Traffic control method, system and device based on large model and storage medium
By acquiring traffic monitoring data and road network information, and using large-scale model analysis and reinforcement learning training, traffic control data is generated, solving the problem that traditional methods are unable to cope with complex traffic flow and improving vehicle driving efficiency.
Patent Information
- Application Number
- CN202511420696.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-09-30
- Publication Date
- 2026-02-10
AI Technical Summary
Traditional traffic signal control methods are ineffective in dealing with complex and ever-changing traffic flows, especially on roads without video surveillance devices, leading to reduced vehicle driving efficiency.
By acquiring traffic monitoring data and road network information, traffic characteristic data is determined, target prompt information is generated and input into a large traffic analysis model, traffic analysis results are output, and finally traffic control data is generated. The large model is trained using reinforcement learning algorithms to improve the accuracy of the analysis.
It improves the accuracy and efficiency of traffic control, especially in enabling intelligent traffic signal adjustment on roads lacking video surveillance.
Smart Images
Figure CN121505844A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of intelligent transportation technology, and in particular to a traffic control method, system, device, and storage medium based on a large model. Background Technology
[0002] With the development of urbanization, traffic congestion has become increasingly prominent. Traditional traffic signal control methods, such as fixed-duration control and rule-based optimization strategies, are no longer effective in dealing with complex and ever-changing traffic flow. Currently, intelligent adjustment methods are generally used to adjust traffic signals at various intersections to improve the efficiency of vehicles driving on urban roads. However, not all roads are equipped with video surveillance devices. Intelligent adjustments on roads that cannot be monitored often reduce the efficiency of vehicles driving on the road.
[0003] The above content is only used to help understand the technical solution of the present invention and does not represent an admission that the above content is prior art. Summary of the Invention
[0004] The main objective of this invention is to provide a traffic control method, system, device, and storage medium based on a large-scale model, aiming to improve vehicle driving efficiency on roads. To achieve the above objective, this invention provides a traffic control method based on a large-scale model, which includes the following steps:
[0005] Acquire traffic monitoring data and road network information, and determine traffic characteristic data based on the traffic monitoring data. The traffic characteristic data includes: U-turn and return data and first environmental data of the monitoring area.
[0006] Generate corresponding target prompt information based on the traffic feature data, road network information, and prompt template;
[0007] Input the target prompt information into the traffic analysis model and output the traffic analysis results;
[0008] The traffic control data is generated based on the traffic analysis results.
[0009] Optionally, before the step of inputting the target prompt information into the traffic analysis model and outputting the traffic analysis results, the method further includes:
[0010] Acquire historical traffic monitoring data and road network information, and generate query data based on the historical traffic monitoring data and road network information;
[0011] Based on the question data, historical road network water accumulation data, and the reward model, a large model to be trained is obtained, thus obtaining the large traffic analysis model.
[0012] Optionally, the step of obtaining the traffic analysis model by training a large model to be trained based on the question data, historical road network water accumulation data, and the reward model includes:
[0013] The question data is input into the large model to be trained, and the large model to be trained is controlled to output the answer data.
[0014] The answer data is scored based on the historical road network water accumulation data and the reward model to obtain an answer score;
[0015] The parameters of the large model to be trained are adjusted according to the reinforcement algorithm and the answer score to obtain the large traffic analysis model.
[0016] Optionally, the step of scoring the response data based on the historical road network waterlogging data and the reward model to obtain the response score includes:
[0017] The scoring input data is generated based on the answer data, the historical road network water accumulation data, and the scoring template.
[0018] The scoring input data is input into the reward model to obtain the output result of the reward model;
[0019] The answer score is determined based on the output.
[0020] Optionally, the step of adjusting the parameters of the large model to be trained based on the reinforcement algorithm and the answer score to obtain the traffic analysis large model includes:
[0021] The word contribution of each word in the answer data is calculated based on the reinforcement learning algorithm and the answer score.
