Airport scene dynamic modeling method based on multi-type target behaviors

Abstract

The invention relates to the technical field of dynamic modeling, in particular to an airport scene dynamic modeling method based on multi-type target behaviors. Comprising the following steps: data acquisition and preprocessing, multi-type target identification and classification, trajectory prediction and preliminary modeling, dynamic behavior modeling based on behavior pattern learning, multi-modal information fusion and model optimization, conflict detection and prediction, adaptive behavior adjustment strategy, model verification and simulation, and real-time updating and iterative optimization. Through the machine learning and deep learning technology, the method can automatically learn and adapt to the dynamic change of the airport scene and the new behavior mode, achieves the precise prediction of the target behavior in combination with radar, video and GPS multi-mode information, Kalman filtering, particle filtering and DQN prediction algorithms, not only pays attention to the behavior of a single target, but also pays attention to the behavior of the target. In addition, interaction behaviors between targets are also considered, including complex scenes such as avoiding of airplanes and vehicles and passing of pedestrians through runways, so that more comprehensive scene dynamic analysis is provided.

Description

Technical Field

[0001] The invention relates to the technical field of dynamic modeling, and in particular to an airport scene dynamic modeling method based on multi-type target behaviors. Background Art

[0002] Multi-type target behavior analysis refers to the process of identifying, classifying, predicting and understanding the behaviors of multiple types of targets in a specific environment. In the airport environment, this analysis is particularly important because the airport is a highly dynamic and complex transportation system involving multiple different types of targets operating simultaneously in a limited space. Its multi-type targets mainly include aircraft, vehicles and pedestrians. The behavior patterns of these targets are different, and there may be complex interactive behaviors between them, such as avoidance of aircraft and vehicles, and maintaining a safe distance between pedestrians and aircraft.

[0003] Therefore, in general, traditional airport scene dynamic analysis modeling methods rely on rule-based systems or simple mathematical models. These methods have problems with flexibility, accuracy and scalability. Rule-based systems are often difficult to adapt to the dynamic changes and emerging behavior patterns of airport scenes, while simple mathematical models may not be able to accurately capture the complex interactive behaviors and nonlinear dynamics between targets. Therefore, traditional methods are often unable to cope with multi-type targets and large-scale data, and are difficult to expand to more complex airport environments.

[0004] In summary, it is necessary to propose an airport scene dynamic modeling method based on multi-type target behaviors to solve the above problems. Summary of the invention

[0005] The purpose of the present invention is to provide an airport scene dynamic modeling method based on multi-type target behaviors to solve the problems raised in the above background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] The airport scene dynamic modeling method based on multi-type target behavior includes the following steps:

[0008] S1. Use radar, lidar, and cameras to collect real-time data on the airport scene. Through data cleaning and format unification, the collected raw data is converted into a standard format that can be used for modeling;

[0009] S2. Apply machine learning algorithms to the preprocessed data to identify targets, distinguish different types of targets such as aircraft, vehicles, and pedestrians, and use classification algorithms to finely classify the identified targets, including distinguishing different types of aircraft or vehicles;

[0010] S3. Use Kalman filter or particle filter algorithm to predict the trajectory of the identified target, and preliminarily establish a dynamic model of the airport scene based on the predicted trajectory, including the position, speed and direction information of the target;

[0011] S4. Apply deep reinforcement learning algorithms to learn and simulate the behavior patterns of different types of targets at airports, including aircraft taxi path selection and vehicle avoidance behavior;

[0012] S5. Combine information from multiple modalities such as radar, video, and GPS, and use a multimodal information fusion algorithm to optimize the preliminary model;

[0013] S6. Develop conflict detection algorithms, predict possible future conflicts based on dynamic models, use spatiotemporal prediction models to predict conflicts, and calculate conflict probabilities;

[0014] S7. Adaptive behavior adjustment strategies based on conflict prediction results, including dynamic adjustment of aircraft taxiing speed and vehicle driving routes, and application of heuristic search algorithms or reinforcement learning algorithms to adjust the solutions;

[0015] S8. Use historical data or simulation data to verify the established dynamic model, evaluate the accuracy and reliability of the model, use simulation software to perform scene dynamic simulation, and observe the performance of the model in different scenarios;

[0016] S9. Deploy the model to the actual airport system to collect new data in real time and update the model, and iterate and optimize the model regularly.

