Calculation method of civil aviation flight arrival data based on ADS-B (Automatic Dependent Surveillance-Broadcast)

Abstract

The invention relates to an ADS-B (Automatic Dependent Surveillance-Broadcast)-based civil aviation flight pre-arrival data calculation method, which integrates ADS-B data, meteorological satellite data, airport runway state monitoring data and air traffic control real-time instruction data through multi-source data fusion, and dynamically adjusts the weight of each data source according to the flight stage of a flight. A dynamic adjustment time synchronization interval mechanism is adopted, a short synchronization interval is set in the take-off stage and the landing stage so as to capture rapidly-changed data, and the synchronization interval is prolonged in the cruise stage so as to optimize computing resources. Meanwhile, the deep learning model based on the Transform architecture utilizes a multi-head self-attention mechanism and position coding to enhance the feature capture capability, and reduces the calculation burden through a sparse activation strategy. The system is provided with a real-time monitoring layer, log records and visual analysis tools are integrated, and interpretability and traceability of the model are ensured. An efficient aviation management decision is supported, and the efficiency and safety of overall aviation traffic management are improved.

Description

Technical Field

[0001] The invention belongs to the field of aerospace technology, and in particular relates to a method for calculating civil aviation flight arrival data based on ADS-B. Background Art

[0002] With the rapid development of the civil aviation industry, the number of flights is increasing, and the complexity and challenges of air traffic management are also increasing. Accurately predicting the estimated time of arrival (ETA) of flights is of great significance for optimizing flight scheduling, improving airport operation efficiency, reducing flight delays and improving passenger satisfaction. Existing technologies mainly use methods based on historical data and real-time monitoring data to predict flight ETA, but there are still deficiencies in data fusion, real-time performance and prediction accuracy.

[0003] Chinese invention patent CN112862171B discloses a flight arrival time prediction method based on spatiotemporal neural network, which belongs to the field of civil aviation technology. The method first collects historical tracks as samples for historical flights within a certain period of time, and obtains a historical track set after processing; the historical tracks are clustered into clusters, each cluster represents a track mode, and the representative track of each track mode is selected. Then, given a target flight, the track that has flown in the terminal area is resampled to obtain the existing track of the target flight, and the predicted track of the target flight is calculated through the track mode. An arrival time prediction model is constructed, and the arrival time prediction model is trained using a cross-validation method. Finally, the predicted track of the target flight is input into the trained arrival time prediction model, and the estimated arrival time of the target flight is output. The invention improves the accuracy of the prediction result by considering the comprehensiveness of the factors affecting the flight arrival time and introducing the spatiotemporal attention mechanism. However, this method mainly relies on the matching of historical track patterns, has limited response capabilities to real-time dynamic changes, and has room for improvement in multi-source data fusion and real-time performance.

[0004] Chinese invention patent CN117272236B discloses a multi-source estimated arrival time fusion method and system based on flight schedule. The system includes: a raw data receiving and parsing module for receiving and parsing raw data; a message queue processing module for playing the role of message queue processing and receiving data by establishing an activemq queue; a multi-data source comprehensive screening module for comprehensive screening of multiple data sources; a business system server for processing the business and verifying the signed information. The multi-source estimated arrival time fusion system based on flight schedule provided by the invention has higher stability, reliability, accuracy and real-time performance, and has good scalability. Through the error correction mechanism and the estimated arrival time of comprehensive multiple data sources, the system meets the estimated arrival time analysis and fusion requirements of stability and data reliability. However, the system fails to fully consider the dynamic adjustment of data source weights, deep learning model optimization, and anomaly monitoring and traceability, which limits its application effect and system reliability in complex aviation traffic environments.

[0005] The above two existing technologies have their own advantages in flight ETA prediction, but still have the following shortcomings: CN112862171B mainly relies on historical track patterns for prediction, lacks dynamic fusion and weight adjustment of multi-source real-time data, resulting in insufficient prediction accuracy and response speed when facing real-time flight status changes. Although CN117272236B achieves the fusion of multi-source data, it lacks advanced algorithm support in the dynamic adjustment and optimization of data source weights, making it difficult to achieve optimal weight distribution.

[0006] Both existing technologies do not adopt advanced deep learning architectures (such as Transformer) for feature capture and prediction optimization, which limits the model's ability to understand and process complex data relationships.

[0007] Both CN112862171B and CN117272236B have not established a complete abnormality monitoring and traceability mechanism, which makes it difficult to discover and locate problems in a timely manner when the model has abnormalities or large prediction errors, affecting the reliability of the system and user trust.

[0008] Based on the above shortcomings, the present invention proposes a method for calculating the expected arrival data of civil aviation flights based on ADS-B, aiming to Summary of the invention

[0009] In view of the above-mentioned prior art, the technical problem to be solved by the present invention is how to solve the deficiencies in the prior art and improve the accuracy, real-time nature and system reliability of flight ETA prediction through multi-source data fusion and dynamic weight adjustment, Transformer-based deep learning optimization and a complete anomaly monitoring and traceability mechanism.

[0010] Dynamically integrate multi-source data such as ADS-B, meteorological satellites, runway status monitoring and air traffic control at different flight stages, and dynamically adjust the weight of each data source according to the real-time flight status to improve the accuracy and flexibility of flight ETA prediction. How to use advanced deep learning architectures (such as Transformer) through multi-head self-attention mechanism and position encoding to enhance the model's understanding and processing capabilities of complex data relationships and time series characteristics, and further improve prediction accuracy.

[0011] Establish a complete anomaly monitoring and traceability mechanism to record and monitor the operating status of the model in real time, promptly detect and respond to abnormal situations, and ensure system reliability and user trust.

[0012] Through sparse activation strategy and intelligent adjustment of time synchronization interval mechanism, efficient utilization of computing resources is achieved, system operating costs are reduced, and high-frequency, high-precision data processing and prediction are provided during critical flight phases.

