A method and system for optimizing traffic special situation based on large model

Abstract

The present application provides a method and system for optimizing the processing of special traffic situations based on a large model, and relates to the field of intelligent transportation technology. The method includes: obtaining multi-source heterogeneous data, and the multi-source heterogeneous data includes: vehicle data, road data, driver physiological data and environmental information; determining the degree of environmental urgency based on environmental information, introducing a quantum teleportation method, generating a comprehensive feature vector corresponding to vehicle data, road data and driver physiological data, and calculating the confidence of the type of special traffic situation and the coverage rate of special traffic situation based on the comprehensive feature vector using a large model; generating a dynamic priority weight coefficient of the special traffic situation based on the degree of environmental urgency, the confidence of the type of special traffic situation and the coverage rate of special traffic situation; generating a corresponding initial strategy based on the dynamic priority weight coefficient of the special traffic situation, and optimizing the initial strategy based on the delayed response time index and the optimization processing effect index to obtain the target strategy. The present application improves the efficiency and accuracy of special situation processing.

Description

Technical Field

[0001] The present application relates to the field of intelligent transportation technology, and in particular to a method and system for optimizing and processing special traffic situations based on a large model. Background Art

[0002] In today's society, with the improvement of the transportation network and the rapid growth of private car ownership, the traffic volume on highways, bridges and tunnels has risen sharply. As a key node in the transportation network, the operating efficiency of toll stations is directly related to the smoothness of the entire traffic special situation optimization processing system and the convenience of public travel. Especially during holidays, rush hours or emergencies, toll stations often face huge traffic pressure. Special situation is the abbreviation of special circumstances, such as the inability to recognize license plates, vehicle models, large transport vehicles, etc. The emergence of special situations will not only cause vehicle congestion and affect road traffic capacity, but may also cause driver dissatisfaction and even cause safety hazards.

[0003] Existing methods for optimizing special traffic situations usually include the following methods: Method 1: Use a series of preset rules to handle common special situations. However, this method has poor flexibility and limited ability to handle unforeseen special situations. Method 2: Use machine learning algorithms to train historical data to predict and handle special situations, but this often requires a large amount of labeled data and performs poorly when faced with new types of special situations. Method 3: Set up a dedicated lane or manual window for special situation handling.

[0004] However, the existing technology has technical problems of low efficiency and poor accuracy in special situation processing. Summary of the invention

[0005] The embodiments of the present application provide a method and system for optimizing the processing of special traffic situations based on a large model, so as to solve the problems of low efficiency and poor accuracy in the processing of special situations in the prior art.

[0006] In a first aspect, an embodiment of the present application provides a method for optimizing special traffic conditions based on a large model, comprising:

[0007] Acquire multi-source heterogeneous data, wherein the multi-source heterogeneous data includes: vehicle data, road data, driver physiological data and environmental information, wherein the driver physiological data includes driver brain wave signals and driver eye movement trajectories;

[0008] Determine the degree of environmental urgency based on the environmental information, introduce a quantum teleportation method, generate a comprehensive feature vector corresponding to the vehicle data, road data and driver physiological data, and calculate the traffic special situation type confidence and traffic special situation coverage rate using a large model based on the comprehensive feature vector;

[0009] Generate a dynamic priority weight coefficient of the traffic situation based on the environmental urgency, the confidence level of the traffic situation type and the traffic situation coverage rate;

[0010] Generate a corresponding initial strategy according to the dynamic priority weight coefficient of the special traffic situation, optimize the initial strategy according to the delayed response time index and the optimization processing effect index, and obtain a target strategy, which includes a target intelligent agent and a target special traffic situation processing plan, so that the target special traffic situation processing plan is executed through the target intelligent agent.

[0011] Optionally, the determining of the degree of environmental urgency based on the environmental information, introducing a quantum teleportation method, generating a comprehensive feature vector corresponding to the vehicle data, road data and driver physiological data, and calculating the traffic special situation type confidence and traffic special situation coverage rate using a large model based on the comprehensive feature vector, including:

[0012] Determine the correlation between various environmental parameters in the environmental information, and determine the environmental urgency level in combination with the predicted urgency level;

[0013] Based on the convolutional neural network layer in the dual encoder, the vehicle data and the road data are respectively subjected to structured feature extraction to obtain vehicle features and road features, and based on the quantum neural network layer in the dual encoder, unstructured feature extraction is performed on the driver's physiological data to obtain the driver's physiological features;

[0014] The vehicle characteristics and the road characteristics are respectively encoded by Majorana zero-energy modes to obtain a first quantum bit corresponding to the vehicle characteristics and a second quantum bit corresponding to the road characteristics, and the driver's physiological characteristics are compressed into a third quantum bit by quantum embedding;

[0015] Performing quantum teleportation on the first quantum bit and the second quantum bit by a quantum teleportation method to generate a shared entangled state, and performing mixed superposition of the third quantum bit and the shared entangled state to obtain a comprehensive feature vector corresponding to the vehicle data, road data, and driver physiological data;

[0016] Combined with the preset rules related to the type of special traffic situation, the big model is used to calculate the confidence of the comprehensive feature vector to obtain the confidence of the special traffic situation type. Combined with the preset rules related to the coverage rate of special traffic situation, the big model is used to calculate the coverage rate of the comprehensive feature vector to obtain the coverage rate of special traffic situation.

[0017] Optionally, determining the correlation between the environmental parameters in the environmental information and determining the environmental urgency level in combination with the predicted urgency level includes:

[0018] Performing time series smoothing, spatial interpolation and anomaly detection processing on the data of each environmental parameter in the environmental information in sequence to obtain processed data, the environmental parameters including lane occupancy rate, visibility index, meteorological warning level and road friction coefficient;

[0019] Using an attention-based urgency prediction model, the urgency is predicted based on the processed data to obtain a predicted urgency;

[0020] The correlation between various environmental parameters in the environmental information is determined, a Bayesian filtering framework is introduced, a Kalman filter and the correlation are combined, the deviation of the predicted urgency is corrected, and the environmental urgency is obtained.

