Adaptive and interpretable traffic situation prediction method and system based on large language model

Abstract

The invention discloses an adaptive and interpretable traffic situation prediction method and system based on a large language model, and the method comprises the following steps: obtaining historical traffic data related to a prediction task, inputting a trained time sequence prediction basic model, and generating candidate traffic state tracks of a plurality of prediction periods based on the prediction task; acquiring context information related to a prediction time period, and evaluating the candidate traffic state trajectory based on a configured large language reasoning model to obtain an optimal trajectory conforming to the context information of the prediction time period; and based on the optimal track and the context information, generating a text report by utilizing the configured large language interpretation model. Probability prediction is carried out through the time sequence prediction basic model, context reasoning and text generation are carried out through the large language model, the accuracy, adaptability and interpretability of traffic prediction under abnormal events are remarkably improved, and powerful support is provided for intelligent traffic.

Description

Technical Field

[0001] The present invention relates to the field of intelligent transportation technology, and more particularly to an adaptive and explainable traffic situation prediction method and system based on a large language model. Background Art

[0002] Currently, accurate traffic forecasting is fundamental to enabling intelligent transportation systems to achieve key functions such as congestion relief, dynamic routing guidance, and emergency management. In recent years, deep learning methods, such as graph neural networks (GNNs) and sequence models, have made significant progress in short-term forecasting under typical traffic conditions. These models effectively learn the spatiotemporal patterns inherent in historical data.

[0003] However, existing technologies face three major bottlenecks in practical applications: (1) Vulnerability under abnormal events: When unusual events such as traffic accidents, severe weather, road construction, or large-scale events occur, the distribution of traffic data changes significantly, causing the performance of prediction models that rely mainly on historical regular patterns to drop sharply. It is precisely at these moments that accurate predictions are crucial for emergency response. (2) Limitations of contextual information fusion: Existing models usually use a fixed input structure and are difficult to flexibly integrate diverse real-time contextual information. For example, dynamically integrating an accident report described in natural language or a structured construction plan into the prediction process often requires complex feature engineering or model retraining, which has poor adaptability. (3) Lack of interpretability and operability: The output of traditional prediction models is usually a high-dimensional numerical matrix (such as the speed or flow of each road section in the future). It does not provide any analysis of the reasons behind the prediction results, nor can it be directly converted into an action guide that traffic managers can understand and implement. This "black box" characteristic limits its application value in actual decision support.

[0004] Therefore, existing traffic forecasting technologies lack robustness and flexibility when facing unexpected scenarios, and are unable to support real-time decision-making needs. There is an urgent need for a solution that can integrate contextual information, has explanatory capabilities, and is generalizable.

[0005] Large language models (LLMs), on the other hand, demonstrate powerful natural language understanding, contextual reasoning, and generative capabilities, offering new possibilities for addressing these challenges. However, despite their powerful textual reasoning and generative capabilities, LLMs are inherently ill-suited for predicting numerical continuous variables, particularly in the absence of numerical perception, which limits their accuracy. Therefore, relying solely on LLMs cannot meet the stringent quantitative accuracy requirements of traffic forecasting.

[0006] Therefore, how to effectively combine the numerical modeling capabilities of time series prediction models with the contextual reasoning capabilities of large language models to develop a new paradigm for traffic prediction that is both accurate and has good adaptability and interpretability is a technical challenge that needs to be solved urgently in this field. Summary of the Invention

[0007] In view of this, the present invention provides an adaptive and interpretable traffic situation prediction method and system based on a large language model. By combining the probabilistic prediction ability of the time series model with the contextual reasoning and generation ability of the large language model, the accuracy, adaptability and interpretability of traffic prediction under abnormal events are significantly improved, providing strong support for the realization of a more efficient and resilient intelligent transportation system.

[0008] In order to achieve the above object, the present invention adopts the following technical solutions:

[0009] In a first aspect, the present invention provides an adaptive and interpretable traffic situation prediction method based on a large language model, comprising the following steps:

[0010] S1. Obtain historical traffic data related to the prediction task, input the trained time series prediction basic model, and generate multiple candidate traffic state trajectories for the prediction period based on the prediction task;

[0011] S2. Obtain context information related to the predicted period, and evaluate the candidate traffic state trajectories based on the configured large language inference model to obtain an optimal trajectory that meets the context information of the predicted period;

[0012] S3. Based on the optimal trajectory and context information, a text report including traffic situation analysis and actionable traffic management suggestions is generated using a configured large language interpretation model.

