Expert model-based expressway data processing method and device

Abstract

The embodiment of the invention provides an expressway data processing method based on an expert model. The method comprises the following steps: receiving an initial increment causal rule periodically issued by an expressway cloud node; inputting the initial increment causal rule into a first expressway expert model to obtain a latest expressway decision rule; and issuing the latest expressway decision rule to the expressway terminal node, enabling the expressway terminal node to update the corresponding second expressway expert model according to the latest expressway decision rule, and processing the latest expressway data collected in real time through the updated second expressway expert model. According to the invention, expressway data can be processed, and the processing efficiency and accuracy are improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation technology, and more specifically, to a method and apparatus for highway data processing based on expert models. Background Technology

[0002] With the rapid development of intelligent transportation systems, highway management platforms, as an important component of the intelligent transportation field, shoulder the important responsibility of improving highway operating efficiency, ensuring traffic safety, and optimizing traffic management. Currently, highway management platforms rely on manual labor to monitor, analyze, and support decision-making regarding traffic conditions.

[0003] However, manual monitoring suffers from problems such as delayed response and susceptibility to subjective judgment, making it difficult to cope with increasing traffic flow and complex and ever-changing road conditions. Summary of the Invention

[0004] In view of this, the purpose of this application is to provide a highway data processing method and apparatus based on expert models, which can process highway data and improve processing efficiency and accuracy.

[0005] In a first aspect, embodiments of this application provide a highway data processing method based on an expert model. This method is applied to any highway edge node in a highway data processing system based on an expert model. The system further includes a highway cloud node and at least one highway terminal node. The method includes: Receive initial incremental causal rules periodically sent by the highway cloud node; each initial incremental causal rule contains a highway event, the cause of the highway event, the adaptation area attribute value of the causal rule, and the causal correlation strength between the highway event and its corresponding cause. The initial incremental causal rules are input into the first highway expert model to obtain the latest highway decision rules. The latest highway decision rules are sent to the highway terminal nodes, enabling the highway terminal nodes to update their corresponding second highway expert models based on the latest highway decision rules, and to process the latest real-time collected highway data through the updated second highway expert models.

[0006] In one possible implementation, the step of inputting the initial incremental causal rule into a first highway expert model to obtain the latest highway decision rule includes: The initial incremental causal rules are input into the spatiotemporal gating mechanism built into the first highway expert model to filter out the initial incremental causal rules whose corresponding adaptation region attribute values ​​satisfy the preset region attribute threshold conditions, and thus obtain the intermediate incremental causal rules. The intermediate incremental causal rules are input into the dynamic temperature coefficient scheduler built into the first highway expert model, so as to allocate the corresponding distillation temperature to the intermediate incremental causal rules according to the corresponding highway event risk level. The intermediate incremental causal rules and their corresponding distillation temperatures are input into the distillation module built into the first highway expert model, so as to perform graded distillation on the intermediate incremental causal rules according to the corresponding distillation temperatures to obtain the target incremental causal rules. The target incremental causal rule is input into the decision rule generation module built into the first highway expert model to generate the latest highway decision rule in combination with the preset causal decision mapping knowledge base.

[0007] In one possible implementation, the step of inputting the initial incremental causal rules into the spatiotemporal gating mechanism built into the first highway expert model to filter out the initial incremental causal rules whose corresponding adaptation region attribute values ​​satisfy the preset region attribute threshold conditions, and obtaining intermediate incremental causal rules, includes: If the correlation between the attribute value of the region to which the highway edge node belongs and the attribute value of the adaptation region is greater than a preset correlation threshold, then the initial incremental causal rule is determined as an intermediate incremental causal rule.

[0008] In one possible implementation, the step of inputting the intermediate incremental causal rule into the dynamic temperature coefficient scheduler built into the first highway expert model, so as to allocate a corresponding distillation temperature to the intermediate incremental causal rule according to the corresponding highway event risk level, includes: The preset distillation temperature corresponding to the risk level of the highway event is determined as the distillation temperature corresponding to the intermediate incremental causal rule assignment.