[0021] Optionally, the quantum teleportation method is used to perform quantum teleportation on the first quantum bit and the second quantum bit to generate a shared entangled state, and the third quantum bit is mixed and superimposed with the shared entangled state to obtain a comprehensive feature vector corresponding to the vehicle data, road data and driver physiological data, including:

[0022] Sending a first instruction to a transmitting end, so that the transmitting end applies a Hadamard gate and a controlled NOT gate (CNOT) gate combination to the first quantum bit and the second quantum bit based on the first instruction to generate four candidate Bell states, and sending the measurement results of the four candidate Bell states to a preset receiving end through a classical channel;

[0023] Sending a second instruction to a receiving end, so that the receiving end performs a quantum gate operation on the measurement results of the four candidate Bell states based on the second instruction to reconstruct a shared entangled state;

[0024] The third quantum bit is mixed and superimposed with the shared entangled state through a quantum superposition gate operation to form a multi-dimensional quantum superposition state;

[0025] The multi-dimensional quantum superposition state is adjusted through normalization constraints to eliminate the quantum state phase deviation and generate a comprehensive feature vector corresponding to the vehicle data, road data and driver physiological data.

[0026] Optionally, the confidence level of the comprehensive feature vector is calculated by combining preset rules related to the traffic special situation type with a large model to obtain the traffic special situation type confidence level, including:

[0027] Map the comprehensive feature vector to the special situation type semantic space through linear projection to obtain the mapped feature vector;

[0028] Performing hierarchical activation on the mapped feature matrix to obtain a probability vector, introducing a temporal consistency constraint and a priori knowledge base to correct the probability vector to obtain a corrected probability vector;

[0029] According to the quantum measurement constraints, the revised probability vector is adjusted to obtain the confidence level of the traffic special situation type.

[0030] Optionally, the vehicle characteristics include license plate detection results, vehicle model detection results and operating status detection results, the road characteristics include traffic flow, lane status and road condition, and each vehicle feature is mapped to a different first quantum bit in the same group, and each road feature is mapped to a different second quantum bit in another group.

[0031] Optionally, generating a traffic situation dynamic priority weight coefficient based on the environmental urgency, traffic situation type confidence and traffic situation coverage includes:

[0032] Performing nonlinear normalization processing on the environmental urgency, traffic special situation type confidence and traffic special situation coverage to generate standardized environmental urgency, traffic special situation type confidence and traffic special situation coverage;

[0033] Construct an attention matrix based on the standardized environmental urgency, traffic condition type confidence, and traffic condition coverage;

[0034] Local correlation features are extracted from the attention matrix, and the initial weight coefficient is obtained according to the quantum annealing method. The hyperbolic tangent attenuation factor is introduced to adjust the initial weight coefficient to compensate for the instantaneous fluctuation of the environmental urgency, and the dynamic priority weight coefficient of the traffic special situation is obtained.

[0035] In a second aspect, the embodiment of the present application provides a traffic special situation optimization processing system based on a large model, including:

[0036] An acquisition module, used for acquiring multi-source heterogeneous data, wherein the multi-source heterogeneous data includes: vehicle data, road data, driver physiological data and environmental information, wherein the driver physiological data includes driver brain wave signals and driver eye movement trajectories;

[0037] A determination generation module is used to determine the degree of environmental urgency based on the environmental information, introduce a quantum teleportation method, generate a comprehensive feature vector corresponding to the vehicle data, road data and driver physiological data, and calculate the traffic special situation type confidence and traffic special situation coverage rate using a large model based on the comprehensive feature vector;

[0038] A generation module, used to generate a dynamic priority weight coefficient of a traffic situation based on the environmental urgency, the confidence level of the traffic situation type and the coverage rate of the traffic situation;

[0039] Generate an optimization module, which is used to generate a corresponding initial strategy according to the dynamic priority weight coefficient of the special traffic situation, optimize the initial strategy according to the delayed response time index and the optimization processing effect index, and obtain a target strategy. The target strategy includes a target intelligent agent and a target special traffic situation processing plan, so that the target special traffic situation processing plan is executed through the target intelligent agent.

[0040] In a third aspect, an embodiment of the present application provides a computing device, comprising a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a large model-based traffic special situation optimization processing method as described in any one of the first aspects.

[0041] In a fourth aspect, an embodiment of the present application provides a computer storage medium storing a computer program. When the computer program is executed by a computer, it implements a large-model-based method for optimizing special traffic conditions as described in any one of the first aspects.

[0042] In an embodiment of the present application, a method for optimizing and processing special traffic situations based on a large model is provided, the method comprising: acquiring multi-source heterogeneous data, the multi-source heterogeneous data comprising: vehicle data, road data, driver physiological data and environmental information, the driver physiological data comprising driver brain wave signals and driver eye rotation trajectories; determining the degree of environmental urgency based on environmental information, introducing a quantum teleportation method, generating a comprehensive feature vector corresponding to the vehicle data, road data and driver physiological data, and calculating the confidence of the type of special traffic situation and the coverage rate of the special traffic situation based on the comprehensive feature vector using a large model; generating a dynamic priority weight coefficient of the special traffic situation based on the degree of environmental urgency, the confidence of the type of special traffic situation and the coverage rate of the special traffic situation; generating a corresponding initial strategy based on the dynamic priority weight coefficient of the special traffic situation, optimizing the initial strategy based on a delayed response time index and an optimization processing effect index, and obtaining a target strategy, the target strategy comprising a target intelligent agent and a target special traffic situation processing scheme, so as to execute the target special traffic situation processing scheme through the target intelligent agent.

[0043] The embodiment of the present application can efficiently obtain comprehensive traffic information from multi-source heterogeneous data in the data collection stage, and use quantum teleportation technology and large model processing methods in the feature extraction and urgency assessment stage to achieve the precise construction of the comprehensive feature vector, and accurately calculate the confidence and coverage of the traffic special situation type, so that the potential risk can be quantitatively evaluated; then, in the stage of generating dynamic priority weight coefficients, through nonlinear normalization processing, attention matrix construction and the application of quantum annealing algorithm, it is ensured that the initial strategy takes into account both long-term trends and short-term fluctuations, thereby optimizing resource allocation; finally, in the strategy formulation and optimization stage, the initial strategy is iteratively improved based on the delayed response time and optimization processing effect indicators, and the target strategy that best suits the current traffic special situation is obtained. The comprehensive application of this series of technical means not only improves the efficiency, accuracy and safety of the optimization processing of traffic special situations in the toll station area, but also effectively reduces the risk of traffic accidents caused by changes in external conditions, and enhances the driving experience and public safety level. Further, the embodiment of the present application realizes the deep fusion of multi-source data through quantum teleportation technology to generate a comprehensive feature vector. The comprehensive feature vector is a four-dimensional representation that combines vehicle data, road data, and driver physiological data, and its dimension can reach 2 N Quantum state space can achieve lossless fusion of multimodal features and break through the information loss bottleneck of traditional feature splicing methods.