[0022] Determine the corresponding target loss function based on the contribution of multiple words;
[0023] The parameters of the large model to be trained are adjusted by gradient descent and directional propagation to obtain the large traffic analysis model.
[0024] Optionally, the step of determining traffic characteristic data based on the traffic monitoring data includes:
[0025] Based on the traffic monitoring data, identify the first vehicle data passing through the monitored area, the vehicle data including the vehicle's direction of travel;
[0026] Collect second vehicle data of vehicles that pass through the monitored area twice within the target time period, and determine the U-turn return data based on the second vehicle data;
[0027] The first environmental data corresponding to the monitoring area is determined based on the traffic monitoring data.
[0028] Optionally, the step of generating the traffic control data based on the traffic analysis results includes:
[0029] When the traffic analysis results indicate that there is a waterlogged area in an adjacent area of the monitored area, the traffic signals leading to the adjacent area are adjusted.
[0030] When the traffic analysis result indicates that there are no waterlogged areas in the adjacent areas of the monitored area, the traffic signals in the monitored area are adjusted based on the traffic flow data.
[0031] Furthermore, to achieve the above objectives, the present invention also provides a traffic control system based on a large model. The traffic control system based on a large model includes:
[0032] The acquisition module is used to acquire traffic monitoring data and road network information, and determine traffic characteristic data based on the traffic monitoring data. The traffic characteristic data includes: U-turn and return data and first environmental data of the monitoring area.
[0033] The monitoring module is used to generate corresponding target prompt information based on the traffic feature data, road network information, and prompt template;
[0034] The analysis module is used to input the target prompt information into the traffic analysis model and output the traffic analysis results;
[0035] The control module is used to generate the traffic control data based on the traffic analysis results.
[0036] Furthermore, to achieve the above objectives, the present invention also provides a traffic control device based on a large model, the traffic control device based on a large model comprising: a memory, a processor, and a traffic control program based on a large model stored in the memory and executable on the processor, the traffic control program based on a large model being configured to implement the steps of the traffic control method based on a large model as described above.
[0037] In addition, to achieve the above objectives, the present invention also provides a storage medium storing a traffic control program based on a large model, wherein when the traffic control program based on the large model is executed by a processor, it implements the steps of the traffic control method based on the large model described above.
[0038] This invention proposes a traffic control method based on a large model. This method acquires traffic monitoring data and road network information, determines traffic feature data based on the traffic monitoring data, generates corresponding target prompt information based on the traffic feature data, road network information, and prompt templates, inputs the target prompt information into a large traffic analysis model, outputs traffic analysis results, and generates traffic control data based on the traffic analysis results, thereby improving the accuracy of traffic control and increasing traffic efficiency. Attached Figure Description
[0039] Figure 1 This is a schematic diagram of the structure of a traffic control device based on a large model, which is part of the hardware operating environment involved in the embodiments of the present invention.
[0040] Figure 2 This is a flowchart illustrating the first embodiment of the traffic control method based on a large model according to the present invention.
[0041] Figure 3 This is a flowchart illustrating the second embodiment of the traffic control method based on a large model according to the present invention.
[0042] Figure 4 This is a flowchart illustrating the fifth embodiment of the traffic control method based on a large model according to the present invention.
[0043] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation
[0044] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
[0045] Reference Figure 1 , Figure 1 This is a schematic diagram of the traffic control equipment structure based on a large model, which is part of the hardware operating environment involved in the embodiments of the present invention.
[0046] like Figure 1As shown, the traffic control device based on a large model may include: a processor 1001, such as a central processing unit (CPU), a communication bus 1002, an interaction device 1003, a network interface 1004, and a memory 1005. The communication bus 1002 is used to enable communication between these components. The interaction device 1003 may include a display screen or an input unit such as a keyboard. Optionally, the interaction device 1003 may also connect to the communication bus via standard wired or wireless interfaces. The network interface 1004 may optionally include standard wired or wireless interfaces (such as a Wi-Fi interface). The memory 1005 may be a high-speed random access memory (RAM) or a stable non-volatile memory (NVM), such as a disk drive. Optionally, the memory 1005 may also be a storage device independent of the aforementioned processor 1001.