[0009] In one possible implementation, the method further includes: Receive candidate action sets generated from the latest highway data uploaded by multiple highway terminal nodes; For each candidate action set, a decision score is calculated based on the resource coverage benefit, scheduling cost, and synergistic gain with at least one other candidate action set. The set of candidate actions with the highest decision score is determined as the first candidate action set; The second candidate action set whose cooperative gain with the first candidate action set is greater than or equal to a preset cooperative gain value is determined as the final decision action set; The latest highway data is processed based on the set of decision actions.

[0010] Secondly, embodiments of this application provide an expert model-based highway data processing method. This method is applied to a highway cloud node in the expert model-based highway data processing system described in the first aspect. The system further includes at least one highway edge node and at least one highway terminal node. The method includes: Receive incremental feature data of the highway periodically reported by at least one highway edge node; All incremental characteristic data of highways are input into the causal analysis model to obtain the initial incremental causal rules; The initial incremental causal rules are then distributed to each highway edge node.

[0011] Thirdly, embodiments of this application provide a highway data processing method based on an expert model. This method is applied to a highway terminal node in the highway data processing system based on an expert model as described in the second aspect. The system further includes a highway cloud node and a highway edge node. The method also includes: Get the latest highway data; The latest highway data is input into the second highway expert model within the highway terminal node to obtain a set of candidate actions; The candidate action set is sent to the corresponding highway edge node so that the highway edge node can select the final decision action set from the candidate action set uploaded by multiple highway terminal nodes.

[0012] Fourthly, embodiments of this application also provide an expert model-based highway data processing device, which is applied to any highway edge node in an expert model-based highway data processing system. The system further includes a highway cloud node and at least one highway terminal node. The device includes: The receiving module is used to receive the initial incremental causal rules periodically sent by the highway cloud node; each initial incremental causal rule includes a highway event, the corresponding cause of the highway event, the adaptation area attribute value of the causal rule, and the causal correlation strength between the highway event and its corresponding cause. The input module is used to input the initial incremental causal rules into the first highway expert model to obtain the latest highway decision rules; The distribution module is used to distribute the latest highway decision rules to the highway terminal nodes, so that the highway terminal nodes update their corresponding second highway expert models according to the latest highway decision rules, and process the latest real-time collected highway data through the updated second highway expert models.

[0013] Fifthly, embodiments of this application also provide an electronic device, including: a processor, a storage medium, and a bus, wherein the storage medium stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor communicates with the storage medium via the bus, and the processor executes the machine-readable instructions to perform the steps of the expert model-based highway data processing method as described in any of the first aspects.

[0014] In a sixth aspect, embodiments of this application also provide a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of the expert model-based highway data processing method as described in any of the first aspects.

[0015] This application provides a highway data processing method based on an expert model. The method includes: receiving initial incremental causal rules periodically distributed by a highway cloud node; inputting the initial incremental causal rules into a first highway expert model to obtain the latest highway decision rules; distributing the latest highway decision rules to highway terminal nodes, causing the highway terminal nodes to update their corresponding second highway expert models according to the latest highway decision rules, and processing the latest real-time collected highway data using the updated second highway expert model. This application enables the processing of highway data, improving processing efficiency and accuracy. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This illustration shows a flowchart of a highway data processing method based on an expert model, provided in an embodiment of this application. Figure 2 This paper illustrates another expert model-based highway data processing flowchart provided in an embodiment of this application. Figure 3This paper illustrates another expert model-based highway data processing flowchart provided in an embodiment of this application. Figure 4 This paper shows a schematic diagram of the structure of a highway data processing device based on an expert model according to an embodiment of this application; Figure 5 A schematic diagram of the structure of an electronic device provided in an embodiment of this application is shown. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the accompanying drawings in this application are for illustrative and descriptive purposes only and are not intended to limit the scope of protection of this application. Furthermore, it should be understood that the schematic drawings are not drawn to scale. The flowcharts used in this application illustrate operations implemented according to some embodiments of this application. It should be understood that the operations in the flowcharts may not be implemented in sequence, and steps without logical contextual relationships may be reversed or implemented simultaneously. In addition, those skilled in the art, guided by the content of this application, may add one or more other operations to the flowcharts, or remove one or more operations from the flowcharts.