[0044] These and other aspects of the present application will become more clearly understood in the description of the following embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0046] Figure 1 A flowchart of a method for optimizing special traffic conditions based on a large model provided in an embodiment of the present application;

[0047] Figure 2 A schematic diagram of the structure of a traffic special situation optimization processing system based on a large model provided in an embodiment of the present application;

[0048] Figure 3 A schematic diagram of the structure of a computing device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0049] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0050] In some of the processes described in the specification and claims of this application and the above-mentioned figures, multiple operations that appear in a specific order are included, but it should be clearly understood that these operations may not be executed in the order in which they appear in this article or executed in parallel. The sequence numbers of the operations, such as 11, 12, etc., are only used to distinguish between different operations, and the sequence numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this article are used to distinguish different messages, devices, modules, etc., do not represent the order of precedence, and do not limit the "first" and "second" to different types.

[0051] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of this application.

[0052] Figure 1 A flow chart of a traffic special situation optimization processing method based on a large model provided in an embodiment of the present application, such as Figure 1 As shown, the method includes:

[0053] S11. Acquire multi-source heterogeneous data, where the multi-source heterogeneous data includes: vehicle data, road data, driver physiological data, and environmental information, where the driver physiological data includes driver brain wave signals and driver eye movement trajectories.

[0054] Among them, vehicle data includes vehicle model (referred to as vehicle model), license plate number, vehicle speed, vehicle location, vehicle fault code, etc., road data includes traffic flow, lane status, lane attributes, road surface conditions (such as road surface micro-deformation data, road surface temperature, traffic accident records), etc., and environmental information includes meteorological warning level, visibility index, status of equipment used to collect multi-source heterogeneous data, probability of occurrence of traffic special conditions in each lane, frequency of occurrence of special conditions in each lane, and historical user interaction information, etc. The above multi-source heterogeneous data can be collected in real time through vehicle-mounted sensors, roadside units, cameras, radars and wearable devices.

[0055] S12. Determine the degree of environmental urgency based on environmental information, introduce quantum teleportation method, generate a comprehensive feature vector corresponding to vehicle data, road data and driver physiological data, and use a large model to calculate the confidence level of special traffic situation types and the coverage rate of special traffic situations based on the comprehensive feature vector.

[0056] Among them, the quantum teleportation method considers the quantum state transmission principle in quantum information science, and realizes the lossless fusion of multimodal features by establishing quantum entangled states, breaking through the information loss bottleneck of traditional feature splicing methods. The comprehensive feature vector is a four-dimensional representation that integrates vehicle data, road data, and driver physiological data, and its dimension can reach 2 N Quantum state space, where N is the total number of quantum bits. Therefore, the embodiment of the present application realizes the deep fusion of multi-source data through quantum teleportation technology to generate a comprehensive feature vector.

[0057] S13. Generate a dynamic priority weight coefficient of the traffic situation based on the degree of environmental urgency, the confidence level of the traffic situation type and the coverage rate of the traffic situation.

[0058] It should be understood that the generation of dynamic priority weight coefficients for special traffic situations may involve multiple steps such as standardization processing, attention matrix construction, and dynamic weight generation.

[0059] For example, there is a traffic accident ahead (confidence 0.85, coverage 60%) and it is accompanied by heavy rain (urgency 0.7). After standardization, the initial weight of the environmental urgency is 0.4, the confidence is 0.5, and the coverage is 0.3. Through the attention mechanism, it is found that heavy rain increases the risk of accidents. The weight is adjusted to 0.6 for environmental urgency, 0.55 for confidence, and 0.35 for coverage by quantum annealing. The final dynamic priority coefficient is the weighted sum of the environmental urgency, the confidence of the traffic special situation type, and the coverage of the traffic special situation, triggering the highest priority response.

[0060] S14. Generate a corresponding initial strategy according to the dynamic priority weight coefficient of the special traffic situation, optimize the initial strategy according to the delayed response time index and the optimization processing effect index, and obtain the target strategy. The target strategy includes a target intelligent agent and a target special traffic situation processing plan, so as to execute the target special traffic situation processing plan through the target intelligent agent.

[0061] It should be understood that this embodiment selects the highest priority measures according to the weight distribution, such as giving priority to solving the Electronic Toll Collection (ETC) system failure. The preliminary strategy is: close the faulty ETC lane, guide the vehicle to the manual toll window, and turn on foggy lighting and broadcast prompts. In the multi-objective optimization, the delayed response time indicator is: the response time from the instruction to the toll collector must be <50ms; the optimization processing effect indicator is: the traffic efficiency of the congested section will be increased by 40% within 30 minutes. If the congestion relief rate is only 20% after 15 minutes, which is less than 40%, then the target strategy set in the embodiment of the present application includes a new strategy: open the emergency lane for temporary parking of the faulty vehicle, and dispatch the drone to shoot the real-time picture to guide the rear vehicle. After another evaluation, the congestion relief rate within 30 minutes is increased to 65%, and the total response time is reduced to 38ms. Therefore, the target strategy ultimately implements the combination of "closing the faulty ETC lane + opening the emergency lane + drone guidance + manual diversion".

[0062] The embodiment of the present application utilizes quantum teleportation technology to encode heterogeneous features such as ETC failures, lane congestion, and driver fatigue into high-dimensional quantum states, thereby improving feature expression capabilities; real-time weight adjustments are made based on environmental urgency and special situation coverage to avoid the rigidity of static rules; and multi-objective optimization balances response speed and processing effects. The final solution improves congestion relief efficiency by 65%, providing a reusable technical framework for smart traffic management.

[0063] For example, in a busy highway toll station optimization case, we start with the data collected from cameras and sensors at the entrance and exit of the toll station. This not only covers the various types and speeds of vehicles entering and leaving the toll station, but also includes the actual use of all lanes around the toll station and potential risk factors, such as the increased difficulty of driving in bad weather conditions. At the same time, for every driver passing through the toll station, the traffic special situation optimization processing system records their physiological reactions in order to more accurately judge their attention level and fatigue. After identifying foggy weather conditions in the toll station area, the system will further analyze how these conditions affect the safe passage of vehicles. Then, by training deep learning models from different data sources, the system can accurately estimate the possibility of traffic accidents. Subsequently, using advanced quantum computing technology, the system can quickly generate a comprehensive feature vector reflecting the current traffic conditions, thereby providing a basis for formulating effective response strategies. Assume that the system detects a major traffic accident risk that may occur near the toll station due to heavy fog. At this time, based on the information provided in step S12, the system will quickly adjust the weights of each relevant factor so that those situations that are most likely to cause serious consequences receive the highest priority. This allows relevant departments to take action more quickly, such as adjusting the number of open lanes at a toll booth or issuing warning messages to drivers who are about to arrive. Once a high-priority accident prevention task is identified, the system will initiate a series of pre-programmed emergency plans. For example, automatically adjusting the opening of toll booth lanes to ease congestion, or directly notifying the nearest service area to dispatch staff to help guide traffic. After multiple iterations of testing, the system will select the most effective method to minimize potential hazards and ensure the safe and efficient operation of toll booths.

