Critical inflection point-based departure flight metering control and time sequence prediction method

Abstract

The embodiment of the invention discloses a critical inflection point-based departure flight metering control and time sequence prediction method. The method comprises the steps of unobstructed taxiing time estimation, saturation critical inflection point identification, departure runway service mode statistics, taxiing time increment factor calculation, flight taxiing time sequence prediction model construction and saturation critical inflection point control. According to the method provided by the embodiment of the invention, accurate measurement of scene traffic is realized on the premise of not increasing the runway capacity by predicting the flight taxiing time at high precision, scientifically regulating and controlling the deducing opportunity and optimizing the deducing-takeoff flight sequence, so that taxiing congestion is effectively relieved, the flow relationship between taxiways and runways is balanced, and the traffic flow rate is improved. And the overall operation efficiency and the safety margin of the complex hub airport are improved.

Description

Technical Field

[0001] This invention relates to the field of flight departure management technology, specifically to a departure flight metering control and timing prediction method based on critical inflection points. Background Technology

[0002] In the complex operational environment of hub airports, departing flights, due to their more diverse control methods and wider adjustment space, exhibit significantly higher operational complexity than arriving flights. Therefore, the departure taxiing phase often becomes a critical link restricting flight release efficiency and overall operational stability. During the departure taxiing phase, the transition from pushback to runway takeoff involves multiple operational stages, including apron control, surface taxiing, and in-flight release. Any incoordination in any stage can lead to a decrease in overall operational efficiency. Furthermore, the departure taxiing process is not only affected by the aircraft's own taxiing characteristics but also by multiple external factors such as runway capacity limitations, surface congestion, and air traffic control instructions. These factors are coupled, making the taxiing phase highly dynamic and complex.

[0003] In current operations, the lack of effective traffic flow coordination and measurement strategies during flight pushback and taxiing phases easily leads to localized congestion in the taxiway network. Especially during peak hours, flights often form excessively long takeoff queues as they taxi to the runway end, resulting in decreased runway utilization. Prolonged idling by aircraft increases fuel consumption and emissions, and can also cause cascading delays between flights, further affecting the overall operational rhythm of the airport and the stability of flight release schedules.

[0004] Patent application number 2025103348607 discloses a method and system for predicting critical time on the airport surface based on gradient boosting decision trees. This method predicts the unobstructed taxiing time of a target aircraft on its planned taxiing path based on historical radar tracks and average unobstructed taxiing speeds, and then incorporates this unobstructed taxiing time into a gradient boosting decision tree prediction model to predict the target aircraft's estimated taxiing time on the airport surface. The drawback of this approach is that current research on taxiing over-point times largely focuses on statistical regression and machine learning models. While the latter can capture complex nonlinear relationships, such learning methods are "black box" models, lacking interpretability and heavily reliant on the selection of historical features.

[0005] Patent application number 2016106733433 discloses a method for predicting aircraft surface taxiing time based on multiple regression analysis. This method extracts the most fundamental factors affecting surface taxiing time by transforming different surface taxiing time data and using different independent variable selection strategies. It then establishes aircraft surface taxiing time prediction models based on multiple linear regression and multiple curvilinear regression analysis, and compares the prediction errors of different models to improve the predictive capability of aircraft surface taxiing processes within airport flight areas. The drawback of this approach is that multiple linear or curvilinear regression assumes a relatively fixed and analytical functional relationship between independent variables and taxiing time. However, the actual taxiing process is affected by various nonlinear, coupled, and dynamic factors, making it difficult for linear models to capture complex interactive effects.

[0006] The paper "Aircraft Taxiing Time Prediction Model Based on Digital Twin" discloses an aircraft taxiing time prediction model based on digital twins. It constructs a virtual twin of the aircraft based on its physical entity, including its geometry and dynamic characteristics. Using influencing factors as input, the model simulates the aircraft's taxiing behavior in the virtual space, obtaining the corresponding taxiing time, thus forming input and output virtual training samples. These samples are then used to train the SVR-BP combined prediction model to achieve aircraft taxiing time prediction. The drawbacks of this approach are: the digital twin model requires the construction of a high-precision 3D simulation environment of the airport surface, taxiing path topology, and dynamic interaction mechanism, resulting in high development and maintenance costs and a technical threshold far exceeding traditional multiple regression or machine learning methods. Furthermore, to achieve consistency with the real-world situation, real-time synchronization of flight status, weather, and pavement occupancy data is necessary. However, these data sources are diverse and have varying latency, leading to a "time lag" discrepancy between the twin system and reality. Summary of the Invention

[0007] The purpose of this invention is to provide a departure flight metering control and timing prediction method based on critical inflection points to solve the problems of taxiing congestion and low release efficiency during the departure phase of complex hub airports.