[0019] Furthermore, the described embodiments are merely some, not all, of the embodiments of this application. The components of the embodiments of this application described and illustrated herein can typically be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0020] To enable those skilled in the art to utilize the content of this application, and in conjunction with the specific application scenario of "intelligent transportation technology," the following implementation methods are provided. For those skilled in the art, the general principles defined herein can be applied to other embodiments and application scenarios without departing from the spirit and scope of this application. Although this application is primarily described within the "intelligent transportation technology field," it should be understood that this is merely an exemplary embodiment.

[0021] It should be noted that the term "comprising" will be used in the embodiments of this application to indicate the presence of the features declared thereafter, but does not exclude the addition of other features.

[0022] The following is a detailed description of a highway data processing method based on an expert model provided in the embodiments of this application.

[0023] Reference Figure 1 The diagram shown is a flowchart illustrating a highway data processing method based on an expert model, provided in an embodiment of this application. This method is applied to any highway edge node in a highway data processing system based on an expert model. The system also includes a highway cloud node and at least one highway terminal node. The exemplary steps of this embodiment are described below: S101, Receive the initial incremental causal rules periodically issued by the highway cloud node.

[0024] In this embodiment, at least one highway edge node is managed under a highway cloud node. Each highway edge node manages at least one other highway edge node (road segment node, rescue station, toll station, variable message sign, etc.). The highway edge node uses various managed sensors to collect multimodal real-time highway data (used to characterize the highway's state, such as fog concentration, temperature, traffic flow, etc.), performs preprocessing such as data anonymization, and periodically reports this data to the corresponding highway edge node to obtain incremental highway data. The highway edge node performs data alignment, feature extraction, and other data operations on the incremental highway data reported by multiple highway terminal nodes to obtain incremental highway feature data. The highway edge node periodically reports the incremental highway feature data to the highway cloud node, enabling the highway cloud node to perform federated learning on the incremental highway feature data reported by all highway edge nodes to obtain initial incremental causal rules. The highway cloud node periodically distributes the initial incremental causal rules to each highway cloud node.

[0025] Each initial incremental causal rule includes a highway event, the corresponding cause of the highway event, the applicable region attribute value of the causal rule, and the causal association strength between the highway event and its corresponding cause. Example of an initial incremental causal rule: Cause: "Specific road section (K105 waterfront section) and sudden weather change (fog)"; Highway event: "Accident risk (rear-end collision)", Causal association strength: 85%. This example initial incremental causal rule indicates that a rear-end collision occurred on the highway due to "specific road section (K105 waterfront section) and sudden weather change (fog)". Causal association strength measures the probability that the cause leads to the highway event. For example, the probability of a rear-end collision on the highway due to the specific road section (K105 waterfront section) and sudden weather change (fog) is 85%. The applicable region attribute value refers to the region that applies to this causal rule, such as mountainous areas or waterfront sections.

[0026] S102. Input the initial incremental causal rules into the first highway expert model to obtain the latest highway decision rules.

[0027] In this embodiment, the highway decision-making rule is a rule for making management decisions about highways based on highway data. For example, during the morning rush hour, if the traffic flow on section K105 is greater than 2800 vehicles / hour, the information board will display "Slow traffic warning, detour recommended". The first highway expert model is a deployed lightweight GNN model. The specific implementation process is as follows: Step 1: Input the initial incremental causal rules into the spatiotemporal gating mechanism built into the first highway expert model, filter out the initial incremental causal rules whose corresponding adaptation region attribute values ​​meet the preset region attribute threshold conditions, and obtain the intermediate incremental causal rules.