[0013] To solve the above problems, the present invention provides a method for calculating civil aviation flight arrival data based on ADS-B, comprising the following steps: S1. Multi-source data fusion steps: Integrate multi-source data from ADS-B data, meteorological satellite data, airport runway status monitoring data and air traffic control real-time instruction data; Dynamically adjust the weights of the above data sources according to the flight phase of the flight; S2. Steps to dynamically adjust the time synchronization interval: Adjust the time synchronization interval according to the flight phase (take-off, cruise, landing); The time synchronization interval is set to A seconds during takeoff and landing, and the time synchronization interval is set to B seconds during cruising; Intelligently adjust the time synchronization interval length according to the changes in real-time flight status; S3. Deep learning optimization steps: Construct an input sequence, including fused multi-source data, dynamically adjusted data source weights, and dynamically adjusted time synchronization intervals; A model based on the Transformer architecture is used to process input sequences through a self-attention mechanism to capture long-distance dependencies. Use feedforward neural network to perform nonlinear transformation on features; Applying positional encoding adds timing information to each element in the input data sequence; A sparse activation strategy is used to activate only some neurons for calculation, and all neurons are activated when an anomaly is detected; S4. Abnormal monitoring and traceability steps: Establish a real-time monitoring layer to record the model’s input, processing, and output results; Reduce the black box problem of models through visual monitoring and logging; When an exception occurs, the computing process can be traced back to help identify the source of the problem; S5. Flight arrival time prediction steps: Based on the optimized data obtained in the above steps S1 to S4, the expected arrival time of the civil aviation flight is predicted.

[0014] In the S1 multi-source data fusion step, ADS-B data includes the flight's real-time position, speed, altitude and heading information. The flight's real-time position uses high-precision positioning technology to improve the data's spatial resolution and accuracy.

[0015] In the S1 multi-source data fusion step, meteorological satellite data includes wind speed, wind direction, airflow and temperature information. Meteorological data is obtained through a real-time data stream interface, and a data interpolation algorithm is used to complete missing data to ensure the continuity and integrity of meteorological information.

[0016] In the S1 multi-source data fusion step, the airport runway status monitoring data includes the runway usage status, slippery degree and snow accumulation conditions. The runway status data is collected in real time through the ground sensor network, and the runway slippery degree is predicted and evaluated using a machine learning algorithm.

[0017] In the S1 multi-source data fusion step, the real-time air traffic control command data includes the relative position of the flight to other flights, landing queue and runway change information. The real-time command data is obtained through the air traffic management system interface and the commands are processed using a priority sorting algorithm to optimize the flight landing order.

[0018] The dynamic data source weight adjustment in the S1 multi-source data fusion step adopts a weighted average or weighted aggregation algorithm, where the weight changes dynamically according to the flight phase and the real-time reliability of the data source. The weight adjustment is based on the Bayesian optimization algorithm to achieve optimal weight distribution.

[0019] In the S2 step of dynamically adjusting the time synchronization interval, the intelligent adjustment mechanism is based on real-time flight data and flight status. When airflow changes or sudden control instructions are detected, the time synchronization interval is automatically shortened, and the sliding window technology is used to dynamically manage the time synchronization interval to improve the real-time and accuracy of the prediction.

[0020] In the S3 deep learning optimization step, the Transformer model includes a multi-layer self-attention mechanism and a fully connected feedforward neural network. The self-attention mechanism uses multi-head attention to enhance the model's ability to capture different features, and the position encoding is generated using sine and cosine functions to provide rich timing information.

[0021] In the S3 deep learning optimization step, the sparse activation strategy selectively activates neurons through a dynamic gating mechanism, activating only a specific proportion of neurons under normal conditions to reduce the amount of computation. When an abnormality is detected, the activation of all neurons is triggered through a global signal to ensure the model's comprehensive responsiveness in emergency situations.

[0022] In the S4 anomaly monitoring and traceability step, the real-time monitoring layer records and displays the model's input data, weight adjustment, time synchronization process and output results at each time step in real time by integrating logging modules and visual analysis tools. The logging uses distributed storage technology to ensure high availability and integrity of the data, and the visual analysis tool supports dynamic query and backtracking of multi-dimensional data to improve the interpretability of the model and the decision support capabilities of operators.

[0023] In summary, this application has the following beneficial effects: 1. By integrating ADS-B data, meteorological satellite data, runway status monitoring data and air traffic control instruction data, comprehensively considering various key factors affecting the flight arrival time, significantly improving the accuracy of the expected arrival time prediction, dynamically adjusting the weight of each data source according to different flight stages, ensuring that the most relevant data source at each stage maximizes its contribution to the prediction model, and further optimizing the prediction results.

[0024] 2. Based on the intelligent adjustment mechanism of real-time flight data and flight status, the system can flexibly respond to emergencies during the flight (such as airflow changes, sudden control instructions), improve the system's adaptability to dynamic changes, and use sliding window technology to dynamically manage the time synchronization interval to ensure the continuity and real-time performance of data processing under different flight conditions, thereby improving the overall flexibility of the system.

[0025] 3. In the cruise phase when the flight status is stable, the time synchronization interval is extended to reduce the calculation frequency and reduce the consumption of system resources; in the critical phases such as takeoff and landing, the time synchronization interval is shortened to achieve high-frequency data processing and prediction. Through the dynamic gating mechanism, only some neurons are activated under normal conditions to reduce the amount of calculation and energy consumption; all neurons are activated under abnormal conditions to ensure efficient use of computing resources while maintaining prediction accuracy.

[0026] 4. Use high-precision positioning technology to obtain ADS-B data, and use data interpolation algorithms to supplement meteorological data to ensure the data quality and continuity of each data source, provide reliable data support for the prediction model, and use ground sensor networks and machine learning algorithms to monitor and predict the runway slipperiness in real time to ensure the accuracy and timeliness of runway status data.

[0027] 5. Establish a real-time monitoring layer, integrate logging modules and visual analysis tools, comprehensively record the model's input, processing process and output results, improve the transparency of the system, reduce black box problems, and help operators deeply analyze model behavior and performance through dynamic query and backtracking of multi-dimensional data, quickly locate and solve abnormal problems, and ensure the reliability and controllability of system operation.

[0028] 6. The multi-head self-attention mechanism is adopted to enhance the model's ability to capture the dependencies between different features and different positions, and to improve the model's ability to understand and process complex flight data. The position encoding generated by sine and cosine functions provides rich timing information to ensure that the model can accurately understand the time sequence and dependencies of the data, and further improve the prediction accuracy.

[0029] 7. Through the real-time anomaly detection and response mechanism, the system can quickly identify and respond to abnormal situations during operation, ensure the accuracy of the prediction results and the stable operation of the system. When an abnormality occurs, the system can trace back the specific calculation process, identify the source of the problem, quickly locate and solve the problem, and improve the reliability and stability of the overall system.

[0030] 8. Provide high-precision and real-time estimated arrival time predictions, provide reliable data support for aviation management departments to optimize flight scheduling and resource allocation, reduce flight delays and resource waste, optimize flight landing order through priority sorting algorithms, improve airport operation efficiency, reduce flight waiting time on the ground and in the air, and improve the efficiency and safety of overall air traffic management.

[0031] 9. Through detailed logging and visual monitoring, the black box problem of the model is reduced, the transparency of the system is improved, and the user's trust in the prediction results is enhanced. The anomaly detection and calculation process backtracking mechanism ensures that the system can still maintain efficient operation and accurate prediction in complex flight environments and emergencies, thereby improving the system's availability and user satisfaction.