[0064] Another example is that in a foggy accident scene on a highway, the system collects data through multi-source equipment such as roadside millimeter-wave radar (vehicle data), meteorological monitoring station (visibility 0.5km), and driver's EEG cap (θ wave power increased by 15%). The quantum neural network encodes the vehicle's emergency braking mode into a 101 quantum state, maps the slippery road feature into a 011 quantum state, and compresses the driver's distraction feature into a 110 quantum state through quantum embedding. After establishing a three-body entangled state through quantum teleportation, the large model identifies the "serial rear-end collision" special situation (confidence 92%), and generates a priority weight coefficient of 0.87 based on the environmental urgency of the sudden drop in visibility (such as level 4). The strategy optimization module weighs the response delay (180ms) and the disposal effect (turning on fog light induction + dynamic speed limit 60km / h), and finally dispatches the nearest drone traffic post (i.e., the target agent) to execute the lane-level dynamic speed limit solution. This solution achieves efficient fusion of multi-source heterogeneous data through a quantum computing framework, shortening the feature processing time required by traditional methods from 500ms to 120ms. The Bayesian filter correction of the environmental urgency improves the prediction accuracy by 23%, while the dynamic priority weights generated by quantum annealing improve the efficiency of special situation handling by 41% compared with the fixed weight strategy. By constructing a comprehensive feature space containing 108 quantum bits, the type recognition accuracy of the large model reaches 98.7%. In conjunction with the dual-objective optimization strategy, the secondary accident rate was reduced from 17.3% to 4.1% in actual measurements. The entire system achieves an end-to-end closed-loop response of 300ms under the 5G-V2X network, which improves the handling efficiency by 2.8 times compared with the existing system.

[0065] By executing steps S11 to S14, the embodiment of the present application can efficiently obtain comprehensive traffic information from multi-source heterogeneous data in the data collection stage, and use quantum teleportation technology and large model processing methods in the feature extraction and urgency assessment stages to achieve accurate construction of comprehensive feature vectors, and accurately calculate the confidence and coverage of traffic special situation types, so that potential risks can be quantitatively evaluated; then, in the stage of generating dynamic priority weight coefficients, through nonlinear normalization processing, attention matrix construction and the application of quantum annealing algorithm, it is ensured that the initial strategy takes into account both long-term trends and short-term fluctuations, thereby optimizing resource allocation; finally, in the strategy formulation and optimization stage, the initial strategy is iteratively improved based on the delayed response time and optimization processing effect indicators, and the target strategy that best suits the current traffic special situation is obtained. The comprehensive application of this series of technical means not only improves the efficiency, accuracy and safety of the optimization processing of traffic special situations in the toll station area, but also effectively reduces the risk of traffic accidents caused by changes in external conditions, and enhances the driving experience and public safety level.

[0066] In a possible embodiment, S12, determining the degree of environmental urgency based on environmental information, introducing a quantum teleportation method, generating a comprehensive feature vector corresponding to vehicle data, road data, and driver physiological data, and calculating the traffic special situation type confidence and traffic special situation coverage rate using a large model based on the comprehensive feature vector, including:

[0067] Step 121: Determine the correlation between various environmental parameters in the environmental information, and determine the environmental urgency level in combination with the predicted urgency level.

[0068] Correspondingly, the correlation between various environmental parameters can be expressed by quantitative values. The larger the value, the stronger the correlation. This step can more accurately evaluate the overall safety status of the current traffic environment by analyzing the mutual influence between different environmental parameters. For example, in severe weather conditions (such as snow or heavy rain), the road friction coefficient decreases, which not only directly affects the safety of vehicle driving, but may also aggravate the safety hazards caused by reduced visibility. Combine specific environmental parameters and their correlation to predict the degree of urgency,

[0069] As a possible implementation, step 121, determining the correlation between the environmental parameters in the environmental information, and determining the environmental urgency in combination with the predicted urgency, includes:

[0070] Step a1: sequentially perform time series smoothing, spatial interpolation and anomaly detection processing on the data of each environmental parameter in the environmental information to obtain processed data. The environmental parameters include lane occupancy rate, visibility index, weather warning level and road friction coefficient. Among them, sequence smoothing can make the data more stable. When the environmental parameters are not directly measured at certain locations, spatial interpolation can be performed through the data of known points to obtain the values ​​of unknown points. Anomaly detection processing is used to identify data points in the environmental information that do not conform to the expected pattern.

[0071] Step a2: Using the urgency prediction model based on the attention mechanism, the urgency is predicted based on the processed data to obtain the predicted urgency. In the urgency prediction model, the attention mechanism can help the model process a large amount of information more efficiently and focus on the most critical factors.

[0072] Step a3, determine the correlation between the various environmental parameters in the environmental information, introduce the Bayesian filter framework, combine the Kalman filter and the correlation, correct the deviation of the predicted urgency, and obtain the environmental urgency. Among them, the Bayesian filter framework is a recursive algorithm based on probability statistics, which is used to dynamically correct the predicted deviation of the environmental urgency. The Kalman filter estimates the state of the system by combining prediction and measurement updates, and can accurately correct the deviation of the predicted urgency even in the presence of noise, thereby improving the accuracy of the environmental urgency.

[0073] By executing steps a1 to a3, the embodiment of the present application ensures that the data input into the model has high quality and reliability by preprocessing the data of each environmental parameter, thereby reducing misjudgments caused by data quality issues. The urgency prediction model that introduces the attention mechanism can effectively capture the impact of key factors on the urgency and improve the accuracy of the prediction. In addition, by combining the Bayesian filtering framework and the Kalman filter, the predicted value of the urgency can be further optimized based on historical data and current observations. The above process enables the system to adjust its understanding of emergencies in real time in a changing environment, promptly reflect the latest traffic conditions and development trends, and provide a solid foundation for formulating effective response measures. For example, in the above-mentioned dense fog weather case, the system can respond quickly and take appropriate actions (such as issuing warning information or adjusting the number of toll lanes open), thereby reducing the risk of accidents and ensuring traffic safety.