[0028] In this embodiment, if the correlation between the attribute value of the region to which the highway edge node belongs and the attribute value of the matching region is greater than a preset correlation threshold (e.g., 85%), then the initial incremental causal rule is determined as an intermediate incremental causal rule. The correlation can be calculated using any traditional similarity calculation method, such as cosine similarity. The specific calculation method can be adjusted according to the actual situation.

[0029] Among them, spatiotemporal gating mechanism refers to introducing gating mechanism (such as the gate in LSTM / GRU or the attention gate in Transformer) into the model to dynamically control and filter the flow of information that contains both time and space dimensions.

[0030] Step 2: Input the intermediate incremental causal rules into the dynamic temperature coefficient scheduler built into the first highway expert model, so as to allocate the corresponding distillation temperature to the intermediate incremental causal rules according to the corresponding highway event risk level.

[0031] In this embodiment, the preset distillation temperature corresponding to the risk level of highway events is determined as the distillation temperature corresponding to the intermediate incremental causal rule assignment. Specifically, for high-risk events such as "fog patches," low-temperature fine distillation (τ=0.5) is used to retain decision details; for routine traffic flow statistics, high-temperature fuzzy distillation (τ=5) is used to abstract the data into rule logic, reducing the model size to 1 / 8 of that in the cloud.

[0032] Step 3: Input the intermediate incremental causal rules and their corresponding distillation temperatures into the distillation module built into the first highway expert model, so as to perform graded distillation on the intermediate incremental causal rules according to the corresponding distillation temperatures, and obtain the target incremental causal rules.

[0033] In this application's implementation, knowledge distillation, also known as model distillation, is used to learn the causal rules issued by the highway cloud nodes. Knowledge distillation is an artificial intelligence model compression technique based on a teacher-student model. By transferring knowledge from a large teacher model to a small student model, it aims to reduce deployment costs and improve inference efficiency.

[0034] Step 4: Input the target incremental causal rules into the decision rule generation module built into the first highway expert model, so as to generate the latest highway decision rules in combination with the preset causal decision mapping knowledge base.

[0035] In this embodiment, the causal decision mapping knowledge base stores causal rules and highway decisions (such as dispatching an ambulance). The highway decision corresponding to the incremental causal rule for the target is retrieved from the preset causal decision mapping knowledge base, and the latest highway decision rule is generated.

[0036] Here, this application employs a hierarchical federated distillation mechanism in real time, enabling student models (highway cloud nodes) to distill and learn causal rules distributed by teacher models (highway cloud nodes). This addresses the difficulty of deploying large cloud models on edge and terminal devices. By constructing a three-tiered collaborative architecture of "cloud-edge-device," it achieves efficient knowledge transfer, adaptive compression, and agile response. The hierarchical federated distillation mechanism is a federated learning framework that combines model hierarchical strategies and knowledge distillation techniques, aiming to solve the collaboration efficiency problems caused by heterogeneous device computing capabilities and diverse model architectures. This mechanism transfers the knowledge of complex models (teachers) to lightweight models (students) in the form of soft labels, feature maps, or parameters, enabling collaborative training of heterogeneous models under privacy protection. Simultaneously, a hierarchical structure is introduced to dynamically allocate model sub-networks based on the client's computing resources, reducing device burden and improving the overall model quality.

[0037] S103. The latest highway decision rules are distributed to the highway terminal nodes, so that the highway terminal nodes update their corresponding second highway expert models according to the latest highway decision rules, and process the latest real-time collected highway data through the updated second highway expert models.

[0038] In this embodiment, a miniature second highway expert model at the highway terminal node is automatically generated through Neural Architecture Search (NAS) and adapted to hardware resources (parameter size <100KB). The second highway expert model makes decisions based on highway data by combining a knowledge graph containing highway decision rules, and its AI model structure can be determined according to the actual situation.

[0039] Optionally, the second highway expert model can utilize an event importance sampling mechanism and a spatiotemporal decay factor (geographic weight a=1 / d). 2 Time decay β=e^( λt)) enables rapid localized incremental learning of highway decision rules issued to highway edge nodes. d represents the geographical distance between the location of the highway event and the geographical location of the highway terminal node. t represents the time of the highway event.