[0017] Preferably, the implementation steps of step S1 are:

[0018] S1.1. Use radar sensors to obtain basic information about the target position and speed at the airport scene, use laser radar to obtain the three-dimensional shape and precise position of the target, and use cameras to capture video images of the airport scene for subsequent target recognition;

[0019] S1.2. Remove noise and outliers from radar and Li DAR data, including erroneous data due to weather or equipment failure, and perform denoising and enhancement on video images to improve image quality;

[0020] S1.3. Convert radar, Li DAR, and camera data into a unified spatiotemporal coordinate system, ensure data consistency, and format the data into a format that can be used as input for machine learning models.

[0021] Preferably, the implementation steps of step S2 are:

[0022] S2.1. Apply convolutional neural network (CNN) to detect targets in preprocessed video images, identify aircraft, vehicles, and pedestrians, and use YOLO target detection algorithm to improve recognition speed and accuracy;

[0023] S2.2. Classify the identified targets using a support vector machine (SVM) to distinguish different types of passenger planes, cargo planes or vehicles, and extract the size, shape and texture of the target as the input of the SVM classifier;

[0024] S2.3. Record the position, speed, and direction attribute information of each target to provide basic data for subsequent trajectory prediction and dynamic modeling.

[0025] Preferably, the implementation steps of step S3 are:

[0026] S3.1. Use the Kalman filter algorithm to predict the trajectory of the identified target and estimate the future position and speed of the target;

[0027] For nonlinear or complex scenarios, particle filtering algorithms are used to improve prediction accuracy;

[0028] S3.2. Based on the predicted trajectory, a preliminary dynamic model of the airport scene is established, including the position, speed, and direction status information of the target, and the target trajectories at different time points are associated using the nearest neighbor algorithm to form a complete motion trajectory;

[0029] S3.3. Store the data of the preliminary dynamic model in a database to facilitate subsequent analysis and optimization.

[0030] Preferably, the implementation steps of step S4 are:

[0031] S4.1. Apply the DQN training model to learn and simulate the behavior patterns of different types of targets at the airport, design the reward function and penalty mechanism, and guide the model learning machine taxi path selection and vehicle avoidance behavior strategy;

[0032] S4.2. Use the trained DRL model to simulate the behavior of the target at the airport, generate simulated trajectory and behavior data, and verify the accuracy and reliability of the model through simulated data;

[0033] S4.3. Analyze simulation data, capture and simulate complex interactive behaviors such as avoidance between aircraft and vehicles and pedestrians crossing the runway, and use behavioral analysis methods to further analyze and understand the interactive behaviors.

[0034] Preferably, the implementation steps of step S5 are:

[0035] S5.1. Combine information from multiple modalities such as radar, video, and GPS, use factor graph models to optimize the preliminary model, design fusion strategies, fuse data from different modalities, and improve the accuracy and robustness of the model;

[0036] S5.2. Use the fused data to optimize the preliminary dynamic model, correct the error caused by a single sensor, and use the gradient descent algorithm to adjust the model parameters to improve the prediction accuracy of the model;

[0037] S5.3. The effectiveness of multimodal information fusion is verified by comparing the model prediction results before and after fusion, and the optimized model is evaluated using the mean square error evaluation indicator.

[0038] Preferably, the implementation steps of step S6 are:

[0039] S6.1. Develop conflict detection algorithms based on dynamic models to predict possible future aircraft taxi path intersections and vehicle collisions, and design conflict detection rules, including distance thresholds and speed difference thresholds, to determine whether a conflict has occurred;

[0040] S6.2. Use the ST-GCN spatiotemporal prediction model to predict conflicts, calculate the probability and possible time points of conflicts, analyze the relative positions and speed changes between targets, and predict the development trend of conflicts;

[0041] S6.3. Generate conflict warning information based on the conflict prediction results and provide it to airport managers for decision support and design warning strategies, including lead time and warning level, to ensure that managers can take timely measures to avoid conflicts.

[0042] Preferably, the implementation steps of step S7 are:

[0043] S7.1. Design adaptive behavior adjustment strategies based on conflict prediction results, including dynamically adjusting aircraft taxi speed and vehicle driving routes, and considering multiple solutions, including changing taxi paths, pausing taxiing, accelerating or decelerating;

[0044] S7.2. Use the A* algorithm or reinforcement learning algorithm to search for the optimal adjustment plan in the solution space, design an evaluation function, evaluate each adjustment plan, and select the optimal plan as the execution strategy;

[0045] S7.3. Send the optimal adjustment plan to relevant targets for execution, including sending it to aircraft or vehicle drivers through the communication system, and monitor the execution of the adjustment plan in real time to ensure that the conflict is effectively resolved.