[0032] 10. Reduce system operating costs and improve overall operational efficiency by rationally allocating computing resources and optimizing data processing processes. Accurate arrival time prediction and optimized flight scheduling mechanism can reduce flight delays and resource waste, and improve the operating efficiency and service quality of airlines. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 The workflow for this application Figure 1 ; Figure 2 The workflow for this application Figure 2 ; Figure 3 This is a weight distribution formula diagram for this application; Figure 4The Bayesian optimization objective function formula diagram of this application; Figure 5 This is the formula diagram of the self-attention mechanism for this application; Figure 6 This is the multi-head attention formula diagram for this application; Figure 7 This is the location code map of this application; Figure 8 A feedforward neural network formula diagram for this application; Fig. 9 This is the sparse activation strategy formula diagram for this application; Fig.10 This is the loss function formula diagram for this application; Fig.11 This is the Adam optimization algorithm update formula diagram for this application; Fig.12 This is a diagram of the distributed storage technology formula for this application; Fig.13 This is the multi-dimensional data dynamic query and backtracking formula diagram for this application; Fig.14 Generate formula graphs for detailed logs and reports for this application; Fig.15 This is the anomaly detection formula diagram for this application; Fig.16 This is the anomaly detection formula diagram for this application; Fig.17 This is a diagram of the distributed storage technology formula for this application; Fig.18 This is the multi-dimensional data dynamic query and backtracking formula diagram for this application; Fig.19 Generate formula graphs for detailed logs and reports for this application; Fig. 20 This is the anomaly detection formula diagram for this application; Fig.21 This is a formula diagram for the backtracking steps of this application. DETAILED DESCRIPTION

[0034] In order to make the objectives, technical solutions and beneficial effects of the present invention more clear, the preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings to facilitate understanding by technicians.

[0035] Embodiment 1: like Figure 1 to Figure 2 As shown, this embodiment describes in detail a method for calculating civil aviation flight arrival time data based on ADS-B, with a particular focus on the fusion of multi-source data and the dynamic weight adjustment mechanism to improve the accuracy of flight arrival time prediction and the adaptability of the system. The specific steps are as follows: Step S1: Multi-source data fusion step Data Integration This step involves integrating data from multiple data sources to build comprehensive flight status information. Specifically, it includes: ADS-B data: collects the real-time position, speed, altitude and heading information of the flight. The real-time position of the flight is obtained through high-precision positioning technology to ensure the spatial resolution and accuracy of the data. ADS-B data is transmitted to the data fusion module in real time through a dedicated communication link.

[0036] Meteorological satellite data: Obtain meteorological information including wind speed, wind direction, airflow and temperature. Meteorological data is obtained through a real-time data stream interface, and a data interpolation algorithm is used to complete possible missing data to ensure the continuity and integrity of meteorological information. Data interpolation algorithms such as linear interpolation and spline interpolation can be selected according to actual conditions.

[0037] Airport runway status monitoring data: including runway usage status, slippery degree, and snow accumulation. Runway status data is collected in real time through a ground sensor network, and machine learning algorithms are used to predict and evaluate the runway slippery degree. The sensor network includes humidity sensors, temperature sensors, and snow detection sensors to collect real-time runway status information.

[0038] Airport runway condition monitoring data Airport runway status monitoring data includes runway usage status, slippery degree, and snow accumulation. Runway status data is collected in real time through a ground sensor network. The sensor network includes humidity sensors, temperature sensors, and snow detection sensors to collect real-time runway status information. In order to accurately predict and evaluate the slippery degree of the runway, the Random Forest machine learning algorithm is used. The Random Forest algorithm can effectively process multi-dimensional sensor data and capture complex nonlinear relationships by constructing multiple decision trees and combining the prediction results of each tree, thereby providing high-precision slippery degree prediction. In addition, the Random Forest has good anti-overfitting capabilities and high computational efficiency, making it suitable for real-time application in runway status monitoring systems.

[0039] A brief description of the implementation steps of the random forest algorithm: Data preprocessing: Collect and organize real-time data from humidity sensors, temperature sensors, and snow detection sensors.

[0040] Perform data cleaning, handle missing values ​​and outliers, and ensure data quality.

[0041] The data is standardized to improve the training effect of the algorithm.

[0042] Feature selection and construction: Select key features, such as humidity, temperature, snow depth, etc., as input variables for the model.

[0043] Construct feature vectors and combine them with time series information to enhance the model's ability to capture runway state changes.

[0044] Model training: A random forest model was trained using historical runway wetness data.

[0045] Through cross-validation, model parameters such as the number and depth of trees are optimized to improve the generalization ability of the model.

[0046] Real-time prediction and evaluation: The real-time collected data is input into the trained random forest model to predict the slipperiness of the runway.

[0047] Based on the prediction results, the slippery conditions of the runway are evaluated and timely feedback is provided to the aviation management system to support flight scheduling and ground operation decisions.

[0048] Air traffic control real-time instruction data: including the relative position of the flight to other flights, landing queue and runway change information. Real-time instruction data is obtained through the air traffic management system interface and processed using a priority sorting algorithm to optimize the landing order of flights. The priority sorting algorithm dynamically adjusts the landing order based on the importance, urgency and current queue status of the flight.

[0049] The specific sorting process and basis are as follows: The specific process of sorting Data collection and preprocessing: Data acquisition: Obtain the relative position of each flight, landing queue information and runway change instructions in real time from the air traffic management system interface.

[0050] Data cleaning: Clean the acquired data to remove noise and outliers to ensure the accuracy and completeness of the data.

[0051] Feature extraction: Extract key features of each flight, including flight importance, urgency, and current queue position.

[0052] Priority score calculation: Weight allocation: The importance, urgency and queue position of the flight are scored according to the preset weight coefficients.

[0053] Comprehensive score: The scores of each item are weighted and summed up to obtain the comprehensive priority score of each flight.

[0054] Flight sorting: Sort in descending order: Sort all flights in descending order based on their comprehensive priority scores, with flights with higher priority scores at the top.

[0055] Dynamic adjustment: When the flight status changes (such as an emergency or runway change instruction), the priority scores of the relevant flights are recalculated and the sorting results are updated.

[0056] Optimize the landing order: Determine landing order: Determine the landing order of each flight according to the sorting results to ensure that high-priority flights land first.

[0057] Instruction release: The optimized landing sequence is fed back to the air traffic management system to guide flights to perform corresponding landing operations.