[0040] Neural Architecture Search (NAS) is an automated technique designed to enable computer algorithms to automatically design and optimize the structure of neural networks to achieve optimal performance on specific tasks. The second highway expert model adjusts its focus on highway decision rules based on their importance, and gives greater attention to more recently generated rules based on a spatiotemporal decay factor, thus updating the second highway expert model. The input to the second highway expert model is various data (such as weather) extracted from highways via sensors, and the output is a set of candidate actions for highway management (e.g., dispatching two rescue vehicles, closing ramp C). The event importance sampling mechanism assigns different weights to different highway events; the greater the risk of a highway event, the higher its weight (e.g., traffic accident events have a higher weight than traffic flow data collection events).

[0041] In addition, highway edge nodes are also used to play a game on candidate action sets generated from the latest highway data uploaded by multiple highway terminal nodes. The specific implementation process is as follows: Step 1: Receive candidate action sets generated from the latest highway data uploaded by multiple highway terminal nodes.

[0042] In this embodiment, the highway terminal node acquires the latest highway data through corresponding sensors; the highway terminal node inputs the latest highway data into the second highway expert model within the highway terminal node to obtain a candidate action set; the highway terminal node sends the candidate action set to the corresponding highway edge node so that the highway edge node selects the final decision action set from the candidate action sets uploaded by multiple highway terminal nodes.

[0043] Step 2: For each candidate action set, calculate the decision score corresponding to the candidate action set based on the resource coverage benefit, scheduling cost, and synergistic gain with at least one other candidate action set.

[0044] In this application's implementation, a potential game framework and a defined utility function are introduced: The decision score for each candidate action set is calculated by weighting and summing the resource coverage gain, scheduling cost, and collaborative gain with other candidate action sets for each candidate action set. , and These are the preset weighting coefficients.

[0045] Resource coverage benefit quantifies the positive value of adopting the candidate action set for the highway and can be implemented by pre-training a corresponding AI model (input is a sample of candidate action sets, output is resource coverage benefit). Scheduling cost measures the cost required to adopt the candidate action set and can also be implemented by pre-training a corresponding AI model. Collaboration gain measures the incremental positive value to the highway when collaborating with other candidate action sets based on a given set. For example, rescue station A has low cost in dispatching an ambulance, but its benefits are limited. When it learns through differential privacy-protected communication that another rescue station B has a heavy crane available for collaborative action, its collaboration gain term will significantly increase, thus incentivizing it to make a dispatch decision.

[0046] Step 3: Determine the candidate action set with the largest decision score as the first candidate action set; determine the second candidate action set whose cooperative gain with the first candidate action set is greater than or equal to a preset cooperative gain value (such as 0) as the final decision action set.

[0047] In the embodiments of this application, the higher the decision score, the better the candidate action set.

[0048] Step 4: Process the latest highway data according to the decision action set.

[0049] Reference Figure 2 The diagram shown is another expert model-based highway data processing flowchart provided in this application embodiment: This method is applied to highway terminal nodes in an expert model-based highway data processing system. The system further includes a highway cloud node and a highway edge node. The method also includes: S201, Obtain the latest highway data.

[0050] In this application embodiment, the latest highway data is collected from various sensors (such as cameras, temperature sensors, etc.) managed by the highway terminal node.

[0051] S202. Input the latest highway data into the second highway expert model within the highway terminal node to obtain a set of candidate actions.

[0052] In this embodiment, the second highway expert model is a model that combines a dynamic highway decision rule base to make highway decisions. The input is highway data, and the output is a set of candidate actions. The network architecture of the model can be determined according to the actual situation.

[0053] The highway decision-making rule base includes multiple highway decision-making rules.

[0054] S203. Send the candidate action set to the corresponding highway edge node so that the highway edge node can select the final decision action set from the candidate action set uploaded by multiple highway terminal nodes.