[0047] Those skilled in the art will understand that Figure 1 The structure shown does not constitute a limitation on traffic control devices based on large models, and may include more or fewer components than shown, or combine certain components, or have different component arrangements.
[0048] like Figure 1 As shown, the memory 1005, which serves as a storage medium, may include an operating system, a data storage module, a network communication module, a user interface module, and a traffic control program based on a large model.
[0049] exist Figure 1 In the large-model-based traffic control device shown, the network interface 1004 is mainly used for data communication with other devices; the interactive device 1003 is mainly used for data interaction with users; the processor 1001 and memory 1005 in the large-model-based traffic control device of the present invention can be set in the large-model-based traffic control device, and the large-model-based traffic control device calls the large-model-based traffic control program stored in the memory 1005 through the processor 1001 and executes the large-model-based traffic control method provided in the embodiment of the present invention.
[0050] This invention provides a traffic control method based on a large model, referring to... Figure 2 , Figure 2 This is a flowchart illustrating the first embodiment of a traffic control method based on a large model according to the present invention.
[0051] In this embodiment, the traffic control method based on a large model includes:
[0052] Step S1: Obtain traffic monitoring data and road network information, and determine traffic characteristic data based on the traffic monitoring data. The traffic characteristic data includes: U-turn and return data and first environmental data of the monitoring area.
[0053] The traffic monitoring data here generally refers to the monitoring data obtained from cameras installed at traffic intersections. The road network information can be a map, including traffic information of the road network. Generally, traffic monitoring data is video data. By extracting the image information from the traffic monitoring data, traffic flow information and the first environmental data of the monitored area can be determined. It should be noted that, compared with general traffic flow information, this application focuses on acquiring traffic feature data, including head-turn and U-turn data. In this embodiment, vehicle feature information, such as license plate, color, direction of travel, and speed, is recorded. By detecting vehicles that pass through the intersection at least twice within a certain period of time, head-turn and U-turn data is determined.
[0054] Step S2: Generate corresponding target prompt information based on the traffic feature data, road network information, and prompt template;
[0055] In this embodiment, the raw traffic monitoring data is not directly input into the traffic analysis model, thus effectively saving the context content of the traffic analysis model. This allows for the use of a more lightweight model. Preferably, the target road network information to be input is divided from the road network information based on the traffic monitoring data. Optionally, the target road network information is: road network information within a preset range from the monitoring area of the monitoring data. For example: road network information within a 1-kilometer range from the monitoring area of the monitoring data. The prompt template here includes: preset prompt words and preset prompt phrases. Optionally, the preset prompt phrases may include: based on the above data analysis of the flooding situation in the monitoring area and adjacent areas, output the analysis results.
[0056] Step S3: Input the target prompt information into the traffic analysis model and output the traffic analysis results;
[0057] The aforementioned target prompt information is input into a large-scale traffic analysis model, which then outputs traffic analysis results. These results may include: the location of flooded roads, water depth, and predicted probability. In some embodiments, results related to road congestion may also be output. It is important to note that even if the traffic feature data and road network information are the same, different prompt templates may result in different output traffic analysis results. Different templates can be used to output the traffic analysis results multiple times; the accuracy of the analysis can be determined by whether the results contain the same information.
[0058] Step S4: Generate the traffic control data based on the traffic analysis results.
[0059] In this embodiment, specifically, when an area adjacent to the monitored area is flooded, the traffic lights indicating traffic towards the flooded area are adjusted. For example, the green light duration is reduced, or the green light is not displayed. Optionally, based on the traffic analysis results, corresponding prompts are generated and can be broadcast to the corresponding traffic intersection via a network. Alternatively, the prompts can be sent to vehicles at the intersection via a vehicle-to-everything (V2X) network, so that vehicles do not pass through flooded areas or congested sections of road.