[0013] Furthermore, the time series prediction basic model described in step S1 is constructed based on the Transformer architecture;

[0014] By segmenting and quantifying numerical traffic time series data, the spatiotemporal dependency of traffic flow is captured, and the congestion probability and vehicle speed in the corresponding time period are predicted; thus, multiple corresponding candidate traffic state trajectories are obtained.

[0015] Furthermore, the historical traffic data in step S1 is numerical time series data, including historical traffic flow and speed data of the predicted road section and surrounding areas; {1:T} Indicates that T is the current moment;

[0016] Based on the K candidate traffic state trajectories generated from the historical traffic data during the prediction period, {X k} {T+1:T+n}Represents; k = 1, ..., K; where k is the number of candidate trajectories; n is the prediction step size.

[0017] Furthermore, the context information related to the prediction period in step S2 includes structured data and unstructured data;

[0018] The structured data includes weather forecasts and scheduled road construction plans;

[0019] The unstructured data includes traffic accident reports and large-scale public event notifications in natural language format.

[0020] Furthermore, the reasoning process of the large language reasoning model in step S2 is:

[0021] Performing semantic understanding on the context information to analyze its causal impact on traffic flow, the causal impact including the location, severity, and duration of the impact;

[0022] An optimal trajectory is selected from the plurality of candidate traffic state trajectories for the prediction period based on the causal influence.

[0023] Furthermore, the step S2 is expressed as follows:

[0024]

[0025] in, represents the optimal trajectory that meets the context information of the prediction period; F LLM represents the adjustment function of the large language inference model; {X k} {T+1:T+n} represents the set of K candidate traffic state trajectories for the prediction period; S represents the context information related to the prediction period.

[0026] Furthermore, the process of obtaining the optimal trajectory that conforms to the context information of the prediction period described in step S2 is equivalent to solving a posterior probability maximization problem conditional on s;

[0027] The solution formula is:

[0028]

[0029] where P(·|s) represents the posterior plausibility score of each candidate trajectory given the context information s, as evaluated by the large language inference model.

[0030] Furthermore, the operational traffic management suggestions in step S3 specifically include:

[0031] Suggestions on adjusting signal light timing for corresponding road sections, traffic diversion or detour strategies, and travel information release recommendations for the public.

[0032] In a second aspect, the present invention provides an adaptive and interpretable traffic situation prediction system based on a large language model, comprising the following modules:

[0033] Trajectory generation module: used to obtain historical traffic data related to the prediction task, input the trained time series prediction basic model, and generate multiple candidate traffic state trajectories for the prediction period based on the prediction task;

[0034] Trajectory evaluation module: used to obtain context information related to the predicted period and evaluate the candidate traffic state trajectories based on the configured large language inference model to obtain the optimal trajectory that meets the context information of the predicted period;

[0035] Trajectory traffic interpretation module: used to generate a text report containing traffic situation analysis and actionable traffic management suggestions based on the optimal trajectory and context information using the configured large language interpretation model.

[0036] As can be seen from the above technical solutions, compared with the prior art, the present invention discloses an adaptive and interpretable traffic situation prediction method and system based on a large language model, which has the following beneficial effects:

[0037] This method uses a time series prediction model to process high-dimensional, noisy numerical data, generating multiple possible trajectories. This is then combined with a large language model to perform reasoning and decision-making based on external, semantically rich contextual information. This avoids requiring the large language model to directly handle precise numerical regression tasks, which it is not adept at, and also avoids the poor generalization of traditional models to unseen external events. Furthermore, it further improves the accuracy of abnormal event prediction.

[0038] The large language model of this invention uses different system instructions and task prompts to perform analytical reasoning and decision interpretation. It processes various structured and unstructured information through natural language prompts, and has zero-shot analysis and response capabilities for new types of events not seen during training. This enhances the model's adaptability and generalization capabilities. It also transforms purely numerical predictions into intelligent decision support with accompanying cause analysis and specific action recommendations, bridging the critical gap between "prediction" and "action," greatly enhancing the practical application value of the prediction system and achieving both interpretability and operability of predictions. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are merely embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0040] Figure 1 Flowchart of the adaptive and interpretable traffic situation prediction method based on a large language model provided by an embodiment of the present invention.