[0058] Sort by Importance of flights: Carrier Priority: Airlines or carriers are scored based on their importance, with flights from key carriers being given higher priority.

[0059] Flight types: such as emergency medical flights, government official flights, etc. have higher importance scores.

[0060] Urgency: Emergency: Whether there is an emergency situation on the flight (such as mechanical failure, passenger health problems, etc.), emergency flights have a higher priority.

[0061] Flight status: Factors such as whether the flight is close to landing and whether an emergency landing is required affect the urgency score.

[0062] Current queue status: Queue position: The position of a flight in the landing queue, with flights at the front having higher priority.

[0063] Estimated landing time: The estimated landing time predicted based on the flight's current position and flight speed. Flights with earlier estimated landing times have higher priority.

[0064] Dynamic weight adjustment Based on data integration, the weight of each data source is dynamically adjusted according to the different flight phases (takeoff, cruising, landing) of the flight to improve the flexibility and accuracy of the prediction model. The specific methods include: Weighted average or weighted aggregation algorithm: Dynamically assign weights based on the flight phase and the real-time reliability of each data source. Weight adjustment is based on the Bayesian optimization algorithm, which continuously optimizes weight allocation through historical data and real-time data to achieve optimal weight allocation. For example, during the takeoff phase, the weight of airport runway status monitoring data is relatively large; during the cruise phase, the weight of meteorological satellite data increases; during the landing phase, the weight of air traffic control instruction data is the largest.

[0065] Weight distribution formula: Figure 3 Shown Bayesian optimization algorithm: The Bayesian optimization algorithm is used to dynamically adjust weights according to the flight phase and the reliability of the data source. Bayesian optimization establishes a probability model to evaluate the effects of different weight combinations and select the optimal weight allocation scheme. This method can efficiently find the optimal solution within a limited number of evaluations and is suitable for the needs of real-time dynamic adjustment.

[0066] Bayesian optimization objective function formula: Figure 4 Shown The specific implementation process of the Bayesian optimization algorithm in weight adjustment is described in detail below.

[0067] Problem Definition The goal is to assign optimal weights W = {W1, W2, W3, W4} to different data sources (ADS-B data, meteorological satellite data, runway status monitoring data, air traffic control instruction data) to minimize the mean square error (MSE) of flight arrival time prediction. i The following constraints must be met: Objective function definition The core of Bayesian optimization is to optimize a black box objective function. In this invention, the objective function is defined as the negative mean square error (Negative Mean Squared Error) of the prediction model: Bayesian Optimization Process Bayesian optimization finds the optimal weight combination W through an iterative process ∗ , the specific steps are as follows: Step 1: Initialization Initial sample point selection: Randomly select several groups of weights W that meet the constraints as the initial sample points.

[0068] Evaluate the initial sample points: For each set of initial weights W, run the prediction model and calculate the corresponding f(W).

[0069] Step 2: Create a proxy model Gaussian Process Regression (GPR): Use Gaussian process regression as a surrogate model to fit the estimated weight combinations The relationship between W and the corresponding objective function value f(W).

[0070] The Gaussian process provides the mean estimate and uncertainty estimate of the objective function, providing a basis for the next step of acquisition function.

[0071] Step 3: Select the next evaluation point Acquisition Function: Commonly used acquisition functions include Expected Improvement (EI), Probability of Improvement (PI), and Knowledge Gradient (KG).

[0072] Taking the expected improvement as an example, select the next weight combination W next , to maximize the expected improvement on the current optimal objective function value: Step 4: Evaluate new sample points Run the prediction model: Use the newly selected weight combination W next Run the prediction model and calculate f(W next ).

[0073] Step 5: Update the proxy model Data update: Combine the new weights W next And its objective function value f(W next ) is added to the evaluated sample set.

[0074] Retrain the surrogate model: Retrain the Gaussian process regression model using the updated sample data and update the estimate of the objective function.

[0075] Step 6: Iteration Repeat steps 3 to 5: Continue to iteratively select new evaluation points, evaluate the objective function value, and update the agent model until the stopping condition is met (such as reaching a predetermined number of iterations or the objective function converges).

[0076] Step 7: Get the optimal weights Determine the optimal weight combination W ∗ : After the iteration, select the weight combination W with the highest objective function value (i.e., the lowest MSE) * Implementation details Data preprocessing Normalize weights: Ensure that weights are normalized in each iteration W satisfies the constraints and Feature Engineering: Construct a weighted feature vector based on the flight phase (takeoff, cruise, landing) and the real-time reliability of the data source.

[0077] Proxy model parameter settings Gaussian process kernel function: Choose an appropriate kernel function (such as RBF kernel) to capture the nonlinear relationship between weights and objective function.

[0078] Hyperparameter optimization: By maximizing the marginal likelihood, the hyperparameters of the Gaussian process are optimized to improve the fitting ability of the surrogate model.

[0079] Collection function optimization Optimize the acquisition function: In each iteration, the next weight combination W is selected by optimizing the acquisition function (such as EI) next , ensuring efficient exploration of the weight space within a limited number of evaluations.

[0080] Real-time dynamic adjustment Online Learning: Implement an online version of Bayesian optimization, supporting real-time adjustment of weights during flight to respond to dynamic changes in flight status.

[0081] Computational efficiency: Optimize algorithm implementation to ensure sufficient computing speed in real-time systems to meet the real-time requirements of aviation management.

[0082] Steps S2 to S5: Since this embodiment mainly focuses on the multi-source data fusion and dynamic weight adjustment of step S1, the following steps S2 to S5 briefly describe their application in this embodiment: Step S2: Dynamically adjust the time synchronization interval Adjust the time synchronization interval according to the flight phase.

[0083] The time synchronization interval is set to A seconds (eg, 5 seconds) during the takeoff and landing phases, and the time synchronization interval is set to B seconds (eg, 30 seconds) during the cruise phase.

[0084] In the step S2 of dynamically adjusting the time synchronization interval, the intelligent adjustment mechanism is dynamically managed based on real-time flight data and flight status. When the system detects airflow changes or sudden control instructions, the time synchronization interval will be automatically shortened to improve the real-time performance and accuracy of the prediction. In order to effectively manage the length of the time synchronization interval, the present invention adopts the sliding window technology, and its specific implementation process is as follows: Window initialization: Set the initial window size: Set the initial time synchronization interval length according to the flight phase (takeoff, cruising, landing) of the flight. For example, set a shorter time synchronization interval (such as 5 seconds) for the takeoff and landing phases, and a longer time synchronization interval (such as 30 seconds) for the cruising phase.

[0085] Real-time data monitoring and event detection: Continuous monitoring of flight status: The system continuously monitors data from ADS-B, weather satellites and other sensors, and analyzes flight parameters in real time.

