration 92%, air humidity 97%, wind speed 5.3m / s; vehicle information: a heavy truck (license plate number "AA") was traveling at a speed of 68km / h, and the distance to the vehicle in front was shortened to 18 meters (the system default safe distance should be more than 45 meters); road information: the road surface in this area is slippery, and the road friction coefficient is reduced to 0.42 (normal is 0.65).

[0157] At 4:13 a.m., the system detected that a heavy truck with license plate number "AA" failed to reduce its speed in a low-visibility environment, and the distance to the vehicle in front was shortened sharply, posing a potential risk of rear-end collision. The system optimized the calculation through the quantum genetic mutation algorithm, conducted a comprehensive analysis of the current road conditions and driving data, and generated an early warning report:

[0158] Warning time: 04:13:27, December 15, 2024; Warning type: Risk of high-speed rear-end collision in low-visibility environment; Trigger reason: The current vehicle distance is lower than the safety threshold and the weather conditions are severe; Recommended measures: It is recommended that the vehicle slow down to 40km / h and maintain a safe distance of at least 50 meters; Driver warning level: Level 3 (medium risk).

[0159] At 4:13:30 in the morning, the system sent the warning information to the driver's smart terminal of Guangdong B8 truck in real time through a multi-node and multi-path data transmission strategy, and pushed it to the on-board terminals of all connected vehicles in the area, and synchronized it to the highway control center. The system also automatically updated the information on the smart electronic traffic sign 2 kilometers ahead: "Heavy fog ahead, the distance between vehicles is too close, please slow down to 40km / h."

[0160] At 4:14 a.m., the driver of the truck with license plate number "AA" received a warning from the on-board terminal and confirmed that there was severe fog ahead. He then took measures to slow down to 42 km / h, while gradually increasing the distance between the vehicle and the vehicle in front.

[0161] At 4:14:50 in the morning, an SUV (license plate number "BB") behind the truck also received the warning message pushed by the system. It actively reduced its speed from 72km / h to 50km / h at a distance of 45 meters behind the truck and adjusted the cruise control mode.

[0162] At 4:15 a.m., the highway control center confirmed that the overall traffic on the section had slowed down, avoiding a series of rear-end collisions caused by low visibility. The system maintained real-time monitoring throughout the process and continued to push road safety status information.

[0163] In order to verify the effectiveness of the present invention, a two-month measured data comparison experiment was carried out on this road section. The experiment was divided into a "traditional method group" and a "method group of the present invention". The key comparison indicators included warning response time, accident rate, false alarm rate, and driver response rate.

[0164] Table 1 Comparison of Key Indicators between the Method Group of the Present Invention and the Traditional Method Group

[0165] Test indicators Traditional method (fixed monitoring + vehicle intelligence) Method of the present invention (vehicle-road collaborative warning) Warning response time (seconds) 5.8 3.9 Accident rate (times / month) 8.2 2.7 False alarm rate (%) 12.5 6.8 Driver response rate (%) 78.3 94.6 Overall warning accuracy rate of the system (%) 83.4 92.8

[0166] Table 1 Test results show that the vehicle-road collaborative highway safety warning system of the present invention shortens the warning response time by 27.3% in low visibility environments, reduces the accident rate by 67.1%, reduces the false alarm rate by 45.6%, and increases the driver response rate by 16.3%, greatly improving highway driving safety.

[0167] This embodiment fully demonstrates the effectiveness of the quantum genetic variation optimization algorithm, multi-objective driving warning optimization model, and multi-node data transmission strategy technology of the present invention in actual highway safety warning. Through real-time data perception, intelligent analysis, and optimization calculation of vehicle-road collaboration, the present invention can provide driving safety warnings faster and more accurately than traditional methods, reduce traffic accidents in low visibility environments, and provide a safer and more efficient solution for intelligent traffic management.

[0168] The present invention uses a quantum genetic variation algorithm to globally search and optimize the driving environment feature set. Through quantum superposition state encoding and quantum rotation update rules, it can efficiently explore the optimal combination of driving environment features in a multi-dimensional feature space. By introducing the probabilistic search mechanism of quantum computing, quantum individuals can explore multiple feature combinations simultaneously and dynamically adjust the search direction, thereby ensuring the global optimality of the optimization results.