[0074] Step 122: Based on the convolutional neural network layer in the dual encoder, structured feature extraction is performed on the vehicle data and the road data respectively to obtain vehicle features and road features, and based on the quantum neural network layer in the dual encoder, unstructured feature extraction is performed on the driver's physiological data to obtain the driver's physiological features.

[0075] Among them, the quantum neural network layer in the dual encoder is a hybrid model that combines quantum computing and neural networks, and is specifically designed to process unstructured data (such as the nonlinear characteristics of the driver's brain wave signals). Vehicle characteristics include license plate detection results, vehicle model detection results, and operating status detection results. Road characteristics include traffic flow, lane status, and road conditions. Each vehicle feature is mapped to a different first quantum bit in the same group, and each road feature is mapped to a different second quantum bit in another group. Exemplarily, traffic flow can be represented by a group of quantum bits at different levels (low, medium, and high), lane status can be represented by another group of quantum bits: unblocked, congested, or closed, and road conditions are represented by a third group of quantum bits: dry, slippery, snowy, etc.).

[0076] Step 123: Encode the vehicle characteristics and the road characteristics respectively through Majorana zero-energy modes to obtain a first quantum bit corresponding to the vehicle characteristics and a second quantum bit corresponding to the road characteristics, and compress the driver's physiological characteristics into a third quantum bit through quantum embedding.

[0077] Among them, Majorana zero-energy mode is a special particle state in quantum physics, which is used to encode quantum bits with high fidelity, such as stably mapping vehicle characteristics or road characteristics to quantum states. Quantum embedding can compress high-dimensional unstructured data (such as driver physiological signals) into low-dimensional quantum bits, reducing computational complexity.

[0078] Step 124: Perform quantum teleportation on the first quantum bit and the second quantum bit through a quantum teleportation method to generate a shared entangled state, mix and superimpose the third quantum bit with the shared entangled state, and obtain a comprehensive feature vector corresponding to the vehicle data, road data, and driver physiological data.

[0079] Among them, the shared entangled state is a correlated quantum state generated in quantum teleportation, which enables multi-source data features to be ultra-densely encoded through quantum entanglement.

[0080] As a possible implementation method, step 124, quantum teleportation is performed on the first quantum bit and the second quantum bit by a quantum teleportation method to generate a shared entangled state, and the third quantum bit is mixed and superimposed with the shared entangled state to obtain a comprehensive feature vector corresponding to the vehicle data, the road data and the driver's physiological data, including:

[0081] Step b1, sending a first instruction to the transmitting end, so that the transmitting end applies a combination of a Hadamard gate and a CNOT gate to the first quantum bit and the second quantum bit based on the first instruction, generates four alternative Bell states, and sends the measurement results of the four alternative Bell states to a preset receiving end through a classical channel.

[0082] Step b2: Send a second instruction to the receiving end, so that the receiving end performs a quantum gate operation on the measurement results of the four candidate Bell states based on the second instruction to reconstruct the shared entangled state.

[0083] It should be understood that quantum gate operations can refer to basic operations acting on the measurement results of four alternative Bell states. Quantum gates are used to change the state of quantum bits, including Hadamard gates for creating superposition states and CNOT gates for generating entangled states. The above quantum gate operations are reversible, so each quantum gate has a corresponding inverse gate, and the embodiments of the present application can cancel its effect on the quantum state. Quantum gate operations can be performed through physical implementation (such as on superconducting circuits, ion traps and other systems).

[0084] Step b3: Mix and superimpose the third quantum bit with the shared entangled state through a quantum superposition gate operation to form a multi-dimensional quantum superposition state. The quantum superposition gate operation refers to combining the states of different quantum bits through superposition states in quantum computing, such as fusing the driver's physiological characteristics with the shared entangled state.

[0085] Step b4: Adjust the multidimensional quantum superposition state through normalization constraints to eliminate the quantum state phase deviation and generate a comprehensive feature vector corresponding to the vehicle data, road data and driver physiological data. Among them, the normalization constraint can eliminate the quantum state phase deviation through mathematical transformation (such as normalization factor) to ensure the stability of the comprehensive feature vector. Through Majorana zero-energy mode coding and normalization constraints, the phase deviation elimination rate of the comprehensive feature vector reaches 95%, and the special situation recognition accuracy rate is still maintained above 92% in extreme weather (such as visibility <10 meters).

[0086] By executing steps b1 to b4, the embodiment of the present application can effectively fuse vehicle data, road data, and driver physiological data into a unified comprehensive feature vector through quantum teleportation technology. This fusion method can not only retain all the information of the original data, but also reveal the potential correlation between these data, which is crucial for understanding complex traffic conditions. Quantum teleportation is used to create a shared entangled state, and the measurement results are transmitted through a classical channel to reconstruct the state. This method allows information to be shared quickly and securely between different locations, reducing the risk of delays and errors in traditional communication methods. The Bell state is generated by combining the Hadamard gate and the CNOT gate, and further combined with the third quantum bit to form a multidimensional quantum superposition state, which enables the system to represent data features in a higher dimension. Compared with the traditional binary coding method, the quantum superposition state can provide a more refined and accurate data representation, which helps to improve the accuracy of subsequent analysis tasks (such as classification or prediction). The quantum gate operation is reversible, which means that any modification to the quantum state can be undone, which is very important for preventing data loss and error propagation. In addition, the quantum error correction mechanism included in the quantum teleportation process can effectively resist noise interference and ensure the quality and reliability of data transmission. In summary, by adopting quantum teleportation and quantum superposition gate operations, not only can the effective integration and accurate representation of complex data be achieved, but also the efficiency and security of information transmission can be improved, thereby optimizing the entire traffic special situation processing process and improving the ability to respond to emergencies.

[0087] Step 125: Combined with the preset rules related to the special traffic situation type, use the big model to calculate the confidence of the comprehensive feature vector to obtain the confidence of the special traffic situation type; combined with the preset rules related to the special traffic situation coverage rate, use the big model to calculate the coverage rate of the comprehensive feature vector to obtain the special traffic situation coverage rate.