[0008] To achieve the above objectives, embodiments of the present invention provide a method for departing flight metering control and timing prediction based on critical inflection points, comprising:

[0009] S1, Unobstructed taxiing time estimation: Obtain historical flight operation dataset, and based on the historical flight operation dataset, estimate the baseline taxiing time for the parking stand-runway combination under the assumptions of no queuing, no waiting, and no control conflicts to obtain the unobstructed taxiing time;

[0010] S2, Saturation Critical Inflection Point Identification: Filter the peak departure traffic flow and time period information of specific runways in the historical flight dataset to obtain the corresponding runway departure peak time slice flight operation dataset, count the taxiing congestion level of each peak time slice, and determine the saturation critical inflection point when the flight operation set of the runway changes from a steady state to a congested state; traverse all runways of the airport to obtain the saturation critical inflection points of all runways of the airport.

[0011] S3, Departure Runway Service Mode Statistics: Based on the flight operation dataset of the departure peak time slice, the departure interval of consecutive departing flight pairs when departing from the runway is statistically analyzed, and the probability quality function of the departure interval is statistically analyzed to obtain the departure runway service mode.

[0012] S4, Quantitative calculation of taxiing time increment factor: Based on the historical flight operation dataset and the departure peak time slice flight operation dataset, combined with different scene congestion conditions, quantitatively evaluate the additional taxiing time caused by conflicts and taxiing interference, and identify the taxiing time increment factor under each congestion level;

[0013] S5, Flight Taxiing Time Prediction Model Construction: A flight taxiing time prediction model is constructed by using unobstructed taxiing time, departure runway service mode, and taxiing time increment factor;

[0014] S6 combines real-time operational data with a flight taxiing timing prediction model to predict flight taxiing duration, calculates surface congestion, and combines the saturation critical inflection point to implement metering and pushback coordinated control of departing flight traffic.

[0015] As a specific implementation of this application, step S1 is as follows:

[0016] Obtain historical flight operation datasets, preprocess the historical flight operation datasets, and filter out parking stand-runway combination data;

[0017] Extract the actual taxiing time of flights corresponding to each parking stand-runway combination;

[0018] For each group of parking stand-runway combination data, the lowest 20th percentile of the actual taxiing time distribution of flights is taken as the unobstructed taxiing time.

[0019] As a specific implementation of this application, step S2 is as follows:

[0020] The peak departure traffic flow and time period information for specific runways are filtered from the historical flight operation dataset, and peak time slots are divided using a preset time as the smallest unit to obtain the runway data. r Peak departure time slots collection ;

[0021] Set up statistics window Where n is the window length, For statistical window time offset;

[0022] Statistics on the level of traffic congestion at each peak time. Among them, the surface taxiing congestion level is defined as the number of departing flights taxiing at time t;

[0023] According to the statistics window Calculate runway departure rate Exit rate here =

Statistics Window

[0024] Iterate through all peak time slots and calculate the congestion level for each scene. Average runway departure rate ;

[0025] Fitted curve Find the x-coordinate of the fitted curve when the slope equals the preset saturation inflection point slope, and determine this point as the critical saturation inflection point of the runway r. The horizontal axis represents the critical inflection point of saturation.

[0026] Repeat the above process to obtain the saturation critical inflection point of all runways in the airport.

[0027] As a specific implementation of this application, step S3 is as follows:

[0028] Select peak time slices And must meet ;

[0029] Calculate each peak time slice sequentially Corresponding statistics window For runway intervals of consecutive departing flight pairs on the same runway, the highest 1% to 5% interval values ​​in the global data are removed.

[0030] Based on the remaining interval data after removing anomalies, the probability quality function of the departure interval is calculated, which is the departure runway service mode for that runway.

[0031] As a specific implementation of this application, step S4 is as follows:

[0032] Based on the historical flight operation dataset, an unobstructed taxiing time is generated for each flight, and the flight is recorded. f Unobstructed gliding time ;

[0033] Based on experience, traffic congestion levels are categorized into low-congestion, medium-congestion, and high-congestion zones, each corresponding to a taxiing time increment factor. ,in This is a factor for increasing taxiing time in low-congestion areas. This is a factor for increasing coasting time in moderately congested areas. This is a factor for increasing taxiing time in highly congested areas;

[0034] Construct simulation functions and execute the simulation process;

[0035] Algorithm selected to update conflict increment factor In this process, the particle swarm optimization algorithm is used to repeat the simulation process in each iteration, traversing the preset parameter range and recording the corresponding chi-square statistics results.

[0036] Select the conflict increment factor corresponding to the minimum value from the set of chi-square statistics. That is, as the optimal conflict increment factor .

[0037] As a specific implementation of this application, the expression for the flight taxiing timing prediction model in step S5 is as follows:

[0038]

[0039] in, For flights f The estimated takeoff time For flights f The actual launch time This refers to the actual level of congestion at the time of flight pushback. This is the optimal conflict increment factor.