[0046] Preferably, the implementation steps of step S8 are:

[0047] S8.1. Use historical data to verify the established dynamic model, evaluate the accuracy and reliability of the model, compare the model prediction results with the actual data, and calculate the error index;

[0048] S8.2. Use OMNeT++ to perform scene dynamic simulation, simulate airport operations in different scenarios, design simulation experiments, and verify the performance of the model under different conditions, including different weather and different traffic flows;

[0049] S8.3. Analyze the simulation results, evaluate the simulation performance and applicability of the model, and further adjust and optimize the model based on the simulation results.

[0050] Preferably, the implementation steps of step S9 are:

[0051] S9.1. Deploy the optimized model to the actual airport system, realize real-time data collection and model update, design data interface and communication protocol, and ensure seamless connection between the model and the airport system;

[0052] S9.2. Collect new data on the airport surface in real time, update and modify the model, and use data synchronization algorithms to ensure the consistency of the model with the new data;

[0053] S9.3. Regularly iterate and optimize the model to adapt to the dynamic changes of the airport surface and emerging target behavior patterns, analyze the model performance indicators, including prediction accuracy and computational efficiency, and continuously improve and optimize the model;

[0054] S9.4. Establish a model maintenance and upgrade mechanism to ensure the long-term stable operation of the model, regularly perform performance testing and troubleshooting on the model, and promptly handle problems that arise with the model.

[0055] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention can automatically learn and adapt to the dynamic changes and emerging behavior patterns of the airport scene through machine learning and deep learning technology, and combine radar, video, GPS multimodal information and Kalman filtering, particle filtering, and DQN prediction algorithm to achieve accurate prediction of target behavior, not only focusing on the behavior of a single target, but also considering the interactive behavior between targets, including complex scenes such as the avoidance of aircraft and vehicles and pedestrians crossing the runway, thereby providing a more comprehensive scene dynamic analysis. At the same time, the method of the present invention can also process multiple types of targets and large-scale data, and is suitable for airport environments of different sizes and complexities. Through real-time data collection and model update mechanisms, the method can keep pace with the airport scene and provide timely decision support for airport managers. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 The flowchart of the airport scene dynamic modeling method based on multi-type target behaviors of the present invention is shown. DETAILED DESCRIPTION

[0057] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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 creative work are within the scope of protection of the present invention.

[0058] Example

[0059] See also Figure 1 The present invention proposes a method for dynamic modeling of airport scenes based on multi-type target behaviors, which specifically includes the following steps:

[0060] S1. Data collection and preprocessing:

[0061] S1.1. Use radar sensors to obtain basic information about the target position and speed at the airport scene, use laser radar to obtain the three-dimensional shape and precise position of the target, and use cameras to capture video images of the airport scene for subsequent target recognition;

[0062] S1.2. Remove noise and outliers from radar and Li DAR data, including erroneous data due to weather or equipment failure, and perform denoising and enhancement on video images to improve image quality;

[0063] S1.3. Convert radar, Li DAR, and camera data into a unified spatiotemporal coordinate system to ensure data consistency and format the data into a format that can be used as input for machine learning models;

[0064] S2. Multi-type target recognition and classification:

[0065] S2.1. Apply convolutional neural network (CNN) to detect targets in preprocessed video images, identify aircraft, vehicles, and pedestrians, and use YOLO target detection algorithm to improve recognition speed and accuracy;

[0066] S2.2. Classify the identified targets using a support vector machine (SVM) to distinguish different types of passenger planes, cargo planes or vehicles, and extract the size, shape and texture of the target as the input of the SVM classifier;

[0067] S2.3. Record the position, speed, and direction attribute information of each target to provide basic data for subsequent trajectory prediction and dynamic modeling;

[0068] S3. Trajectory prediction and preliminary modeling:

[0069] S3.1. Use the Kalman filter algorithm to predict the trajectory of the identified target and estimate the future position and speed of the target. For nonlinear or complex scenes, use the particle filter algorithm to improve the prediction accuracy;

[0070] S3.2. Based on the predicted trajectory, a preliminary dynamic model of the airport scene is established, including the position, speed, and direction status information of the target, and the target trajectories at different time points are associated using the nearest neighbor algorithm to form a complete motion trajectory;