[0088] As a possible implementation method, step 125, combining preset rules related to the traffic special situation type, using a large model to calculate the confidence of the comprehensive feature vector, to obtain the traffic special situation type confidence, includes:

[0089] Step c1, map the comprehensive feature vector to the special situation type semantic space through linear projection to obtain the mapped feature vector. Step c2, perform hierarchical activation on the mapped feature matrix to obtain the probability vector, introduce the temporal consistency constraint and the prior knowledge base to correct the probability vector, and obtain the corrected probability vector. Step c3, adjust the corrected probability vector according to the quantum measurement constraint to obtain the confidence of the traffic special situation type. It should be understood that with the advantage of quantum coding and transmission time of less than 50μs, combined with the lightweight inference architecture of the large model, a millisecond-level closed loop (<100ms) from data collection to policy execution is achieved, which is suitable for high-time scenarios such as highway toll stations.

[0090] By executing steps c1 to c3, the embodiment of the present application maps the comprehensive feature vector generated by complex multi-source data through quantum teleportation and other processes to a specially defined special situation type semantic space. This conversion helps to capture the features related to the specific traffic special situation type, making the subsequent probability estimation more accurate. By performing hierarchical activation on the mapped feature matrix and correcting the probability vector in combination with the temporal consistency constraint and the prior knowledge base, noise interference can be effectively reduced and the stability of the prediction model can be improved. This step ensures that the system can make a more reliable judgment even in the face of incomplete or ambiguous data input. According to the measurement principle in quantum mechanics, the corrected probability vector is adjusted. The advantage of this is that the unique perspective provided by quantum theory can be used to optimize the results of traditional machine learning methods. This method can not only provide more accurate probability estimates, but also reflect the uncertainty in the actual physical world to a certain extent. The entire process enhances the calculation ability of the confidence level of the traffic special situation type through the above-mentioned technical means, thereby providing a decision support tool for traffic management departments. These tools can help managers identify potential problem areas more quickly, deploy preventive measures in advance, reduce the incidence of traffic accidents, and ensure public safety.

[0091] In the embodiment of the present application, combined with large model analysis, quantum teleportation can help integrate data from different sources more effectively, form a more comprehensive and accurate comprehensive feature vector, and thus improve prediction accuracy and decision support capabilities. In addition, the application of quantum teleportation technology to the optimization and processing process of special traffic conditions not only breaks through the limitations of traditional communication means, but also ensures the security of information transmission through quantum encryption technology. At the same time, the unique efficiency of quantum teleportation is used to realize the instant sharing and rapid processing of special traffic information. Specifically, the quantum state contains phase and amplitude information, retains the nonlinear correlation between multimodal data (such as the quantum coherence of driver pupil contraction and brake response delay), and after the entangled state is prepared, the feature transmission does not need to rely on the bandwidth of the classical channel, and the synchronization delay of multi-node data at the toll station is reduced from 120ms to 8.7ms. In addition, the vehicle characteristics (such as speed, position), the driver's physiological signals (such as brain electroencephalogram θ waves), and the road state (such as friction coefficient) are encoded as quantum bits φ1, φ2, and φ3, respectively, and an N-dimensional superposition state is formed through quantum entanglement. For example, in the toll station scenario, a comprehensive feature vector of 24 quantum bits can represent 2^24≈16 million state combinations, far exceeding the 1000 dimensions provided by traditional methods. The CNOT gate is used to establish the entanglement relationship between the vehicle's sudden brake 101 and the driver's sudden heart rate increase 110, directly mining the implicit causal chain that is difficult to discover with traditional algorithms. Quantum teleportation bypasses the TCP / IP protocol stack and directly transmits the Bell state measurement results through quantum channels (such as polarized photons in optical fibers), reducing the end-to-end delay of toll stations from the 200ms level to the 10ms level. Majorana zero-energy mode protection uses topological quantum coding (such as weaving Majorana fermions) to make the vehicle's characteristic quantum state resist electromagnetic interference (such as the 50Hz power frequency noise of high-voltage equipment at toll stations) during transmission, and the bit error rate is reduced from 10 -3 Down to 10 -7 . Perform surface code error correction on the quantum state after transmission. Even if a single quantum bit flips (such as due to electromagnetic pulses in thunderstorms), the original feature vector can still be restored. Establish a quantum key before feature transmission to encrypt the Bell state measurement results in the classical channel to prevent attackers from forging ETC fault data or tampering with the driver's vital signs. Based on the anti-interference coding scheme of Majorana zero-energy mode, 99.3% quantum state fidelity is achieved in complex electromagnetic environments (toll booths).

[0092] It should be noted that this method has good adaptability and flexibility, and can flexibly adjust its algorithm and rule set according to different environmental parameters, traffic patterns and types of emergencies. Therefore, this method can not only be applied to specific scenarios such as toll stations, but can also be extended to other traffic management systems.

[0093] By executing steps 121 to 125, the embodiment of the present application determines the correlation between environmental parameters, which helps to reveal how different factors work together on traffic conditions, and provides a richer information basis for subsequent prediction models. The urgency prediction model based on the attention mechanism can focus on the most critical factors, thereby providing more accurate urgency prediction results. The Bayesian filtering framework and Kalman filter are used to correct the prediction results, which can effectively reduce the prediction error and obtain a more reliable environmental urgency assessment. The shared entangled state is generated by the quantum teleportation method, and multiple data sources are fused into a comprehensive feature vector. This method not only improves the data fusion efficiency, but also enhances the ability of feature representation. It can not only improve the ability to understand and react to complex traffic conditions, but also provide strong support for the formulation of effective preventive measures. This method improves traffic safety and efficiency, and has good scalability and adaptability. It can flexibly adjust its algorithms and rule sets according to different application scenarios to meet diverse needs.

[0094] In a possible embodiment, S13, generating a traffic situation dynamic priority weight coefficient based on the environmental urgency, the traffic situation type confidence and the traffic situation coverage, includes:

[0095] Step 131: Perform nonlinear normalization processing on the environmental urgency, traffic special situation type confidence, and traffic special situation coverage to generate standardized environmental urgency, traffic special situation type confidence, and traffic special situation coverage. The nonlinear normalization processing can map data of different dimensions such as environmental urgency, traffic special situation type confidence, and traffic special situation coverage to a unified interval (such as [0,1]) through a nonlinear function (such as a logarithm, power function, or sigmoid function), thereby eliminating the influence of dimension differences on weight calculation.