[0169] The present invention proposes a multi-objective driving warning optimization model. The multi-objective driving warning optimization model comprehensively considers three core indicators: minimizing the warning time, maximizing the warning accuracy, and optimizing driving comfort, and achieves dynamic balance through an adaptive weight adjustment mechanism. By dynamically adjusting the multi-objective weight parameters, in low visibility environments, it can perform real-time optimization and adjustment according to traffic flow conditions and driving behaviors, ensuring that the warning system can quickly respond to emergencies and avoid causing excessive interference to drivers.

[0170] The present invention adopts a multi-node and multi-path data transmission strategy in the warning information transmission link, which can effectively reduce network congestion and data loss problems, and improve the transmission efficiency and reliability of warning information. In the vehicle-road collaborative system, the timeliness of warning information is crucial. By constructing an adaptive routing optimization mechanism, according to the road information flow provided by the vehicle-road collaborative platform, it dynamically selects the optimal transmission path and realizes data redundancy backup among multiple paths, ensuring that warning information can be transmitted to the driver terminal or vehicle control system with low latency and high reliability.

[0171] As described above, it is only the preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, making equivalent substitutions or changes should be covered within the protection scope of the present invention.

Claims

1. A road safety warning system based on vehicle-road cooperation, characterized in that It includes the following modules: The data acquisition module is used to acquire the driving environment data set in low visibility environment, and perform time synchronization, noise filtering and normalization processing on the driving environment data set to form a normalized driving environment data set; The multi-dimensional environment feature construction module is used to construct a multi-dimensional environment feature model based on the normalized driving environment data set, and extract a set of driving environment features that have a key impact on driving safety through feature selection methods; The quantum genetic variation optimization module is used to globally search and optimize the set of driving environment features by using the quantum genetic variation algorithm to generate an optimized set of driving environment features; Encoding the quantum superposition state based on the set of driving environment characteristics to construct a quantum environment characteristic population , and each quantum individual in the quantum environment characteristic population represents a combination of driving environment characteristics; Perform quantum measurement on the quantum environment characteristic population to obtain a set of classical driving environment characteristic individuals ; For each individual of driving environment characteristics Calculate the fitness value : Among them, is the warning response time calculated based on the individual characteristics of the driving environment and is the shortest warning response time. is the warning accuracy rate of the individual characteristics of the driving environment and is the calculation result of the road environment complexity of the individual characteristics of the driving environment . is the dynamic weight parameter, which is adjusted according to the current low visibility environment characteristics to improve the warning optimization effect under different environmental conditions; ​ Construct a quantum rotation update rule based on driving dynamic perception, combine with the road information flow provided by the vehicle-road collaborative platform, dynamically adjust the search direction of quantum individuals, and update quantum individuals ; Judge whether the quantum population converges according to the optimized convergence criterion. If the convergence condition is satisfied, output the finally optimized set of driving environment characteristics , otherwise return to continue iterative optimization; The multi-objective driving warning optimization module is used to construct a multi-objective driving warning optimization model based on the optimized set of driving environment features, solve the multi-objective driving warning optimization model, and obtain the optimal driving warning strategy; The warning information transmission module is used to transmit the optimal driving warning strategy to the vehicle control system or the driver terminal in a multi-node and multi-path data transmission manner in real time to realize the road-vehicle collaborative highway safety warning.

2. A road safety warning method based on vehicle-road cooperation, applied to the road safety warning system based on vehicle-road cooperation described in claim 1, characterized in that, It includes the following steps S1. Acquire the driving environment data set in low visibility environment, perform time synchronization, noise filtering and normalization processing on the driving environment data set to form a normalized driving environment data set; S2. Based on the normalized driving environment data set, construct a multi-dimensional environment feature model, and extract a set of driving environment features that have a key impact on driving safety through feature selection methods; S3. Use the quantum genetic variation algorithm to globally search and optimize the set of driving environment features to obtain an optimized set of driving environment features; S31. Based on the set of driving environment characteristics Perform quantum superposition state encoding to construct a quantum environment characteristic population , where each quantum individual in the quantum environment characteristic population represents a combination of driving environment characteristics; S32. Perform quantum measurement on the quantum environment characteristic population to obtain a set of classical driving environment characteristic individuals ; For each individual of driving environment characteristics Calculate the fitness value : in, Based on the driving environment characteristics The calculated warning response time, The shortest warning response time is Driving environment characteristic individuals The warning accuracy rate Driving environment characteristic individuals The calculation results of the road environment complexity are as follows: It is a dynamic weight parameter, which is adjusted according to the current low-visibility environmental characteristics to improve the warning optimization effect under different environmental conditions; S33. Based on the dynamic perception of driving, construct a quantum rotation update rule, combine with the road information flow provided by the vehicle-road collaborative platform, dynamically adjust the search direction of quantum individuals, and update quantum individuals ; S34. Determine whether the quantum population converges according to the optimized convergence criterion. If the convergence condition is met, output the finally optimized set of driving environment characteristics , otherwise return to step S32 to continue iterative optimization; S4. Based on the optimized set of driving environment features, construct a multi-objective driving warning optimization model; S5. Solve the multi-objective driving warning optimization model to obtain the optimal driving warning strategy; S6. Transmit the optimal driving warning strategy to the vehicle control system or the driver terminal in a multi-node and multi-path data transmission manner in real time to realize the road-vehicle collaborative highway safety warning.