[0055] Furthermore, the second highway expert model can also utilize counterfactual reinforcement learning techniques for decision-making, integrating causal inference with deep reinforcement learning to construct a causal-driven decision simulation environment. This enables highly reliable policy deduction, allowing the system to "review" and "pre-play" scenarios. For example, the system builds a high-fidelity virtual environment based on a pre-constructed dynamic causal graph (containing causal rules for multiple highways). Then, counterfactual inference is performed using Do-Calculus (intervention calculus), which raises "what-if" questions, such as: "If we had limited the speed to 60 km / h on that section of road when visibility dropped to 100 meters, how much would the subsequent congestion duration have been reduced?" By running simulations of real and counterfactual scenarios in parallel, the system calculates the causal effect size (CES) of the intervention, such as "the expected reduction in congestion duration Δt = 42 minutes." This allows the policy network to learn from "virtual, potentially better outcomes." During training, the system actively injects extreme virtual events (what-if) through an adversarial perturbation generator to improve the model's robustness to rare risks. All counterfactual simulation data is synthesized using a differential privacy generative adversarial network (DP-GAN) to ensure irreversible anonymization of individual-level information (k Anonymity ≥ 50%.

[0056] Here, refer to Figure 3 The diagram shown is another flowchart of highway data processing based on an expert model provided in this application embodiment. Its specific implementation process is as follows: S301. Receive incremental feature data of highways periodically reported by at least one highway edge node.

[0057] In this embodiment, highway edge nodes utilize various managed sensors to collect real highway data (characterizing highway conditions such as fog concentration, temperature, and traffic flow). A dynamic desensitization technique driven by a Generative Adversarial Network (GAN) is used to blur information such as license plates (with an accuracy of 99.2%), yielding incremental highway data. The highway edge nodes then use a dual-stream Transformer network to uniformly encode spatial coordinates and timestamps, generating spatiotemporal embedding vectors. Subsequently, an adversarial modal projection network, through adversarial training, maps features from video, radar, and text to a shared latent space, achieving second-level correlation confirmation of incremental highway data. Simultaneously, the system dynamically calculates the weights of each incremental highway data point based on a data quality assessment model (including 12 indicators such as signal-to-noise ratio and sampling completeness), ensuring that the weighted average alignment error of cross-domain feature calibration is controlled within 5%.

[0058] S302. Input all incremental characteristic data of highways into the causal analysis model to obtain the initial incremental causal rules.

[0059] In this embodiment, the trained improved multimodal BERT++ module extracts semantic-level information from incremental highway feature data through pre-training tasks (such as cross-modal matching of video clips and accident reports, and reconstruction of occluded license plates). Then, by combining Granger causality tests and Bayesian networks to mine potential causal rules (such as "continuous heavy rainfall for 3 hours increases the risk of landslides by 72%), initial incremental causal rules are obtained. The causal analysis model consists of two parts: the multimodal BERT++ module and Granger causality tests combined with Bayesian networks. Bayesian networks, also known as confidence networks, are among the most effective theoretical models in the field of uncertain knowledge representation and reasoning. Granger causality tests are statistical hypothesis testing methods used to determine whether one time series can significantly improve the predictive ability of another time series. Its core logic is: if the past values ​​of variable X can help predict the future values ​​of variable Y, then X is said to be a Granger cause of Y.

[0060] In addition, highway cloud nodes can also use the global model of Transformer-XL to identify high-value event features in highway incremental feature data through federated aggregation and knowledge entropy evaluation algorithm (the inverse of the similarity between highway incremental feature data and historical highway feature data can be used as the knowledge entropy evaluation value. The higher the knowledge entropy evaluation value, the higher the value of highway incremental feature data). This allows for the joint extraction of causal rules, forming a causal rule library, which is then distributed to highway edge nodes for learning.

[0061] Alternatively, the causal analysis model can also be an AI model with any network structure, which can be selected according to the actual situation, as long as its function can be realized.

[0062] S303. The initial incremental causal rules are sent to each highway edge node.