[0086] Event triggering: Once airflow changes (such as sudden turbulence) are detected or sudden air traffic control instructions (such as runway changes, landing queue adjustments) are received, the system immediately triggers the adjustment of the time synchronization interval.

[0087] Dynamic Window Adjustment: Shorten the time synchronization interval: After detecting the above events, the system automatically shortens the current time synchronization interval from a longer B second (such as 30 seconds) to a shorter A second (such as 10 seconds) to respond more quickly to rapid changes in flight status.

[0088] Sliding window update: Using sliding window technology, the system adds new data points to the window at each time step and removes the oldest data points to ensure that the window always contains the latest flight status information. In this way, the window can dynamically adapt to changes in flight status and reflect the current flight environment in real time.

[0089] Window recovery and optimization: Recovery after the event: When the airflow changes or the sudden command is processed and the flight status returns to stability, the system will gradually restore the time synchronization interval to a longer B seconds to optimize the use of computing resources.

[0090] Sliding window smooth transition: Through the smooth transition mechanism, frequent switching of window size can be avoided to prevent interference with the prediction model, ensuring data continuity and prediction stability.

[0091] Improvements in real-time performance and accuracy: Quick response: Sliding window technology enables the system to capture the latest flight status changes in a short time, quickly adjust the input data range of the prediction model, and improve the real-time performance of the prediction.

[0092] Data continuity: The sliding window ensures the continuity of data and the integrity of the time series, enabling the forecasting model to make accurate predictions based on the latest and continuous data.

[0093] Intelligently adjust the time synchronization interval length according to changes in real-time flight status.

[0094] In step S2 of the present invention, the dynamic adjustment of the time synchronization interval is specifically based on the following real-time flight status factors: Airflow changes: Sudden changes in wind speed and direction: By real-time monitoring of meteorological satellite data, when a sharp change in wind speed or direction is detected, it indicates that there may be airflow disturbance or turbulence, and the time synchronization interval needs to be shortened for rapid response.

[0095] Air Traffic Control Instructions: Emergency instructions: The issuance of emergency air traffic control instructions such as runway changes and landing queue instructions requires flights to quickly adjust their flight paths or landing sequences, and the time synchronization interval needs to be shortened to update the forecast in real time.

[0096] Rapid changes in flight status: Rapid changes in speed and altitude: Flights typically experience large speed and altitude adjustments during takeoff and landing. When rapid changes in these parameters are detected, the time synchronization interval needs to be shortened to accurately capture the dynamic changes in flight status.

[0097] Emergency: Emergency situations: Emergency situations such as mechanical failures and passenger health issues require the flight to respond quickly and shorten the time synchronization interval to ensure that the prediction model can update the estimated arrival time in a timely manner.

[0098] By monitoring and analyzing the above-mentioned real-time flight status factors, the system can intelligently adjust the length of the time synchronization interval to ensure high-precision estimated arrival time prediction during critical or abnormal flight phases, while optimizing the use of computing resources when the flight status is stable.

[0099] Step S3: Deep learning optimization step Construct an input sequence, including fused multi-source data, dynamically adjusted data source weights, and dynamically adjusted time synchronization intervals.

[0100] A model based on the Transformer architecture is used to process the input sequence through the self-attention mechanism to capture long-distance dependencies.

[0101] A feed-forward neural network is used to perform nonlinear transformation on features.

[0102] Applying positional encoding adds timing information to each element in the input data sequence.

[0103] A sparse activation strategy is adopted to activate only some neurons for calculation and activate all neurons when an anomaly is detected.

[0104] Step S4: Abnormal monitoring and traceability step Establish a real-time monitoring layer to record the model’s input, processing, and output results.

[0105] Reduce the black box problem of models through visual monitoring and logging.

[0106] When an exception occurs, the computing process can be traced back to help identify the source of the problem.

[0107] Step S5: Flight arrival time prediction step Based on the optimized data obtained in the above steps S1 to S4, the expected arrival time of the civil aviation flight is predicted.

[0108] Working process Data collection and integration: Get the latest flight status information through real-time data collection interfaces from various data sources (ADS-B, meteorological satellites, runway status monitoring, air traffic control).

[0109] Data cleaning and preprocessing techniques are used to ensure data quality and consistency across data sources, including removing noise data, filling missing values, and standardizing data formats.

[0110] Dynamic Weight Allocation: According to the flight phase of the current flight, the Bayesian optimization algorithm is used to dynamically adjust the weight of each data source.

[0111] The weight adjustment process responds to changes in flight status in real time, ensuring that the most relevant data sources contribute to the prediction model to the maximum extent at different flight stages.

[0112] Data fusion and feature construction: Through weighted average or weighted aggregation algorithm, the data from each data source are fused according to weights to construct comprehensive flight status features.

[0113] Construct an input sequence containing dynamic weight adjustment information and time synchronization interval adjustment information to provide comprehensive input for the deep learning model.

[0114] Deep learning model processing: The constructed input sequence is input into a deep learning model based on the Transformer architecture.

[0115] The model captures long-distance dependencies in input data through a self-attention mechanism, uses a feedforward neural network for nonlinear feature transformation, and enhances temporal information through positional encoding.

[0116] Abnormal monitoring and response: Monitor the model's input, processing, and output results in real time, and record detailed logs.

[0117] Through visual monitoring tools, operators can view the operating status of the model in real time and identify potential anomalies.

[0118] When an abnormal situation is detected, the system can trace back the calculation process, locate the source of the problem, and ensure the reliability of the prediction.

[0119] Estimated arrival time prediction and output: Based on the optimized data and the processing results of the deep learning model, the estimated arrival time prediction of civil aviation flights is generated.

[0120] The prediction results are provided to relevant aviation management systems through a user interface or interface for flight scheduling and resource optimization.

[0121] By integrating multi-source data and comprehensively considering multiple factors such as the flight's real-time location, weather conditions, runway status, and air traffic instructions, the accuracy of the estimated arrival time prediction can be significantly improved.

[0122] The dynamic weight adjustment mechanism ensures that the most relevant data sources are fully utilized at different flight stages to further optimize the prediction results.

[0123] The Bayesian optimization algorithm is used for weight allocation, which can flexibly adjust the weight allocation according to real-time data and flight status, and improve the system's adaptability to dynamic changes.

[0124] The intelligent adjustment of the time synchronization interval mechanism enables the system to provide high-precision predictions during critical flight phases while optimizing computing resources during the flight stabilization phase.

[0125] Through weighted average or weighted aggregation algorithms, computing resources can be reasonably allocated to avoid unnecessary data processing and improve the overall operating efficiency of the system.