3. The method for highway safety warning based on vehicle-road cooperation according to claim 2, wherein The S1 includes the following steps: S11. Collect the driving environment dataset in low visibility environment : Among them, represents the th driving environment data, is the total number of collected driving environment data, is the timestamp corresponding to the driving environment data, is the numerical information of the driving environment data, is the device identifier of the driving environment data source, is the category label of the driving environment data, including vehicle dynamic data collected by on-vehicle sensors, road condition and environmental data collected by road infrastructure, meteorological data collected by meteorological equipment, and traffic flow data collected by traffic monitoring systems; S12. Perform time synchronization processing on the driving environment data set, construct a unified time reference, and correct the time deviation of each data source through interpolation and time alignment methods to obtain the time-synchronized driving environment data set; S13. According to the category labels of the driving environment data in the driving environment data set, perform noise filtering processing on the vehicle dynamic data, road condition and environment data, meteorological data and traffic flow data respectively, and use the adaptive filtering method to remove outliers to obtain the noise-filtered driving environment data set; S14. Normalize the driving environment dataset after noise filtering using a standardization method to unify the numerical scales of various driving environment data, and obtain a standardized driving environment dataset ; S15. Remove redundant features from the standardized driving environment dataset based on the correlation analysis of driving environment data, calculate the correlation matrix of driving environment data to quantify the correlation between driving environment data, and obtain the final standardized driving environment dataset : Among them, represents the th driving environment data after final standardization, represents the adjusted standard timestamp, represents the numerical information of the driving environment data after removing redundant features.

4. The method for highway safety warning based on vehicle-road cooperation according to claim 3, characterized in that, The S2 includes the following steps: S21. For each piece of driving environment data in the finally standardized driving environment data set perform feature extraction to construct a multi-dimensional environment feature vector : ​ Among them, represents the meteorological features extracted from the th driving environment data, represents the road condition and environmental features extracted from the th driving environment data, represents the traffic flow features extracted from the th driving environment data, represents the vehicle dynamic features extracted from the th driving environment data; S22. Construct a multi-dimensional environmental feature model ; S23. Use a feature selection method to process each multi-dimensional environmental feature vector in the multi-dimensional environmental feature model to extract a set of driving environment features that have a key impact on driving safety : Among them, is the k-th key feature selected from the multi-dimensional environmental feature model, represents the set of key feature indexes.

5. The highway safety warning method based on vehicle-road cooperation according to claim 4, characterized in that, The S3 includes the following steps: S31. Based on the set of driving environment characteristics Perform quantum superposition state encoding to construct a quantum environment characteristic population , where each quantum individual in the quantum environment characteristic population represents a combination of driving environment characteristics: ； Among them, is the quantum population size, represents the th quantum individual, including the combination of driving environment characteristics in a low visibility environment, and are both the quantum amplitude coefficients corresponding to the th key feature in the th quantum individual; S32. Perform quantum measurement on the quantum environment characteristic population to obtain a set of classical driving environment characteristic individuals : Among them, is the th driving environment feature individual obtained after measurement, is the set of feature indexes selected after measurement; For each individual of driving environment characteristics Calculate the fitness value : Among them, is the warning response time calculated based on the individual driving environment characteristics ; is the shortest warning response time ; is the warning accuracy rate of the individual driving environment characteristics ; is the calculation result of the road environment complexity of the individual driving environment characteristics is a dynamic weight parameter, which is adjusted according to the current low visibility environment characteristics to improve the warning optimization effect under different environmental conditions; S33. Construct a quantum rotation update rule based on driving dynamic perception, combine the road information flow provided by the road-vehicle collaborative platform, dynamically adjust the search direction of quantum individuals, and update quantum individuals: ； Among them, represents the -th quantum individual in the (t + 1)-th generation, represents the -th quantum individual in the t-th generation, is an adaptive quantum rotation gate based on vehicle-road collaborative environment factors, is the step size control parameter, is the random perturbation factor of Lévy distribution, is the difference in quantum states between the current quantum individual and the population's optimal individual, used to enhance the search ability. The rotation angle is calculated as follows: Among them, is the quantum rotation step control parameter, represents the low visibility environment impact factor, is the maximum low visibility environment impact factor, is the road environment complexity, is the maximum road complexity, is the intensity of road monitoring information provided by the vehicle-road collaborative platform, is the optimal fitness value in the current population; S34. Determine whether the quantum population converges according to the optimized convergence criterion. If the convergence condition is met, output the finally optimized set of driving environment characteristics , otherwise return to step S32 to continue iterative optimization.