[0060] In this embodiment, by acquiring traffic monitoring data and road network information, determining traffic characteristic data based on the traffic monitoring data, generating corresponding target prompt information based on the traffic characteristic data, road network information, and prompt template, inputting the target prompt information into a large traffic analysis model, outputting traffic analysis results, and generating traffic control data based on the traffic analysis results, the accuracy of traffic control and traffic efficiency can be improved.
[0061] Furthermore, based on the first embodiment, a second embodiment of the traffic control method based on a large model of the present invention is proposed. In this embodiment, reference is made to... Figure 3 Before the step of inputting the target prompt information into the traffic analysis model and outputting the traffic analysis results, the method further includes:
[0062] Step S301: Obtain historical traffic monitoring data and road network information, and generate query data based on the historical traffic monitoring data and road network information;
[0063] In this embodiment, a large-scale traffic analysis model is obtained by training a large model using historical data. Preferably, the historical traffic monitoring data here refers to monitoring data recorded before the current time, and the road network information can be obtained from a database. The question data is generated using a prompt template, the historical traffic monitoring data, and the road network information. Historical traffic feature data corresponding to the historical traffic monitoring data is extracted, and the historical traffic feature data here is of the same type as the traffic feature data.
[0064] Step S302: Based on the question data, historical road network water accumulation data, and the reward model, train the large model to be trained to obtain the traffic analysis large model.
[0065] Specifically, the traffic analysis model can be obtained by training the large training model using reinforcement learning. During reinforcement learning, a reward model scores the output of the large training model, allowing for parameter adjustments to improve the accuracy of the traffic analysis model.
[0066] In this embodiment, by acquiring historical traffic monitoring data and road network information, generating query data based on the historical traffic monitoring data and road network information, and training a large model to be trained based on the query data, historical road network water accumulation data, and a reward model, the traffic analysis large model is obtained, thereby improving the accuracy of the traffic analysis results obtained from the traffic analysis large model.
[0067] Furthermore, based on the first or second embodiment, a third embodiment of the traffic control method based on a large model of the present invention is proposed. In this embodiment, the step of training the large model to be trained based on the query data, historical road network water accumulation data, and reward model to obtain the traffic analysis large model includes:
[0068] The question data is input into the large model to be trained, and the large model to be trained is controlled to output the answer data.
[0069] The answer data is scored based on the historical road network water accumulation data and the reward model to obtain an answer score;
[0070] The parameters of the large model to be trained are adjusted according to the reinforcement algorithm and the answer score to obtain the large traffic analysis model.
[0071] In this embodiment, a general-purpose large-scale reward model is used, such as GPT or Deepseek. A model with a large dataset is also used. Historical road network flooding data corresponding to historical traffic monitoring data is used as the scoring metric. In this embodiment, it is not necessary to train a dedicated large-scale model for scoring; instead, the historical road network flooding data is used as an external knowledge base, employing a general-purpose large-scale model for scoring. This reduces the resources required to train a standard reward model. Specifically, the response data and the historical road network flooding data are input into the reward model, and it is prompted to determine the accuracy of the response data based on the historical road network flooding data. A score is then calculated. For example, when the flooded area determined by the response data completely overlaps with the historical road network flooding data, the response score is determined as the first score; when the flooded area determined by the response data does not overlap with the historical road network flooding data, the response score is determined as the second score, where the first score is higher than the second score. When the waterlogged area determined by the answer data does not overlap with the historical road network waterlogging data, the numerical value of the second score is determined based on the area of overlap between the waterlogged area determined by the answer data and the historical road network waterlogging data, and / or the distance.