[0041] Figure 2 A schematic diagram of the implementation flow of the adaptive and explainable traffic situation prediction method based on a large language model provided in an embodiment of the present invention.

[0042] Figure 3 This is a structural diagram of the adaptive and explainable traffic situation prediction system based on a large language model provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0043] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0044] Example 1

[0045] The embodiment of the present invention discloses an adaptive and interpretable traffic situation prediction method based on a large language model, referring to Figure 1 As shown, the following steps are included:

[0046] S1. Obtain historical traffic data related to the prediction task, input the trained time series prediction basic model, and generate multiple candidate traffic state trajectories for the prediction period based on the prediction task;

[0047] S2. Obtain contextual information related to the prediction period and evaluate candidate traffic state trajectories based on the configured large language inference model to obtain the optimal trajectory that meets the contextual information of the prediction period;

[0048] S3. Based on the optimal trajectory and context information, a configured large language interpretation model is used to generate a text report containing traffic situation analysis and actionable traffic management suggestions.

[0049] This embodiment is used to predict a planned road construction event. The traffic management center receives a notification: "A certain city's main road will close one inner lane of each of the eight lanes in both directions from 9:00 am to 1:00 pm tomorrow due to road maintenance."

[0050] The embodiment adopts a multi-stage collaborative framework to predict the traffic situation during road maintenance; separates numerical prediction from context reasoning, adopts a modular design idea, and completes the prediction process through three stages of "numerical prediction-context reasoning-decision generation", which has clear structure, strong interpretability and good adaptability.

[0051] The implementation steps of the embodiment are:

[0052] Step 1: Probabilistic trajectory generation: based on historical traffic time series data, a set of multiple (K, K > 1) candidate future traffic state trajectories is generated through a time series prediction base model; wherein the multiple candidate trajectories collectively constitute a probabilistic prediction of the future traffic state;

[0053] Step 2: Context reasoning and trajectory calibration: obtain real-time context information related to the prediction period, input the context information and the multiple candidate trajectory set generated in step 1 into a large language model configured for analysis and selection tasks; the large language model performs causal reasoning on the context information, evaluates the matching degree of each candidate trajectory with the causal reasoning, and accordingly selects or adjusts an optimal future traffic state prediction trajectory;

[0054] Step 3: Explanatory report and suggestion generation: input the optimal prediction trajectory determined in step 2 and the context information into a large language model configured for explanation and generation tasks, and generate a human-readable text report containing traffic situation analysis and actionable management suggestions.

[0055] The implementation steps of the embodiment are described in detail as follows:

[0056] According to step 1, first obtain historical traffic time series data.

[0057] The embodiment first obtains the historical traffic flow and speed data of the road section and the surrounding area in the past few hours before the prediction starts (for example, at 8:30 am), and forms a historical traffic state sequence X {1:T} for processing. Wherein, X {1:t} ={x1,...,x T}, t = 1,...,T; T is the current time; x t is the traffic state vector at time t on the monitoring position N.

[0058] Secondly, a time series prediction base model is constructed based on the Transformer architecture; through tokenization and quantization processing of numerical traffic time series data, the traffic flow spatio-temporal dependence is captured, and the congestion probability and vehicle speed of the corresponding time period are predicted; and multiple candidate traffic state trajectories are obtained.

[0059] The basic time series forecasting model specifically extracts spatiotemporal patterns from historical traffic data and generates multiple probabilistic future trajectories, providing a decision space for subsequent contextual reasoning in the large language model. Its input is structured numerical time series data (speed, flow, occupancy); its output is multiple candidate future traffic state trajectories (representing uncertainty). The advantage of the basic time series forecasting model lies in its use of word segmentation and Transformer to overcome the limitations of traditional time series models in modeling long-range dependencies and probabilistic prediction.

[0060] Among them, the data segmentation and quantization processing specifically includes: numerical discretization and spatiotemporal embedding coding.