[0096] Step 132: Construct an attention matrix based on the standardized environmental urgency, traffic special situation type confidence, and traffic special situation coverage. The attention matrix is ​​a two-dimensional matrix used to quantify the correlation weights between the environmental urgency, special situation type confidence, and coverage. For example, environmental urgency (original value: 0.8, based on a prediction of 30 meters of fog visibility), ETC fault confidence (0.75), and coverage (0.6, affecting 3 of the 4 lanes). After standardization: urgency = 0.92, confidence = 0.87, coverage = 0.78.

[0097] Step 133, extract local correlation features from the attention matrix, obtain the initial weight coefficient according to the quantum annealing method, introduce the hyperbolic tangent attenuation factor to adjust the initial weight coefficient to compensate for the instantaneous fluctuation of the environmental urgency, and obtain the dynamic priority weight coefficient of the special traffic situation. It should be understood that the indirect impact between the environmental urgency and the coverage rate (such as the risk of lane congestion caused by heavy fog) is quantified through the attention mechanism. Quantum annealing escapes the local extreme value by simulating the quantum tunneling effect to avoid strategy misjudgment due to initial weight deviation (such as short-term fluctuations in ETC failure rate). The dynamic compensation mechanism of the hyperbolic tangent attenuation factor performs exponential attenuation compensation (such as 60% attenuation rate when τ=10 seconds) for the instantaneous fluctuation of the environmental urgency (such as short duration of heavy fog but severe impact), avoiding over-response to short-term noise. In addition, based on the quantum annealing algorithm and the hyperbolic tangent attenuation factor, the weight coefficient of the environmental urgency can be iteratively updated within 10ms in the embodiment of the present application, which is 10 times faster than the static rule response speed.

[0098] By executing steps 131 to 133, the embodiment of the present application solves the defects of the traditional dynamic weight generation method in data adaptability, global optimization capability and real-time stability through the deep integration of nonlinear normalization, attention mechanism and quantum annealing, and provides an intelligent transportation system with a special traffic situation optimization processing method with both accuracy and robustness.

[0099] Figure 2 A structural diagram of a traffic special situation optimization processing system based on a large model provided in an embodiment of the present application is shown in FIG. Figure 2 As shown, the system includes:

[0100] The acquisition module 21 is used to acquire multi-source heterogeneous data, which includes: vehicle data, road data, driver physiological data and environmental information. The driver physiological data includes driver brain wave signals and driver eye movement trajectories.

[0101] A determination generation module 22 is used to determine the degree of environmental urgency based on environmental information, introduce a quantum teleportation method, generate a comprehensive feature vector corresponding to vehicle data, road data and driver physiological data, and calculate the confidence level of the traffic special situation type and the coverage rate of the traffic special situation based on the comprehensive feature vector using a large model.

[0102] The generating module 23 is used to generate a dynamic priority weight coefficient of a traffic situation based on the degree of environmental urgency, the confidence level of the traffic situation type and the coverage rate of the traffic situation.

[0103] Generate an optimization module 24, which is used to generate a corresponding initial strategy according to the dynamic priority weight coefficient of the traffic special situation, optimize the initial strategy according to the delayed response time index and the optimization processing effect index, and obtain a target strategy. The target strategy includes a target intelligent agent and a target traffic special situation processing plan, so as to execute the target traffic special situation processing plan through the target intelligent agent.

[0104] Figure 2 The traffic special situation optimization processing system based on the large model can be executed Figure 1 The implementation principle and technical effect of the traffic special situation optimization processing method based on a large model described in the illustrated embodiment will not be repeated. The specific manner in which each module and unit performs operations in the traffic special situation optimization processing system based on a large model in the above embodiment has been described in detail in the embodiment of the method, and will not be elaborated here.

[0105] In one possible design, Figure 2 The traffic special situation optimization processing system based on the large model of the embodiment shown can be implemented as a computing device, such as Figure 3 As shown, the computing device may include a storage component 31 and a processing component 32 .

[0106] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are called and executed by the processing component 32 .

[0107] The processing component 32 is used to: obtain multi-source heterogeneous data, the multi-source heterogeneous data includes: vehicle data, road data, driver physiological data and environmental information, the driver physiological data includes driver brain wave signals and driver eye movement trajectories; determine the environmental urgency based on the environmental information, introduce the quantum teleportation method, generate a comprehensive feature vector corresponding to the vehicle data, road data and driver physiological data, and calculate the traffic special situation type confidence and traffic special situation coverage rate using a large model based on the comprehensive feature vector; generate a traffic special situation dynamic priority weight coefficient based on the environmental urgency, traffic special situation type confidence and traffic special situation coverage rate; generate a corresponding initial strategy according to the traffic special situation dynamic priority weight coefficient, optimize the initial strategy according to the delayed response time index and the optimization processing effect index, and obtain a target strategy, the target strategy includes a target intelligent agent and a target traffic special situation processing plan, so that the target traffic special situation processing plan is executed through the target intelligent agent.

[0108] The present application also provides a computer storage medium storing a computer program, wherein the computer program can achieve the above-mentioned Figure 1 The illustrated embodiment is a method for optimizing special traffic conditions based on a large model.

[0109] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0110] The system embodiment described above is merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art may understand and implement it without creative work.

[0111] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0112] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for optimizing special traffic conditions based on a large model, characterized in that: include: Acquire multi-source heterogeneous data, wherein the multi-source heterogeneous data includes: vehicle data, road data, driver physiological data and environmental information, wherein the driver physiological data includes driver brain wave signals and driver eye movement trajectories; Determine the correlation between various environmental parameters in the environmental information, and determine the environmental urgency level in combination with the predicted urgency level; Based on the convolutional neural network layer in the dual encoder, the vehicle data and the road data are respectively subjected to structured feature extraction to obtain vehicle features and road features, and based on the quantum neural network layer in the dual encoder, unstructured feature extraction is performed on the driver's physiological data to obtain the driver's physiological features; The vehicle characteristics and the road characteristics are respectively encoded by Majorana zero-energy modes to obtain a first quantum bit corresponding to the vehicle characteristics and a second quantum bit corresponding to the road characteristics, and the driver's physiological characteristics are compressed into a third quantum bit by quantum embedding; Performing quantum teleportation on the first quantum bit and the second quantum bit by a quantum teleportation method to generate a shared entangled state, and performing mixed superposition of the third quantum bit and the shared entangled state to obtain a comprehensive feature vector corresponding to the vehicle data, road data, and driver physiological data; Combined with the preset rules related to the traffic special situation type, the big model is used to calculate the confidence of the comprehensive feature vector to obtain the traffic special situation type confidence, and combined with the preset rules related to the traffic special situation coverage rate, the big model is used to calculate the coverage rate of the comprehensive feature vector to obtain the traffic special situation coverage rate; Generate a dynamic priority weight coefficient of the traffic situation based on the environmental urgency, the confidence level of the traffic situation type and the traffic situation coverage rate; Generate a corresponding initial strategy according to the dynamic priority weight coefficient of the special traffic situation, optimize the initial strategy according to the delayed response time index and the optimization processing effect index, and obtain a target strategy, which includes a target intelligent agent and a target special traffic situation processing plan, so that the target special traffic situation processing plan is executed through the target intelligent agent.