[0072] In this embodiment, by inputting the question data into the large model to be trained and controlling the large model to output the answer data, the answer data is scored according to the historical road network water accumulation data and the reward model to obtain the answer score. The parameters of the large model to be trained are adjusted according to the reinforcement algorithm and the answer score to obtain the traffic analysis large model. This can reduce the resources required for training and improve the accuracy of the model based on real data.
[0073] Furthermore, the step of scoring the response data based on the historical road network waterlogging data and the reward model to obtain the response score includes:
[0074] The scoring input data is generated based on the answer data, the historical road network water accumulation data, and the scoring template.
[0075] The scoring input data is input into the reward model to obtain the output result of the reward model;
[0076] The answer score is determined based on the output.
[0077] It should be noted that, for the sake of scoring consistency, multiple scoring prompt templates can be used to generate multiple scoring input data for a single answer, thereby obtaining multiple output results. The average of the multiple output results is calculated, and the average value is used as the answer score.
[0078] In this embodiment, multiple scoring input data are generated by using multiple scoring prompt templates, thereby obtaining multiple output results. The average value of the multiple output results is calculated, which can avoid the instability of data caused by the biased scoring of the templates and further improve the accuracy of the answer scoring.
[0079] Furthermore, based on any of the above embodiments, a fourth embodiment of the traffic control method based on a large model of the present invention is proposed. In this embodiment, the step of adjusting the parameters of the large model to be trained according to the reinforcement algorithm and the response score to obtain the traffic analysis large model includes:
[0080] The word contribution of each word in the answer data is calculated based on the reinforcement learning algorithm and the answer score.
[0081] Determine the corresponding target loss function based on the contribution of multiple words;
[0082] The parameters of the large model to be trained are adjusted by gradient descent and directional propagation to obtain the large traffic analysis model.
[0083] In this embodiment, the reinforcement learning algorithm can be Proximal Policy Optimization (PPO), a policy gradient method and currently the most mainstream algorithm for fine-tuning large-scale reinforcement learning models. This training method has strong generality. Specifically,
[0084] The objective loss function is transformed into an optimizable mathematical objective based on the aforementioned word contribution. In an embodiment using the PPO algorithm, specifically, the objective loss function comprises a composite objective function of policy loss, value function loss, and entropy reward. Furthermore, the PPO algorithm can also use a pruning mechanism to limit the probability ratio within a reasonable range, thereby ensuring smooth and robust updates. The parameters of the large model to be trained are adjusted using gradient descent and direction propagation, where a lower learning rate can be set to stabilize the learning process. Through the above methods, the large traffic analysis model can be obtained.
[0085] In this embodiment, the word contribution of each word in the answer data is calculated according to the reinforcement learning algorithm and the answer score. The corresponding target loss function is determined based on the multiple word contributions. The parameters of the large model to be trained are adjusted according to gradient descent and directional propagation to obtain the traffic analysis large model. This reduces the computational complexity. Furthermore, as the capabilities of the new general model continue to improve, training can be completed quickly by using the new general model. Therefore, the stability of the traffic analysis large model during the iteration process is improved based on the method of this embodiment.
[0086] Furthermore, based on any of the above embodiments, a fifth embodiment of the traffic control method based on a large model of the present invention is proposed. In this embodiment, reference is made to... Figure 4 The step of determining traffic characteristic data based on the traffic monitoring data includes:
[0087] Step S11: Identify the first vehicle data passing through the monitoring area based on the traffic monitoring data, wherein the vehicle data includes the vehicle's direction of travel.
[0088] Specifically, the vehicle's direction of travel is determined through image recognition. In this embodiment, the image recognition algorithm is not limited; it can be any mainstream vehicle recognition algorithm.
[0089] Step S12: Collect second vehicle data of vehicles that pass through the monitoring area twice during the target time period, and determine the U-turn return data based on the second vehicle data;
[0090] Step S13: Determine the first environmental data corresponding to the monitoring area based on the traffic monitoring data.