[0126] In non-critical stages, extend the time synchronization interval, reduce the calculation frequency, and reduce the system burden.

[0127] High-precision positioning technology and data interpolation algorithms ensure the data quality and continuity of each data source, providing reliable data support for the prediction model.

[0128] Ground sensor networks and machine learning algorithms monitor and predict runway status in real time to ensure the accuracy and timeliness of runway status data.

[0129] The real-time monitoring layer records detailed model operation data. Through visual monitoring tools, operators can intuitively understand the working process of the model and reduce black box problems.

[0130] The anomaly monitoring and traceability mechanism can quickly locate the source of problems when they occur, improving system reliability and user trust.

[0131] Accurate ETA forecasts provide reliable data support for aviation management departments, optimizing flight scheduling and resource allocation, and reducing flight delays and resource waste.

[0132] The priority sorting algorithm optimizes landing order, improves airport operation efficiency, and reduces flight waiting time on the ground and in the air.

[0133] By implementing the above specific methods, this embodiment demonstrates an efficient multi-source data fusion and dynamic weight adjustment mechanism, which significantly improves the accuracy of civil aviation flight estimated arrival time prediction based on ADS-B and the adaptability of the system. This method integrates multiple data sources and uses advanced optimization algorithms to dynamically adjust data weights to ensure that the prediction model can fully utilize the most relevant data sources in different flight phases to provide high-precision prediction results. At the same time, the system's abnormal monitoring and traceability design improves the system's reliability and user trust, and provides strong data support and decision-making basis for aviation management departments.

[0134] Embodiment 2: like Figure 1 to Figure 2 As shown, this embodiment describes in detail a method for calculating the expected arrival data of civil aviation flights based on ADS-B, with a particular focus on the Transformer model and its multi-head attention mechanism, position encoding, and sparse activation strategy in the deep learning optimization step (step S3); at the same time, the real-time monitoring layer, logging module, visualization analysis tools, and distributed storage technology in the abnormality monitoring and traceability step are introduced. This method is further improved on the basis of Example 1, focusing on model optimization and abnormality monitoring mechanisms to improve prediction accuracy and system reliability.

[0135] Step S3: Deep learning optimization step Input sequence construction: Fusion data and dynamic adjustment information: The comprehensive data obtained by multi-source data fusion and dynamic weight adjustment in step S1 of Example 1 is combined with the time synchronization interval information dynamically adjusted in step S2 to construct an input sequence.

[0136] The input sequence includes the weighted aggregation results of multi-source data, the dynamic weight of each data source, and the length of the current time synchronization interval.

[0137] Transformer model architecture: Multi-layer self-attention mechanism: The Transformer model consists of a multi-layer self-attention mechanism and a fully connected feedforward neural network, each layer of which includes a multi-head attention sublayer and a feedforward network sublayer.

[0138] Multi-Head Attention: By using multiple attention heads in parallel (e.g. 8 heads), the model's ability to capture dependencies between different features and different positions is enhanced. Each attention head calculates the attention weight independently, and finally the outputs of each head are concatenated and linearly transformed to obtain rich feature representations.

[0139] Self-attention mechanism formula: Figure 5 As shown; Multi-head attention formula: Figure 6 Shown Position Encoding: Since the Transformer model itself does not have the ability to process the order of sequence data, position encoding is introduced to provide timing information.

[0140] The position encoding generated by the sine and cosine functions is: Figure 7 Shown Feedforward Neural Network (Feedforward Neural Network Part): Nonlinear feature transformation: In each feedforward neural network sublayer of the Transformer layer, a fully connected layer is applied for nonlinear transformation, which usually includes two linear transformation layers and a ReLU activation function to increase the expressiveness of the model and capture complex nonlinear relationships.

[0141] Feedforward neural network formula: Figure 8 Shown Sparse activation strategy: Dynamic gating mechanism: A dynamic gating mechanism is used to implement a sparse activation strategy, which activates only some neurons (for example, the activation ratio is 30%) under normal conditions to reduce the amount of calculation and improve operating efficiency.

[0142] Sparse activation strategy formula: Fig. 9 Shown Anomaly Detection and Global Neuron Activation: When the system detects an anomaly (such as drastic changes in flight status, emergencies, etc.), it triggers the activation of all neurons through global signals to ensure that the model can respond comprehensively in emergency situations, improving the accuracy of predictions and the reliability of the system.

[0143] Model training and optimization: Loss function and optimization algorithm: The mean square error (MSE) is used as the loss function to measure the difference between the predicted value and the actual estimated arrival time.

[0144] Loss function formula: Fig.10As shown; The Adam optimization algorithm is used to update the model parameters to accelerate convergence and improve model performance.

[0145] Adam optimization algorithm update formula: Fig.11 Shown Step S4: Abnormal monitoring and traceability step Real-time monitoring layer construction (step S4, real-time monitoring layer part): Integrated logging module: A logging module is set up to record the model’s input data, weight adjustment, time synchronization process and output results in real time.

[0146] Distributed storage technology formula: Fig.12 As shown; Visual analysis tools: Integrate visual analysis tools (such as Grafana, Kibana, etc.) to display the running status of the model at each time step in real time.

[0147] Multi-dimensional data dynamic query and backtracking formula: Fig.13 Shown Reduction of black box problems (step S4, transparency improvement part): Improved transparency: Through real-time recording and visual monitoring, operators can intuitively understand the model's calculation process and decision-making basis, reducing black box problems.

[0148] Detailed log and report generation formula: Fig.14 Shown Exception handling and traceability (step S4, exception detection and response part): Anomaly Detection and Response: When the system detects anomalies in the operation of the model (such as excessive prediction error, abnormal input data, etc.), it automatically triggers the alarm mechanism to notify relevant personnel to take measures.

[0149] Anomaly detection formula: Fig.15 Shown Calculation process backtracking: When an anomaly occurs, the real-time monitoring layer can trace back to the specific calculation process, identify possible problems in the data source, algorithm adjustment or model inference process, and help quickly locate and solve the problem.

[0150] Backtracking step formula: Fig.16 As shown; Working process Input sequence preparation and model processing: The multi-source data, dynamic weights and time synchronization interval information processed in steps S1 and S2 are integrated to construct an input sequence.

[0151] The input sequence is fed into a deep learning model based on the Transformer architecture to predict the expected arrival time.

[0152] Model run with sparse activation: Under normal circumstances, the model processes data through a multi-head self-attention mechanism and a feedforward neural network, activating only some neurons to reduce the computational burden.

[0153] When an anomaly is detected, the dynamic gating mechanism triggers a full neuron activation pattern, ensuring that the model can fully respond to complex flight state changes.