[0061] Numeric discretization converts continuous values ​​into discrete intervals: the original traffic data is divided into B ordered intervals. For example, the vehicle speed [0,120] km / h is divided into intervals of 5 km / h each. B = 24, including the intervals [0,5), [5,10), …, [115,120]. Each interval is assigned a unique ID to construct a traffic semantic vocabulary. The discrete token processing paradigm of the Transformer is then adapted to preserve the numerical ordinal ID relationship and avoid information loss.

[0062] Next, time embedding is performed; the input timestamp t is decomposed into [hour, day of the week, whether it is a holiday], and the output is d-dimensional spatial embedding; the road section topology map is used as input, and the graph neural network generates a d-dimensional position encoding; the feature splicing result is: Token = Embedding (numeric semantics of feature interval ID ⊕ time context t ⊕ spatial position relationship).

[0063] The time series prediction basic model of this embodiment does not output a single deterministic prediction, but generates a set of K candidate future traffic state trajectories {X k} {T+1:T+n} , where k is the index of the candidate trajectory k = 1, ..., K; n is the number of future time steps to predict. This set represents the probabilistic distribution of future traffic states and captures the inherent uncertainty of the system.

[0064] The model in this embodiment outputs 20 candidate traffic state trajectories for the period from 9:00 a.m. to 2:00 p.m. Among these trajectories, some show a normal post-peak congestion pattern, while others may show slightly increased congestion due to minor disturbances in historical data.

[0065] This embodiment also evaluates these trajectories; three evaluation indicators are used to evaluate the effectiveness of spatiotemporal traffic state prediction: (1) mean absolute error (MAE); (2) mean absolute percentage error (MAPE); and (3) mean standard error (RMSE).

[0066] By selecting the short-term (15 minutes), medium-term (30 minutes), and long-term (60 minutes) prediction results, we calculated the three evaluation indicators and obtained the results shown in Table 1:

[0067]

[0068] Table 1 Comparison of the results of three evaluation indicators

[0069] It can be seen that the prediction in step 1 of this embodiment is accurate and reliable in terms of multiple evaluation indicators and multiple dimensions, laying a good foundation for subsequent steps.

[0070] Next, follow step 2, refer to Figure 2 As shown in Figure 2, the optimal trajectory is obtained by combining a large language model with contextual information for evaluation.

[0071] In this embodiment, the multimodal context information s (including text, table, etc.) related to the prediction period obtained in real time is combined with the candidate trajectory set {X k} {T+1:T+n} The data is then fed into a large language model. This large language model is a pre-configured large language reasoning model for analysis and selection tasks. By performing causal reasoning on contextual information, the matching degree between each candidate trajectory and the causal reasoning is evaluated, and the optimal trajectory for predicting future traffic conditions is selected or adjusted accordingly.

[0072] The large language inference model of this embodiment is used as a context adjustment function F LLM , perform comprehensive reasoning on the information and output an optimal prediction trajectory after context correction The process can be expressed as: The function F LLM The internal logic of is to first understand the causal impact of the event described in s, and then evaluate each candidate trajectory X k The degree of matching with the impact ultimately selects or integrates the most reasonable trajectory.

[0073] This embodiment combines the following information into a prompt and inputs it into the large language base model (such as Qwen2.5-14B-Instruct).

[0074] Context: "Event: Planned construction. Location: Main Road X. Time: 9:00 AM to 1:00 PM. Details: Closure of one inner lane of each of the eight lanes in both directions, reducing lane capacity by 25%."

[0075] Candidate trajectory summary: Provides summary information for 20 candidate trajectories, such as "Candidate 1: Congestion peaks at 9:30, vehicle speed 35 km / h", "Candidate 2: Congestion continues until 11:00, minimum speed 30 km / h", etc.

[0076] Mission Instructions: "Based on the above event information, please evaluate its impact on traffic and select the most likely candidate trajectory from the following."

[0077] The large language model performs the following reasoning:

[0078] "Closing 25 per cent of the lanes occurs during the off-peak period just after the morning rush hour, but the reduced capacity will create a bottleneck effect."

[0079] "Congestion levels will be significantly higher than on a normal working day, speeds will be significantly reduced, and the duration will cover the entire construction period."

[0080] "The projections for Candidates 1 and 2 are overly optimistic and underrepresent the severe impact of lane closures."