2. The method according to claim 1, characterized in that The determining of the correlation between the environmental parameters in the environmental information and determining the environmental urgency in combination with the predicted urgency includes: Performing time series smoothing, spatial interpolation and anomaly detection processing on the data of each environmental parameter in the environmental information in sequence to obtain processed data, the environmental parameters including lane occupancy rate, visibility index, meteorological warning level and road friction coefficient; Using an attention-based urgency prediction model, the urgency is predicted based on the processed data to obtain a predicted urgency; The correlation between various environmental parameters in the environmental information is determined, a Bayesian filtering framework is introduced, a Kalman filter and the correlation are combined, the deviation of the predicted urgency is corrected, and the environmental urgency is obtained.

3. The method according to claim 1, characterized in that The method of performing quantum teleportation on the first quantum bit and the second quantum bit to generate a shared entangled state, and performing mixed superposition of the third quantum bit and the shared entangled state to obtain a comprehensive feature vector corresponding to the vehicle data, the road data, and the driver's physiological data includes: Sending a first instruction to a transmitting end, so that the transmitting end applies a combination of a Hadamard gate and a CNOT gate to the first quantum bit and the second quantum bit based on the first instruction to generate four candidate Bell states, and sending the measurement results of the four candidate Bell states to a preset receiving end through a classical channel; Sending a second instruction to a receiving end, so that the receiving end performs a quantum gate operation on the measurement results of the four candidate Bell states based on the second instruction to reconstruct a shared entangled state; The third quantum bit is mixed and superimposed with the shared entangled state through a quantum superposition gate operation to form a multi-dimensional quantum superposition state; The multi-dimensional quantum superposition state is adjusted through normalization constraints to eliminate the quantum state phase deviation and generate a comprehensive feature vector corresponding to the vehicle data, road data and driver physiological data.

4. The method according to claim 1, characterized in that: The confidence level of the traffic special situation type is calculated by combining the preset rules related to the traffic special situation type with the large model to obtain the traffic special situation type confidence level, including: Map the comprehensive feature vector to the special situation type semantic space through linear projection to obtain the mapped feature vector; Performing hierarchical activation on the mapped feature matrix to obtain a probability vector, introducing a temporal consistency constraint and a priori knowledge base to correct the probability vector to obtain a corrected probability vector; According to the quantum measurement constraints, the revised probability vector is adjusted to obtain the confidence level of the traffic special situation type.

5. The method according to claim 1, characterized in that The vehicle characteristics include license plate detection results, vehicle model detection results and operating status detection results, and the road characteristics include traffic flow, lane status and road condition, and each vehicle characteristic is mapped to a different first quantum bit in the same group, and each road characteristic is mapped to a different second quantum bit in another group.

6. The method according to claim 1, characterized in that The generating of the traffic special situation dynamic priority weight coefficient based on the environmental urgency, the traffic special situation type confidence and the traffic special situation coverage rate comprises: Performing nonlinear normalization processing on the environmental urgency, traffic special situation type confidence and traffic special situation coverage to generate standardized environmental urgency, traffic special situation type confidence and traffic special situation coverage; Construct an attention matrix based on the standardized environmental urgency, traffic condition type confidence, and traffic condition coverage; Local correlation features are extracted from the attention matrix, and the initial weight coefficient is obtained according to the quantum annealing method. The hyperbolic tangent attenuation factor is introduced to adjust the initial weight coefficient to compensate for the instantaneous fluctuation of the environmental urgency, and the dynamic priority weight coefficient of the traffic special situation is obtained.

7. A traffic special situation optimization processing system based on a large model, characterized in that: include: An acquisition module, used for acquiring multi-source heterogeneous data, wherein the multi-source heterogeneous data includes: vehicle data, road data, driver physiological data and environmental information, wherein the driver physiological data includes driver brain wave signals and driver eye movement trajectories; A determination generation module is used to execute the following process: determining the correlation between various environmental parameters in the environmental information, and determining the environmental urgency degree in combination with the predicted urgency degree; Based on the convolutional neural network layer in the dual encoder, the vehicle data and the road data are respectively subjected to structured feature extraction to obtain vehicle features and road features, and based on the quantum neural network layer in the dual encoder, unstructured feature extraction is performed on the driver's physiological data to obtain the driver's physiological features; The vehicle characteristics and the road characteristics are respectively encoded by Majorana zero-energy modes to obtain a first quantum bit corresponding to the vehicle characteristics and a second quantum bit corresponding to the road characteristics, and the driver's physiological characteristics are compressed into a third quantum bit by quantum embedding; Performing quantum teleportation on the first quantum bit and the second quantum bit by a quantum teleportation method to generate a shared entangled state, and performing mixed superposition of the third quantum bit and the shared entangled state to obtain a comprehensive feature vector corresponding to the vehicle data, road data, and driver physiological data; Combined with the preset rules related to the traffic special situation type, the big model is used to calculate the confidence of the comprehensive feature vector to obtain the traffic special situation type confidence, and combined with the preset rules related to the traffic special situation coverage rate, the big model is used to calculate the coverage rate of the comprehensive feature vector to obtain the traffic special situation coverage rate; A generation module, used to generate a dynamic priority weight coefficient of a traffic situation based on the environmental urgency, the confidence level of the traffic situation type and the coverage rate of the traffic situation; Generate an optimization module, which is used to generate a corresponding initial strategy according to the dynamic priority weight coefficient of the special traffic situation, optimize the initial strategy according to the delayed response time index and the optimization processing effect index, and obtain a target strategy. The target strategy includes a target intelligent agent and a target special traffic situation processing plan, so that the target special traffic situation processing plan is executed through the target intelligent agent.

8. A computing device, characterized in that It comprises a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a large model-based traffic special situation optimization processing method as described in any one of claims 1 to 6.

9. A computer storage medium, characterized in that A computer program is stored, and when the computer program is executed by a computer, a traffic special situation optimization processing method based on a large model as described in any one of claims 1 to 6 is implemented.

Images