6. The highway safety warning method based on vehicle-road cooperation according to claim 5, characterized in that, The S34 includes the following steps: S341. Calculate the fitness mean of the current population based on the quantum environment characteristic population and fitness values and the fitness mean of the previous generation , and calculate the fitness change rate : Among them, is the quantum population size, is the th individual of the driving environment characteristics, at the th generation fitness value; S342. Define the quantum population diversity metric function , and calculate the population mean square deviation based on the amplitude vectors of the current-generation quantum individuals: Among them, is the characteristic dimension of the quantum individual encoding, and both are the amplitude parameters of the th characteristic dimension in the th quantum individual, and both are the average amplitudes of all quantum individuals in this characteristic dimension, reflecting the diversity of the quantum population. The smaller the value, the more convergent the population tends to be; S343. Calculate the convergence determination parameter of the quantum population according to the fitness change rate and the quantum population diversity metric function : Among them, and are convergence determination weight parameters; S344. Define the optimization termination threshold When the following conditions are met, it is determined that the quantum population has reached the optimization convergence state: Otherwise, return to step S341 to continue the optimization iteration; S345. After the optimization convergence condition is satisfied, select the individual with the highest fitness from the current quantum population as the finally optimized driving environment feature set : ； Among them, in the t-th optimization iteration, the q-th driving environment feature individual in the population represents the complete population of driving environment feature individuals in the t-th optimization iteration, and argmax is to find the fitness function.

7. The method for highway safety warning based on vehicle-road cooperation according to claim 6, characterized in that, The S4 includes the following steps: S41. Based on the set of driving environment characteristics , define a multi-objective driving warning optimization model , the optimization objectives of the multi-objective driving warning optimization model include minimizing the warning time, maximizing the warning accuracy, and optimizing the driving comfort. The objective function of the multi-objective driving warning optimization model is defined as follows: Among them, is the driving warning strategy to be optimized, is the warning time of the driving warning strategy, is the warning accuracy rate of the driving warning strategy, is the degree of interference of the driving warning strategy to the driver, is the multi-objective weight parameter; S42. Define the warning time minimization objective, which is the warning time of the driving warning strategy It is affected by the set of driving environment characteristics and the influencing factors of low visibility environment Among them, represents the detectable distance in the current low visibility environment, represents the transmission speed of the driving warning signal, is the processing time of the vehicle control system; S43. Define the goal of maximizing the warning accuracy rate, and the warning accuracy rate is affected by the set of driving environment characteristics and the road monitoring information provided by the vehicle-road collaborative platform: Among them, represents the number of false alarms or missed alarms in the current driving environment, is the total number of warnings; S44. Define the driving comfort optimization goal and the degree of driving comfort interference Affected by the warning frequency and the driver's warning response time: Among them, represents the number of warnings triggered per unit time, represents the average reaction time of the driver to the warning signal; S45. Build an adaptive weight adjustment mechanism and define dynamic weight adjustment parameters Make the dynamic weight adjustment parameters adapt to different low visibility environments: Among them, is the optimized weight parameter of the (t + 1)-th generation, is the weight parameter of the t-th generation, is the learning rate, is the low visibility environment impact factor, is the maximum low visibility environment impact factor, is the intensity of road monitoring information provided by the vehicle-road collaborative platform, is the maximum intensity of road monitoring information.

Images