[0081] Based on this reasoning, the large language model of this embodiment selects or synthesizes a new trajectory that shows that the vehicle speed drops sharply to below 20 km / h starting at 9:00 and continues to slowly recover after 1:00 p.m. This trajectory is the optimal predicted trajectory.

[0082] Finally, according to step 3, refer to Figure 2 As shown, the optimal predicted trajectory and context information determined in step 2 are input into a large language model configured for interpretation and generation tasks, which generates a human-readable text report containing traffic situation analysis and actionable management recommendations.

[0083] In this embodiment, the optimal prediction trajectory determined in step 2 is The context information s that led to the adjustment is input into the large language model again. The large language model has been configured as a large language interpretation model for interpretation and generation tasks through prompt instructions. The model is instructed to generate two parts: (i) a concise, human-readable traffic situation analysis report that explains the expected traffic conditions and their causes (for example, "Affected by the G6 Expressway accident, congestion on this section of road is expected to worsen in the next hour, with the average vehicle speed dropping below 20 kilometers per hour"); (ii) a series of specific and actionable traffic management suggestions (for example, "It is recommended to open the upstream ramp control of the accident and guide vehicles to detour from the XX exit through the information board").

[0084] In this example, the optimal predicted trajectory and construction information obtained in step 2 are input into the large language interpretation model, and the instruction is issued: "Generate a traffic situation report and management recommendations based on the predicted results and construction events." The large language interpretation model outputs the following text report:

[0085] Situation Analysis Report: "Warning: Major Road X will close some lanes due to construction from 9:00 AM to 1:00 PM, causing significant congestion. Forecast data indicates that the average speed on this section will drop below 20 km / h, and delays may exceed 30 minutes. Congestion is expected to spread upstream to Intersection Y."

[0086] Actionable management suggestions:

[0087] Public information release: "It is recommended to immediately issue warnings through traffic broadcasts and mobile map apps to prompt drivers to plan detour routes in advance."

[0088] Traffic diversion: "Consider implementing temporary traffic control at intersection Z upstream of main road X to redirect some traffic onto the parallel alternative route E."

[0089] Signal coordination and control: "It is recommended to adjust the signal light timing plan at each intersection along Alternative Route E to increase its green light time to accommodate the additional detour traffic flow."

[0090] Through this embodiment, the method of the present invention successfully transforms a simple construction notice into an accurate and adaptive prediction of future traffic conditions, and ultimately generates decision support information with high interpretability and practical guidance value, fully demonstrating the technical concept and superior effects of the present invention.

[0091] Example 2

[0092] The embodiment of the present invention discloses an adaptive and interpretable traffic situation prediction system based on a large language model, referring to Figure 3 As shown, it includes the following modules:

[0093] Trajectory generation module: used to obtain historical traffic data related to the prediction task, input the trained time series prediction basic model, and generate multiple candidate traffic state trajectories for the prediction period based on the prediction task;

[0094] Trajectory evaluation module: used to obtain context information related to the predicted period and evaluate the candidate traffic state trajectories based on the configured large language inference model to obtain the optimal trajectory that meets the context information of the predicted period;

[0095] Trajectory traffic interpretation module: used to generate a text report containing traffic situation analysis and actionable traffic management suggestions based on the optimal trajectory and context information using the configured large language interpretation model.

[0096] The prediction system in this embodiment features a unique task decoupling and collaborative design. First, the trajectory generation module decomposes the complex traffic prediction task into two parts: the time series prediction base model processes high-dimensional, noisy numerical data and generates multiple possibilities; then, the large language model reasones and makes decisions based on external, semantically rich contextual information. This avoids requiring the large model to directly handle precise numerical regression tasks, which it is not adept at, and also avoids the dilemma of traditional models' poor generalization when faced with unseen external events.

[0097] Secondly, contextual reasoning and trajectory calibration are performed through the trajectory evaluation module. The large language model is used as an analytical reasoning engine, whose input is many-to-many multiple trajectories and multi-source information. The task goal is to converge to an optimal solution.

[0098] Finally, explanatory reports and suggestions are generated through the trajectory traffic interpretation module. The large language model is used as a narrative generation engine with a single input (an optimal trajectory) and the task goal is to generate explanatory and instructive texts in a divergent manner.