[0063] Furthermore, this application's embodiments achieve end-to-end privacy protection and efficient storage: privacy protection is maintained throughout. Data interaction between highway terminal nodes and highway edge nodes employs homomorphic encryption for transmission parameters. When data aggregation occurs, highway cloud nodes inject Gaussian noise compliant with GDPR standards to achieve differential privacy. For storage, real-time event graphs (used to store highway data) are stored in an in-memory database to support millisecond-level response times, while historical highway data is archived in a graph database, enabling decade-level time-series backtracking, which can be used for causal rule analysis.

[0064] Based on the same inventive concept, this application also provides a highway data processing device based on an expert model, which corresponds to the highway data processing method based on an expert model. Since the principle of the device in this application is similar to the highway data processing method based on an expert model described above, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be described again.

[0065] Reference Figure 4 The diagram shown is a structural schematic of a highway data processing device based on an expert model, provided in an embodiment of this application. This device is applied to any highway edge node in a highway data processing system based on an expert model. The system also includes a highway cloud node and at least one highway terminal node. The device includes: The receiving module 401 is used to receive the initial incremental causal rules periodically sent by the highway cloud node; each initial incremental causal rule includes a highway event, the corresponding cause of the highway event, the adaptation area attribute value of the causal rule, and the causal correlation strength between the highway event and its corresponding cause. Input module 402 is used to input the initial incremental causal rules into the first highway expert model to obtain the latest highway decision rules; The distribution module 403 is used to distribute the latest highway decision rules to the highway terminal node, so that the highway terminal node updates its corresponding second highway expert model according to the latest highway decision rules, and processes the latest real-time collected highway data through the updated second highway expert model.

[0066] like Figure 5As shown in the embodiment of this application, an electronic device 500 includes a processor 501, a memory 502, and a bus. The memory 502 stores machine-readable instructions executable by the processor 501. When the electronic device is running, the processor 501 communicates with the memory 502 via the bus. The processor 501 executes the machine-readable instructions to perform the steps of the highway data processing method based on the expert model described above.

[0067] Specifically, the memory 502 and processor 501 mentioned above can be general-purpose memory and processor, without any specific limitations. When the processor 501 runs the computer program stored in the memory 502, it can execute the above-mentioned highway data processing method based on the expert model.

[0068] Corresponding to the above-described expert model-based highway data processing method, this application embodiment also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of the above-described expert model-based highway data processing method.

[0069] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and devices described above can be referred to the corresponding processes in the method embodiments, and will not be repeated here. In the several embodiments provided in this application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed mutual coupling or direct coupling or communication connection can be through some communication interfaces; the indirect coupling or communication connection of devices or modules can be electrical, mechanical, or other forms.

[0070] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0071] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0072] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, ROM, RAM, magnetic disks, or optical disks.

[0073] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A highway data processing method based on expert models, characterized in that, This method is applied to any highway edge node in an expert model-based highway data processing system, the system also including a highway cloud node and at least one highway terminal node, and the method includes: Receive initial incremental causal rules periodically sent by the highway cloud node; each initial incremental causal rule contains a highway event, the cause of the highway event, the adaptation area attribute value of the causal rule, and the causal correlation strength between the highway event and its corresponding cause. The initial incremental causal rules are input into the first highway expert model to obtain the latest highway decision rules. The latest highway decision rules are sent to the highway terminal nodes, enabling the highway terminal nodes to update their corresponding second highway expert models based on the latest highway decision rules, and to process the latest real-time collected highway data through the updated second highway expert models.

2. The highway data processing method based on expert models according to claim 1, characterized in that, The step of inputting the initial incremental causal rule into the first highway expert model to obtain the latest highway decision rule includes: The initial incremental causal rules are input into the spatiotemporal gating mechanism built into the first highway expert model to filter out the initial incremental causal rules whose corresponding adaptation region attribute values ​​satisfy the preset region attribute threshold conditions, and thus obtain the intermediate incremental causal rules. The intermediate incremental causal rules are input into the dynamic temperature coefficient scheduler built into the first highway expert model, so as to allocate the corresponding distillation temperature to the intermediate incremental causal rules according to the corresponding highway event risk level. The intermediate incremental causal rules and their corresponding distillation temperatures are input into the distillation module built into the first highway expert model, so as to perform graded distillation on the intermediate incremental causal rules according to the corresponding distillation temperatures to obtain the target incremental causal rules. The target incremental causal rule is input into the decision rule generation module built into the first highway expert model to generate the latest highway decision rule in combination with the preset causal decision mapping knowledge base.