[0154] Abnormal monitoring and logging: The real-time monitoring layer records every calculation detail of the model, including input data, weight adjustment, time synchronization process and output results.

[0155] The visual analysis tool displays the model operation status in real time and supports operators to perform dynamic queries and backtracking analysis.

[0156] Estimated arrival time prediction and feedback: Based on the output of the model, accurate civil aviation flight arrival time prediction results are generated.

[0157] The prediction results are provided to relevant aviation management systems through a user interface or interface for flight scheduling and resource optimization.

[0158] The multi-head self-attention mechanism enhances the model's ability to capture different features and can better understand and process complex flight data relationships.

[0159] Positional encoding provides rich temporal information, ensuring that the model can accurately understand the temporal order and dependencies of the data.

[0160] The sparse activation strategy uses a dynamic gating mechanism to activate only some neurons under normal conditions, significantly reducing the amount of computation and energy consumption.

[0161] Under abnormal circumstances, the model can quickly switch to full neuron activation mode to ensure the accuracy of predictions and the responsiveness of the system.

[0162] The real-time monitoring layer improves the interpretability of the system by integrating logging modules and visual analysis tools, helping operators understand and verify the calculation process of the model.

[0163] Distributed storage technology ensures high availability and integrity of log data and supports efficient management and query of large-scale data.

[0164] Through detailed logging and visual monitoring, the system can quickly identify and respond to operational anomalies, improving the reliability and stability of the overall system.

[0165] The exception handling and traceability mechanisms ensure that the system can still maintain efficient operation and accurate prediction in complex flight environments and emergencies.

[0166] High-precision and real-time estimated arrival time predictions provide aviation management departments with reliable data support, optimize flight scheduling and resource allocation, and reduce flight delays and resource waste.

[0167] The traceability and transparency of the system improve users' trust in the forecast results and enhance the decision-making support capabilities of aviation management.

[0168] By implementing the above specific method, this embodiment demonstrates a Transformer-based deep learning optimization and anomaly monitoring method, which significantly improves the accuracy of civil aviation flight arrival time prediction based on ADS-B and the reliability of the system. Specifically: The application of multi-head attention mechanism and position encoding enables the model to better understand and process complex relationships and time series characteristics in multi-source data, thereby improving prediction accuracy.

[0169] The sparse activation strategy uses a dynamic gating mechanism to activate some neurons only when necessary, reducing the computational burden and energy consumption and improving the system's operating efficiency.

[0170] Intelligent anomaly monitoring and traceability mechanisms enable the system to respond quickly to emergencies, maintain efficient and stable operation, and ensure the reliability of prediction results.

[0171] Real-time monitoring and visual analysis tools provide comprehensive model operation information, helping operators to understand the system status in a timely manner and make accurate management decisions.

[0172] Through detailed logging and visual monitoring, the black box problem of the model is reduced, the transparency of the system and the user's trust are improved, and more reliable technical support is provided to aviation management departments.

[0173] In summary, this embodiment significantly improves the performance and reliability of the civil aviation flight estimated arrival time prediction method based on ADS-B by introducing advanced Transformer model architecture, sparse activation strategy, and comprehensive anomaly monitoring and traceability mechanism, providing strong technical support and decision-making basis for air traffic management.

[0174] Embodiment 3: like Figure 1 to Figure 2As shown, this embodiment describes in detail a method for calculating civil aviation flight arrival data based on ADS-B, with a particular focus on the abnormal monitoring and traceability step (step S4) to ensure the interpretability of the model and the reliability of the system. This method is further improved on the basis of Example 1 and Example 2, focusing on the construction of the real-time monitoring layer, the integration of the log recording module, and the application of the visual analysis tool. The specific steps are as follows: Step S4: Abnormal monitoring and traceability step Real-time monitoring layer construction Integrated logging module: A logging module is set up to record the model’s input data, weight adjustment, time synchronization process and output results in real time.

[0175] Distributed storage technology formula: Fig.17 Shown Visual analysis tools: Integrate visual analysis tools (such as Grafana, Kibana, etc.) to display the running status of the model at each time step in real time.

[0176] Multi-dimensional data dynamic query and backtracking formula: Fig.18 Shown Reduce black box problems Improved transparency: Through real-time recording and visual monitoring, operators can intuitively understand the model's calculation process and decision-making basis, reducing black box problems.

[0177] Detailed log and report generation formula: Fig.19 Shown Exception handling and traceability Anomaly Detection and Response: When the system detects anomalies in the operation of the model (such as excessive prediction error, abnormal input data, etc.), it automatically triggers the alarm mechanism to notify relevant personnel to take measures.

[0178] Anomaly detection formula: Fig. 20 Shown Calculation process backtracking: When an anomaly occurs, the real-time monitoring layer can trace back to the specific calculation process, identify possible problems in the data source, algorithm adjustment or model inference process, and help quickly locate and solve the problem.

[0179] Backtracking step formula: Fig.21 Shown Working process Model operation and anomaly monitoring: During normal operation, the model monitors input data, weight adjustment, time synchronization process and output results in real time, and records detailed logs.

[0180] The visual analysis tool displays the operating status of the model in real time, and operators can dynamically view and analyze the operation of the model through multi-dimensional query functions.

[0181] Abnormal detection and alarm: The system uses a preset anomaly detection formula to monitor the difference between the predicted results and the actual results in real time. When the predicted error exceeds the set threshold, the system automatically triggers the alarm mechanism and notifies relevant personnel to intervene.

[0182] Calculation process backtracking and problem location: When an exception occurs, the system uses the log records and visual analysis tools of the real-time monitoring layer to trace back the computing process before and after the exception event.

[0183] By analyzing input data, weight adjustment, time synchronization process and output results, you can quickly identify the source of problems, such as data source anomalies, weight adjustment errors or model inference deviations.

[0184] Problem Solving and System Optimization: Relevant personnel take corresponding measures to solve the problem based on the results of retrospective analysis, such as repairing the data source, adjusting the algorithm or retraining the model.

[0185] The system optimizes logging and anomaly detection mechanisms based on processing results to improve overall system reliability and prediction accuracy.

[0186] The real-time monitoring layer records and displays the running status of the model at each time step by integrating logging modules and visual analysis tools, helping operators to intuitively understand the calculation process of the model and reduce black box problems.

[0187] The multi-dimensional data dynamic query and backtracking function supports operators to deeply analyze model behavior and performance, and quickly locate and solve abnormal problems.

[0188] Through real-time anomaly detection and response mechanism, the system can quickly identify and respond to abnormal situations during operation, ensuring the accuracy of prediction results and the stable operation of the system.