[0071] S3.3. Store the data of the preliminary dynamic model in a database to facilitate subsequent analysis and optimization;

[0072] S4. Dynamic behavior modeling based on behavior pattern learning

[0073] S4.1. Apply the DQN training model to learn and simulate the behavior patterns of different types of targets at the airport, design the reward function and penalty mechanism, and guide the model learning machine taxi path selection and vehicle avoidance behavior strategy;

[0074] S4.2. Use the trained DRL model to simulate the behavior of the target at the airport, generate simulated trajectory and behavior data, and verify the accuracy and reliability of the model through simulated data;

[0075] S4.3. Analyze simulation data, capture and simulate complex interactive behaviors such as avoidance between aircraft and vehicles and pedestrians crossing the runway, and use behavioral analysis methods to further analyze and understand the interactive behaviors;

[0076] S5. Multimodal information fusion and model optimization

[0077] S5.1. Combine information from multiple modalities such as radar, video, and GPS, use factor graph models to optimize the preliminary model, design fusion strategies, fuse data from different modalities, and improve the accuracy and robustness of the model;

[0078] S5.2. Use the fused data to optimize the preliminary dynamic model, correct the error caused by a single sensor, and use the gradient descent algorithm to adjust the model parameters to improve the prediction accuracy of the model. Specifically, in multimodal information fusion, use the gradient descent algorithm to optimize the model parameters to minimize the error between the fused data and the model prediction;

[0079] Let the model parameter be θ, the fused data be y, and the model prediction be Then the loss function is as follows:

[0080]

[0081] The gradient descent algorithm iteratively updates the parameters θ to minimize the loss function L(θ).

[0082]

[0083] Where:

[0084] θ is the parameter of the model, which in this step serves as the weights and biases in the neural network;

[0085] y is the fused data, which is the information fusion result of multiple modalities such as radar, video, and GPS in this step;

[0086] is the prediction result of the model based on the parameter θ;

[0087] L(θ) is the loss function, which measures the error between the model prediction and the real data;

[0088] α is the learning rate, which controls the step size of parameter update;

[0089] It is the gradient of the loss function with respect to the parameter θ, indicating the direction of parameter update;

[0090] In each iteration, the gradient of the loss function is calculated, and the model parameters are updated according to the gradient until the loss function converges to the minimum value or reaches the preset number of iterations. The model parameters are optimized through the gradient descent algorithm to improve the accuracy and robustness of the model.

[0091] S5.3. Verify the effectiveness of multimodal information fusion by comparing the model prediction results before and after fusion, and use the mean square error evaluation index to evaluate the optimized model;

[0092] S6. Conflict Detection and Prediction

[0093] S6.1. Develop conflict detection algorithms based on dynamic models to predict possible future aircraft taxi path intersections and vehicle collisions, and design conflict detection rules, including distance thresholds and speed difference thresholds, to determine whether a conflict has occurred;

[0094] S6.2. Use the ST-GCN spatiotemporal prediction model to predict conflicts, calculate the probability and possible time points of conflicts, analyze the relative positions and speed changes between targets, and predict the development trend of conflicts;

[0095] S6.3. Generate conflict warning information based on the conflict prediction results and provide it to airport managers for decision support and design of warning strategies, including lead time and warning level, to ensure that managers can take timely measures to avoid conflicts;

[0096] S7. Adaptive behavior adjustment strategy

[0097] S7.1. Design adaptive behavior adjustment strategies based on conflict prediction results, including dynamically adjusting aircraft taxi speed and vehicle driving routes, and considering multiple solutions, including changing taxi paths, pausing taxiing, accelerating or decelerating;

[0098] S7.2. Use the A* algorithm or reinforcement learning algorithm to search for the optimal adjustment plan in the solution space, design an evaluation function, evaluate each adjustment plan, and select the optimal plan as the execution strategy;

[0099] S7.3. Send the optimal adjustment plan to the relevant targets for execution, including sending it to the aircraft or vehicle pilots through the communication system, and monitor the execution of the adjustment plan in real time to ensure that the conflict is effectively resolved;

[0100] S8. Model verification and simulation:

[0101] S8.1. Use historical data to verify the established dynamic model, evaluate the accuracy and reliability of the model, compare the model prediction results with the actual data, and calculate the error index;

[0102] S8.2. Use OMNeT++ to perform scene dynamic simulation, simulate airport operations in different scenarios, design simulation experiments, and verify the performance of the model under different conditions, including different weather and different traffic flows;