[0091] In this embodiment, the order of steps S11 and S13 is not limited. That is, the turnaround data can be determined first, and then the first environmental data can be determined. When the first environmental data is determined first, and then the turnaround data is determined, the target time period can be preset, or a different target time period can be set according to the first environmental data.
[0092] In this embodiment, the traffic monitoring data identifies first vehicle data passing through the monitoring area, the vehicle data including the vehicle's direction of travel; second vehicle data of vehicles that pass through the monitoring area twice within a target time period is statistically analyzed, and the U-turn data is determined based on the second vehicle data; and first environmental data corresponding to the monitoring area is determined based on the traffic monitoring data, thereby improving the accuracy of obtaining traffic feature data.
[0093] Furthermore, based on any of the above embodiments, a sixth embodiment of the traffic control method based on a large model of the present invention is proposed. In this embodiment, the step of generating the traffic control data based on the traffic analysis results includes:
[0094] When the traffic analysis results indicate that there is a waterlogged area in an adjacent area of the monitored area, the traffic signals leading to the adjacent area are adjusted.
[0095] When the traffic analysis result indicates that there are no waterlogged areas in the adjacent areas of the monitored area, the traffic signals in the monitored area are adjusted based on the traffic flow data.
[0096] Specifically, the traffic analysis results can be second environmental data, and traffic control data can be determined based on the first and second environmental data. For example, if the first environmental data in the monitored area shows water accumulation in some lanes, the timing of the traffic lights for the corresponding lanes can be adjusted. It should be noted that because historical road network water accumulation data was used during the training of the large traffic analysis model, the corresponding traffic analysis results are related to the water accumulation. In other embodiments, when historical concert congestion hotspot data is used instead of historical road network water accumulation data, and the prompt questions in the prompt template are adjusted accordingly, the obtained traffic analysis results are related to the traffic congestion hotspots.
[0097] In this embodiment, when the traffic analysis result indicates that there is a waterlogged area in the adjacent area of the monitored area, the traffic signal leading to the adjacent area is adjusted. When the traffic analysis result indicates that there is no waterlogged area in the adjacent area of the monitored area, the traffic signal in the monitored area is adjusted according to the traffic flow data. In fact, it can identify the waterlogging situation in areas outside the monitored area and adjust the traffic signal accordingly, thereby improving the efficiency of vehicle driving on the road by intelligently adjusting traffic lights when road network information is insufficient.
[0098] Furthermore, this invention also proposes a traffic control system based on a large model, which includes:
[0099] The acquisition module is used to acquire traffic monitoring data and road network information, and determine traffic characteristic data based on the traffic monitoring data. The traffic characteristic data includes: U-turn and return data and first environmental data of the monitoring area.
[0100] The monitoring module is used to generate corresponding target prompt information based on the traffic feature data, road network information, and prompt template;
[0101] The analysis module is used to input the target prompt information into the traffic analysis model and output the traffic analysis results;
[0102] The control module is used to generate the traffic control data based on the traffic analysis results.
[0103] Furthermore, embodiments of the present invention also propose a traffic control device based on a large model, the traffic control device based on a large model comprising: a memory, a processor, and a traffic control program based on a large model stored in the memory and executable on the processor, the traffic control program based on a large model being configured to implement the steps of an embodiment of the traffic control method based on a large model as described above.
[0104] Furthermore, embodiments of the present invention also propose a storage medium storing a traffic control program based on a large model, wherein when the traffic control program based on the large model is executed by a processor, it implements the steps of any of the embodiments of the traffic control method based on the large model described above.
[0105] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.
[0106] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.
[0107] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.
[0108] The above are merely preferred embodiments of the present invention and do not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.
Claims
1. A traffic control method based on a large model, characterized in that, The traffic control method based on a large model includes the following steps: Acquire traffic monitoring data and road network information, and determine traffic characteristic data based on the traffic monitoring data. The traffic characteristic data includes: U-turn and return data and first environmental data of the monitoring area. Generate corresponding target prompt information based on the traffic feature data, road network information, and prompt template; Input the target prompt information into the traffic analysis model and output the traffic analysis results; The traffic control data is generated based on the traffic analysis results.