[0189] The calculation process backtracking mechanism helps to quickly locate the source of the problem, reduce system downtime, and improve overall reliability.

[0190] Detailed logging and visual analysis tools provide operators with comprehensive model operation information, supporting more accurate and timely aviation management decisions.

[0191] Accurate estimated arrival time prediction and efficient exception handling mechanism optimize flight scheduling and resource allocation, and reduce flight delays and resource waste.

[0192] Through transparent model operation process and detailed log records, users' trust in prediction results is improved, and the system's usability and user satisfaction are enhanced.

[0193] Visual monitoring tools provide an intuitive display of system operation status, helping operators to promptly identify and handle potential problems and ensure efficient operation of the system.

[0194] By implementing the above specific methods, this embodiment demonstrates an anomaly monitoring and traceability mechanism, which significantly improves the system transparency and reliability of the civil aviation flight estimated arrival time prediction method based on ADS-B. The real-time monitoring layer provides comprehensive model operation information through integrated logging modules and visual analysis tools, helping operators to timely understand and verify the calculation process of the model and reduce black box problems. At the same time, the anomaly detection and calculation process backtracking mechanism ensures that the system can quickly locate and solve problems when anomalies occur, improving the overall reliability of the system and user trust. This mechanism provides strong technical support and decision-making basis for aviation management departments, optimizes flight scheduling and resource allocation, and improves the efficiency and safety of air traffic management.

[0195] Finally, it should be noted that the above preferred embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail through the above preferred embodiments, those skilled in the art should understand that various changes can be made in form and details without departing from the scope defined by the claims of the present invention.

Claims

1. A method for calculating civil aviation flight arrival data based on ADS-B, characterized by: The following steps are involved: S1. Multi-source data fusion steps: Integrate multi-source data from ADS-B data, meteorological satellite data, airport runway status monitoring data and air traffic control real-time instruction data; Dynamically adjust the weights of the above data sources according to the flight phase of the flight; S2. Steps to dynamically adjust the time synchronization interval: Adjust the time synchronization interval according to the flight phase (take-off, cruise, landing); The time synchronization interval is set to A seconds during takeoff and landing, and the time synchronization interval is set to B seconds during cruising; Intelligently adjust the time synchronization interval length according to the changes in real-time flight status; S3. Deep learning optimization steps: Construct an input sequence, including fused multi-source data, dynamically adjusted data source weights, and dynamically adjusted time synchronization intervals; A model based on the Transformer architecture is used to process input sequences through a self-attention mechanism to capture long-distance dependencies. Use feedforward neural network to perform nonlinear transformation on features; Applying positional encoding adds timing information to each element in the input data sequence; A sparse activation strategy is used to activate only some neurons for calculation, and all neurons are activated when an anomaly is detected; S4. Abnormal monitoring and traceability steps: Establish a real-time monitoring layer to record the model’s input, processing, and output results; Reduce the black box problem of models through visual monitoring and logging; When an exception occurs, the computing process can be traced back to help identify the source of the problem; S5. Flight arrival time prediction steps: Based on the optimized data obtained in the above steps S1 to S4, the expected arrival time of the civil aviation flight is predicted.

2. The method for calculating the expected arrival data of civil aviation flights based on ADS-B according to claim 1, characterized in that: In the S1 multi-source data fusion step, the ADS-B data includes the real-time position, speed, altitude and heading information of the flight. The real-time position of the flight adopts high-precision positioning technology to improve the spatial resolution and accuracy of the data.

3. The method for calculating the expected arrival data of civil aviation flights based on ADS-B according to claim 1, characterized in that: In the S1 multi-source data fusion step, the meteorological satellite data includes wind speed, wind direction, airflow and temperature information. The meteorological data is acquired through a real-time data stream interface, and a data interpolation algorithm is used to supplement missing data to ensure the continuity and integrity of meteorological information.

4. The method for calculating the expected arrival data of civil aviation flights based on ADS-B according to claim 1, characterized in that: In the S1 multi-source data fusion step, the airport runway status monitoring data includes runway usage status, slippery degree and snow accumulation conditions. The runway status data is collected in real time through a ground sensor network, and a machine learning algorithm is used to predict and evaluate the runway slippery degree.

5. The method for calculating the expected arrival data of civil aviation flights based on ADS-B according to claim 1, characterized in that: In the S1 multi-source data fusion step, the air traffic control real-time instruction data includes the relative position of the flight to other flights, landing queue and runway change information. The real-time instruction data is obtained through the air traffic management system interface, and the instructions are processed using a priority sorting algorithm to optimize the flight landing order.

6. The method for calculating the expected arrival data of civil aviation flights based on ADS-B according to claim 1, characterized in that: The dynamic data source weight adjustment in the S1 multi-source data fusion step adopts a weighted average or weighted aggregation algorithm, wherein the weight changes dynamically according to the flight phase of the flight and the real-time reliability of the data source. The weight adjustment is based on a Bayesian optimization algorithm to achieve optimal weight distribution.

7. The method for calculating the expected arrival data of civil aviation flights based on ADS-B according to claim 1, characterized in that: In the S2 step of dynamically adjusting the time synchronization interval, the intelligent adjustment mechanism is based on real-time flight data and flight status. When airflow changes or sudden control instructions are detected, the time synchronization interval is automatically shortened, and the sliding window technology is used to dynamically manage the time synchronization interval to improve the real-time and accuracy of the prediction.

8. The method for calculating the expected arrival data of civil aviation flights based on ADS-B according to claim 1, characterized in that: In the S3 deep learning optimization step, the Transformer model includes a multi-layer self-attention mechanism and a fully connected feedforward neural network. The self-attention mechanism uses multi-head attention to enhance the model's ability to capture different features, and the position encoding is generated using sine and cosine functions to provide rich timing information.

9. The method for calculating the expected arrival data of civil aviation flights based on ADS-B according to claim 1, characterized in that: In the S3 deep learning optimization step, the sparse activation strategy selectively activates neurons through a dynamic gating mechanism, activating only a specific proportion of neurons under normal conditions to reduce the amount of computation, and triggers the activation of all neurons through a global signal when an abnormality is detected to ensure the comprehensive responsiveness of the model in emergency situations.

10. The method for calculating the expected arrival data of civil aviation flights based on ADS-B according to claim 1, characterized in that: In the S4 anomaly monitoring and traceability step, the real-time monitoring layer records and displays the model's input data, weight adjustment, time synchronization process and output results at each time step in real time by integrating a logging module and a visual analysis tool. The logging uses distributed storage technology to ensure high availability and integrity of the data, and the visual analysis tool supports dynamic query and backtracking of multi-dimensional data to improve the interpretability of the model and the decision support capabilities of operators.

Images