[0099] The large language model of this embodiment uses the same LLM base, but achieves clear functional distinction through different system instructions and task prompts. Through the dual functional positioning of LLM, it together constitutes the complete technical chain of the present invention from accurate prediction to intelligent decision-making.

[0100] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Reference can be made to the common and similar parts between the various embodiments. The system devices disclosed in the embodiments correspond to the methods disclosed in the embodiments, so the description is relatively simple. For relevant parts, refer to the method description.

[0101] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not limited to the embodiments shown herein but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An adaptive and interpretable traffic situation prediction method based on a large language model, characterized by: The following steps are involved: S1. Obtain historical traffic data related to the prediction task, input the trained time series prediction basic model, and generate multiple candidate traffic state trajectories for the prediction period based on the prediction task; S2. Obtain context information related to the predicted period, and evaluate the candidate traffic state trajectories based on the configured large language inference model to obtain the optimal trajectory that meets the context information of the predicted period; S3. Based on the optimal trajectory and context information, a text report including traffic situation analysis and actionable traffic management suggestions is generated using a configured large language interpretation model.

2. The adaptive and interpretable traffic situation prediction method based on a large language model according to claim 1, characterized in that: The time series prediction basic model described in step S1 is built based on the Transformer architecture; By segmenting and quantifying numerical traffic time series data, the spatiotemporal dependency of traffic flow is captured, and the congestion probability and vehicle speed in the corresponding time period are predicted; thus, multiple candidate traffic state trajectories are obtained.

3. The adaptive and interpretable traffic situation prediction method based on a large language model according to claim 1, characterized in that: The historical traffic data in step S1 is numerical time series data, including historical traffic flow and speed data of the predicted road section and surrounding areas; {1:T} Indicates that T is the current moment; Based on the K candidate traffic state trajectories generated from the historical traffic data during the prediction period, {X k } {T+1:T+n} Represents; k = 1, ..., K; where k is the number of candidate trajectories; n is the prediction step size.

4. The adaptive and interpretable traffic situation prediction method based on a large language model according to claim 1, characterized in that: The context information related to the prediction period in step S2 includes structured data and unstructured data; The structured data includes weather forecasts and scheduled road construction plans; The unstructured data includes traffic accident reports and large-scale public event notifications in natural language format.

5. The adaptive and interpretable traffic situation prediction method based on a large language model according to claim 4, characterized in that: The reasoning process of the large language reasoning model in step S2 is: Performing semantic understanding on the context information to analyze its causal impact on traffic flow, the causal impact including the location, severity, and duration of the impact; An optimal trajectory is selected from the plurality of candidate traffic state trajectories for the prediction period based on the causal influence.

6. The adaptive and interpretable traffic situation prediction method based on a large language model according to claim 5, characterized in that: The step S2 is expressed by the formula: in, represents the optimal trajectory that meets the context information of the prediction period; F LLM represents the adjustment function of the large language inference model; {X k } {T+1:T+n} represents the set of K candidate traffic state trajectories for the prediction period; s represents the context information related to the prediction period.

7. The adaptive and interpretable traffic situation prediction method based on a large language model according to claim 6, characterized in that: The process of obtaining the optimal trajectory that conforms to the context information of the prediction period described in step S2 is equivalent to solving a posterior probability maximization problem conditional on s; The solution formula is: where P(·|s) represents the posterior plausibility score of each candidate trajectory given the context information s, as evaluated by the large language inference model.

8. The adaptive and interpretable traffic situation prediction method based on a large language model according to claim 1, characterized in that: The actionable traffic management suggestions in step S3 specifically include: Suggestions on adjusting signal light timing for corresponding road sections, traffic diversion or detour strategies, and travel information release recommendations for the public.

9. An adaptive and interpretable traffic situation prediction system based on a large language model, characterized by: Includes the following modules: Trajectory generation module: used to obtain historical traffic data related to the prediction task, input the trained time series prediction basic model, and generate multiple candidate traffic state trajectories for the prediction period based on the prediction task; Trajectory evaluation module: used to obtain context information related to the predicted period and evaluate the candidate traffic state trajectories based on the configured large language inference model to obtain the optimal trajectory that meets the context information of the predicted period; Trajectory traffic interpretation module: used to generate a text report containing traffic situation analysis and actionable traffic management suggestions based on the optimal trajectory and context information using the configured large language interpretation model.

Images