2. The traffic control method based on a large model as described in claim 1, characterized in that, Before the step of inputting the target prompt information into the traffic analysis model and outputting the traffic analysis results, the method further includes: Acquire historical traffic monitoring data and road network information, and generate query data based on the historical traffic monitoring data and road network information; Based on the question data, historical road network water accumulation data, and the reward model, a large model to be trained is obtained, thus obtaining the large traffic analysis model.
3. The traffic control method based on a large model as described in claim 2, characterized in that, The steps of obtaining the traffic analysis model by training the large model to be trained based on the question data, historical road network water accumulation data, and the reward model include: The question data is input into the large model to be trained, and the large model to be trained is controlled to output the answer data. The answer data is scored based on the historical road network water accumulation data and the reward model to obtain an answer score; The parameters of the large model to be trained are adjusted according to the reinforcement algorithm and the answer score to obtain the large traffic analysis model.
4. The traffic control method based on a large model as described in claim 3, characterized in that, The step of scoring the response data based on the historical road network water accumulation data and the reward model to obtain the response score includes: The scoring input data is generated based on the answer data, the historical road network water accumulation data, and the scoring template. The scoring input data is input into the reward model to obtain the output result of the reward model; The answer score is determined based on the output.
5. The traffic control method based on a large model as described in claim 3, characterized in that, The step of adjusting the parameters of the large model to be trained based on the reinforcement algorithm and the answer score to obtain the traffic analysis large model includes: The word contribution of each word in the answer data is calculated based on the reinforcement learning algorithm and the answer score. Determine the corresponding target loss function based on the contribution of multiple words; The parameters of the large model to be trained are adjusted by gradient descent and directional propagation to obtain the large traffic analysis model.
6. The traffic control method based on a large model as described in claim 1, characterized in that, The step of determining traffic characteristic data based on the traffic monitoring data includes: Based on the traffic monitoring data, identify the first vehicle data passing through the monitored area, the vehicle data including the vehicle's direction of travel; Collect second vehicle data of vehicles that pass through the monitored area twice within the target time period, and determine the U-turn return data based on the second vehicle data; The first environmental data corresponding to the monitoring area is determined based on the traffic monitoring data.
7. The traffic control method based on a large model as described in any one of claims 1 to 6, characterized in that, The step of generating the traffic control data based on the traffic analysis results includes: When the traffic analysis results indicate that there is a waterlogged area in an adjacent area of the monitored area, the traffic signals leading to the adjacent area are adjusted. When the traffic analysis result indicates that there are no waterlogged areas in the adjacent areas of the monitored area, the traffic signals in the monitored area are adjusted based on the traffic flow data.
8. A traffic control system based on a large model, characterized in that, The large-model-based traffic control system includes: The acquisition module is used to acquire traffic monitoring data and road network information, and determine traffic characteristic data based on the traffic monitoring data. The traffic characteristic data includes: U-turn and return data and first environmental data of the monitoring area. The monitoring module is used to generate corresponding target prompt information based on the traffic feature data, road network information, and prompt template; The analysis module is used to input the target prompt information into the traffic analysis model and output the traffic analysis results; The control module is used to generate the traffic control data based on the traffic analysis results.
9. A traffic control device based on a large model, characterized in that, The large-model-based traffic control device includes: a memory, a processor, and a large-model-based traffic control program stored in the memory and executable on the processor, the large-model-based traffic control program being configured to implement the steps of the large-model-based traffic control method as described in any one of claims 1 to 7.
10. A storage medium, characterized in that, The storage medium stores a traffic control program based on a large model, which, when executed by a processor, implements the steps of the traffic control method based on a large model as described in any one of claims 1 to 7.