[0103] S8.3. Analyze the simulation results, evaluate the simulation performance and applicability of the model, and further adjust and optimize the model based on the simulation results;

[0104] S9. Real-time update and iterative optimization:

[0105] S9.1. Deploy the optimized model to the actual airport system, realize real-time data collection and model update, design data interface and communication protocol, and ensure seamless connection between the model and the airport system;

[0106] S9.2. Collect new data on the airport surface in real time, update and modify the model, and use data synchronization algorithms to ensure the consistency of the model with the new data;

[0107] S9.3. Regularly iterate and optimize the model to adapt to the dynamic changes of the airport surface and emerging target behavior patterns, analyze the model performance indicators, including prediction accuracy and computational efficiency, and continuously improve and optimize the model;

[0108] S9.4. Establish a model maintenance and upgrade mechanism to ensure the long-term stable operation of the model, regularly perform performance testing and troubleshooting on the model, and promptly handle problems that arise with the model.

[0109] Through the above steps, the airport scene dynamic modeling method based on multi-type target behaviors of the present invention provides a more flexible, accurate and comprehensive scene dynamic analysis method than traditional methods by combining advanced machine learning and prediction technologies, which is of great significance for improving the operating efficiency and safety of airports.

[0110] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. The airport scene dynamic modeling method based on multi-type target behavior is characterized by: The following steps are involved: S1. Use radar, lidar, and cameras to collect real-time data on the airport scene. Through data cleaning and format unification, the collected raw data is converted into a standard format that can be used for modeling; S2. Apply machine learning algorithms to the preprocessed data to identify targets, distinguish different types of targets such as aircraft, vehicles, and pedestrians, and use classification algorithms to finely classify the identified targets, including distinguishing different types of aircraft or vehicles; S3. Use Kalman filter or particle filter algorithm to predict the trajectory of the identified target, and preliminarily establish a dynamic model of the airport scene based on the predicted trajectory, including the position, speed and direction information of the target; S4. Apply deep reinforcement learning algorithms to learn and simulate the behavior patterns of different types of targets at airports, including aircraft taxi path selection and vehicle avoidance behavior; S5. Combine information from multiple modalities such as radar, video, and GPS, and use a multimodal information fusion algorithm to optimize the preliminary model; S6. Develop conflict detection algorithms, predict possible future conflicts based on dynamic models, use spatiotemporal prediction models to predict conflicts, and calculate conflict probabilities; S7. Adapt behavior adjustment strategies based on conflict prediction results, including dynamically adjusting aircraft taxiing speed and vehicle driving routes, and applying heuristic search algorithms or reinforcement learning algorithms to adjust the solutions; S8. Use historical data or simulation data to verify the established dynamic model, evaluate the accuracy and reliability of the model, use simulation software to perform scene dynamic simulation, and observe the performance of the model in different scenarios; S9. Deploy the model to the actual airport system to collect new data in real time and update the model, and iterate and optimize the model regularly.

2. The airport scene dynamic modeling method based on multi-type target behavior according to claim 1 is characterized in that: The implementation steps of step S1 are: S1.

1. Use radar sensors to obtain basic information about the target position and speed at the airport scene, use laser radar to obtain the three-dimensional shape and precise position of the target, and use cameras to capture video images of the airport scene for subsequent target recognition; S1.

2. Remove noise and outliers from radar and Li DAR data, including erroneous data due to weather or equipment failure, and denoise and enhance video images; S1.

3. Convert the data from radar, Li DAR, and camera into a unified spatiotemporal coordinate system to ensure data consistency.

3. The airport scene dynamic modeling method based on multi-type target behavior according to claim 2 is characterized in that: The implementation steps of step S2 are: S2.

1. Apply convolutional neural network to detect targets on preprocessed video images and identify targets such as aircraft, vehicles, and pedestrians; S2.

2. Classify the identified targets using SVM to distinguish different types of passenger planes, cargo planes or vehicles, and extract the size, shape and texture of the targets as input to the SVM classifier; S2.

3. Record the position, speed, and direction attribute information of each target.

4. The airport scene dynamic modeling method based on multi-type target behavior according to claim 3 is characterized in that: The implementation steps of step S3 are: S3.

1. Use the Kalman filter algorithm to predict the trajectory of the identified target and estimate the future position and speed of the target. For nonlinear or complex scenes, use the particle filter algorithm to improve the prediction accuracy; S3.