3. The highway data processing method based on expert models according to claim 2, characterized in that, The process involves inputting the initial incremental causal rules into the spatiotemporal gating mechanism built into the first highway expert model to filter out the initial incremental causal rules whose corresponding adaptive region attribute values ​​satisfy the preset region attribute threshold conditions, thereby obtaining intermediate incremental causal rules, including: If the correlation between the attribute value of the region to which the highway edge node belongs and the attribute value of the adaptation region is greater than a preset correlation threshold, then the initial incremental causal rule is determined as an intermediate incremental causal rule.

4. The highway data processing method based on expert models according to claim 2, characterized in that, The step of inputting the intermediate incremental causal rules into the dynamic temperature coefficient scheduler built into the first highway expert model, so as to allocate the corresponding distillation temperature to the intermediate incremental causal rules according to the corresponding highway event risk level, includes: The preset distillation temperature corresponding to the risk level of the highway event is determined as the distillation temperature corresponding to the intermediate incremental causal rule assignment.

5. The highway data processing method based on expert models according to claim 2, characterized in that, The method further includes: Receive candidate action sets generated from the latest highway data uploaded by multiple highway terminal nodes; For each candidate action set, a decision score is calculated based on the resource coverage benefit, scheduling cost, and synergistic gain with at least one other candidate action set. The set of candidate actions with the highest decision score is determined as the first candidate action set; The second candidate action set whose cooperative gain with the first candidate action set is greater than or equal to a preset cooperative gain value is determined as the final decision action set; The latest highway data is processed based on the set of decision actions.

6. A highway data processing method based on an expert model, characterized in that, This method is applied to a highway cloud node in a highway data processing system based on an expert model as described in any one of claims 1 to 5, wherein the system further includes at least one highway edge node and at least one highway terminal node, and the method comprises: Receive incremental feature data of highways periodically reported by at least one highway edge node; All incremental characteristic data of highways are input into the causal analysis model to obtain the initial incremental causal rules; The initial incremental causal rules are then distributed to each highway edge node.

7. A highway data processing method based on an expert model, characterized in that, This method is applied to a highway terminal node in a highway data processing system based on an expert model as described in any one of claims 1 to 6, wherein the system further includes a highway cloud node and a highway edge node, and the method further includes: Get the latest highway data; The latest highway data is input into the second highway expert model within the highway terminal node to obtain a set of candidate actions; The candidate action set is sent to the corresponding highway edge node so that the highway edge node can select the final decision action set from the candidate action set uploaded by multiple highway terminal nodes.

8. A highway data processing device based on an expert model, characterized in that, This device is applied to any highway edge node in an expert model-based highway data processing system, which also includes a highway cloud node and at least one highway terminal node. The device comprises: The receiving module is used to receive the initial incremental causal rules periodically sent by the highway cloud node; each initial incremental causal rule includes a highway event, the corresponding cause of the highway event, the adaptation area attribute value of the causal rule, and the causal correlation strength between the highway event and its corresponding cause. The input module is used to input the initial incremental causal rules into the first highway expert model to obtain the latest highway decision rules; The distribution module is used to distribute the latest highway decision rules to the highway terminal nodes, so that the highway terminal nodes update their corresponding second highway expert models according to the latest highway decision rules, and process the latest real-time collected highway data through the updated second highway expert models.

9. An electronic device, characterized in that, include: The device includes a processor, a storage medium, and a bus, wherein the storage medium stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor communicates with the storage medium via the bus, and the processor executes the machine-readable instructions to perform the steps of the expert model-based highway data processing method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, performs the steps of the highway data processing method based on an expert model as described in any one of claims 1 to 7.

Images