2. Based on the predicted trajectory, a preliminary dynamic model of the airport scene is established, including the position, speed, and direction status information of the target, and the target trajectory at different time points is associated using the nearest neighbor algorithm. S3.

3. Store the data of the preliminary dynamic model in a database.

5. The airport scene dynamic modeling method based on multi-type target behavior according to claim 4 is characterized in that: The implementation steps of step S4 are: S4.

1. Apply the DQN training model to learn and simulate the behavior patterns of different types of targets at the airport, design the reward function and penalty mechanism, and guide the model learning machine taxi path selection and vehicle avoidance behavior strategy; S4.

2. Use the trained DRL model to simulate the behavior of the target at the airport and generate simulated trajectory and behavior data; S4.

3. Analyze simulation data, capture and simulate complex interactive behaviors such as avoidance between aircraft and vehicles and pedestrians crossing the runway, and use behavioral analysis methods to further analyze and understand the interactive behaviors.

6. The airport scene dynamic modeling method based on multi-type target behavior according to claim 5 is characterized in that: The implementation steps of step S5 are: S5.

1. Combine information from multiple modalities such as radar, video, and GPS, use factor graph models to optimize the preliminary model, design fusion strategies, and fuse data from different modalities; S5.

2. Use the fused data to optimize the preliminary dynamic model, correct the error caused by a single sensor, and adjust the model parameters using the gradient descent algorithm; S5.

3. The effectiveness of multimodal information fusion is verified by comparing the model prediction results before and after fusion, and the optimized model is evaluated using the mean square error evaluation indicator.

7. The airport scene dynamic modeling method based on multi-type target behavior according to claim 6 is characterized in that: The implementation steps of step S6 are: S6.

1. Develop conflict detection algorithms based on dynamic models to predict possible future aircraft taxi path intersections and vehicle collisions, and design conflict detection rules, including distance thresholds and speed difference thresholds, to determine whether a conflict has occurred; S6.

2. Use the ST-GCN spatiotemporal prediction model to predict conflicts, calculate the probability and possible time points of conflicts, analyze the relative positions and speed changes between targets, and predict the development trend of conflicts; S6.

3. Generate conflict warning information based on the conflict prediction results and provide it to airport managers for decision support and design of warning strategies, including lead time and warning level.

8. The airport scene dynamic modeling method based on multi-type target behavior according to claim 7 is characterized in that: The implementation steps of step S7 are: S7.

1. Design adaptive behavior adjustment strategies based on conflict prediction results, including dynamically adjusting aircraft taxi speed and vehicle driving routes, and considering multiple solutions, including changing taxi paths, pausing taxiing, accelerating or decelerating; S7.

2. Use the A* algorithm or reinforcement learning algorithm to search for the optimal adjustment solution in the solution space, design an evaluation function, and evaluate each adjustment solution; S7.

3. Send the optimal adjustment plan to relevant targets for execution, including sending it to aircraft or vehicle drivers through the communication system, and monitor the execution of the adjustment plan in real time.

9. The airport scene dynamic modeling method based on multi-type target behavior according to claim 8 is characterized in that: The implementation steps of step S8 are: S8.

1. Use historical data to verify the established dynamic model, evaluate the accuracy and reliability of the model, compare the model prediction results with the actual data, and calculate the error index; S8.

2. Use OMNeT++ to perform scene dynamic simulation, simulate airport operations in different scenarios, design simulation experiments, and verify the performance of the model under different conditions, including different weather and different traffic flows; S8.

3. Analyze the simulation results, evaluate the simulation performance and applicability of the model, and further adjust and optimize the model based on the simulation results.

10. The airport scene dynamic modeling method based on multi-type target behavior according to claim 9 is characterized in that: The implementation steps of step S9 are: S9.

1. Deploy the optimized model to the actual airport system to achieve real-time data collection and model update; S9.

2. Collect new data on the airport surface in real time, update and modify the model, and use data synchronization algorithms to ensure the consistency of the model with the new data; S9.

3. Regularly iterate and optimize the model to adapt to the dynamic changes of the airport surface and emerging target behavior patterns, analyze the model performance indicators, including prediction accuracy and computational efficiency, and continuously improve and optimize the model; S9.

4. Establish a model maintenance and upgrade mechanism to ensure the long-term stable operation of the model, and regularly perform performance testing and troubleshooting on the model.

Images