[0040] As a specific implementation of this application, step S6 is as follows:

[0041] (1) The real-time flight schedule information is adjusted according to the expected wheel-shifting time. Arrange the data in ascending order to generate an initial virtual queue;

[0042] (2) Determine whether the virtual queue is an empty set. If it is, proceed to step (7). If it is not empty, calculate the surface taxiing congestion level corresponding to the first flight in the departure virtual queue based on the planned execution queue. The planned execution queue is the set of planned execution flights with allocated departure time slots.

[0043] (3) Determine whether the surface taxiing congestion level is lower than the runway saturation critical inflection point. If so, proceed to step (4); otherwise, proceed to step (6).

[0044] (4) Call the flight taxiing timing prediction model to predict the taxiing time of the first flight in the virtual queue, and calculate the expected take-off time and release order of the flight accordingly;

[0045] (5) Add the flight to the flight execution queue set, remove it from the virtual queue, and return (2);

[0046] (6) Find the flight with the earliest departure time in the flight execution queue, take the sum of the departure time of the flight and the predefined delay time as the expected launch time of the first flight in the virtual queue, and return (2).

[0047] (7) Output the final plan execution queue.

[0048] The method provided in this invention can accurately predict flight taxiing time, scientifically control pushback timing, and optimize takeoff release sequences. It can achieve "precise measurement" of ground traffic without increasing runway capacity, thereby effectively alleviating taxiing congestion, balancing the flow relationship between taxiways and runways, and improving the overall operational efficiency and safety margin of complex hub airports. Attached Figure Description

[0049] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the specific embodiments or the prior art will be briefly introduced below.

[0050] Figure 1 This is a flowchart of the departure flight metering control and timing prediction method based on critical inflection points provided in the embodiments of the present invention. Detailed Implementation

[0051] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0052] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0053] Under the existing operating model, due to the significant uncertainty in taxiing times for ground flights, controllers struggle to promptly grasp the saturation status of taxiways and runways. This leads to poor coordination between pushback, taxiing, and takeoff release, resulting in uneven ground traffic flow, localized congestion, and insufficient runway utilization. This invention, by identifying the critical saturation inflection point of ground traffic flow and combining it with a high-precision taxiing time prediction model, achieves metering-based coordinated control and taxiing sequence optimization for departing flights, thereby achieving:

[0054] 1. Anticipate traffic congestion trends in advance and scientifically regulate flight pushback and taxiing rhythms to achieve "precise quantitative management" of airport ground operations without increasing runway capacity, thereby avoiding problems such as idling taxiing and excessively long takeoff queues at the runway end;

[0055] 2. To address the problem of the disconnect between pushback control and taxiing prediction, this embodiment of the invention forms a closed-loop management mechanism of "prediction-measurement-control", which deeply couples taxiing pass point prediction with pushback time control, taxiing scheduling and runway allocation, realizing the transformation from static prediction to dynamic and executable control strategy.

[0056] 3. Compared with the "black box" method that relies on artificial intelligence algorithms such as deep learning, the solution proposed in this embodiment of the invention is based on quantifiable traffic characteristic analysis and theoretical modeling, which has clear physical meaning and good interpretability. The model has a simple structure, high computational efficiency, and strong real-time performance. It can provide stable and reliable results while ensuring prediction accuracy, and provide solid theoretical support for operation management decisions.

[0057] The technical solutions provided by the embodiments of the present invention mainly include:

[0058] (1) Unobstructed taxiing time estimation: that is, estimating the baseline taxiing time for the "parking stand-runway" combination under the assumption of "no queuing, no waiting, and no control conflicts";

[0059] (2) Identification of saturation critical inflection point: Capture the dynamic evolution trend of throughput capacity from increasing to slowing down to saturation and limit, and identify the critical inflection point of the transition from steady state to congested state during operation;

[0060] (3) Statistical analysis of departure runway service patterns: Statistical analysis of the service intervals of consecutive departing flight pairs when they depart from the runway, and construction of a probability distribution model of flight departure intervals;

[0061] (4) Quantitatively calculate the taxiing time increment factor: Combine different scene congestion conditions to quantitatively evaluate the extra taxiing time caused by conflicts and taxiing interference, and identify the characteristics of taxiing time increment under each congestion level;

[0062] (5) Construct a flight taxiing time prediction model: Construct a flight taxiing time prediction model by using unobstructed taxiing time, runway gate queuing time and taxiing time increment factors;

[0063] (6) Saturation critical inflection point control: By predicting the taxiing time sequence of planned flights, and combining the flight taxiing time prediction model and the characteristics of saturation critical inflection point, the departure flight traffic is measured and pushed out in a coordinated manner.

[0064] Please refer to Figure 1 The present invention provides a method for departing flight metering control and timing prediction based on critical inflection points, comprising the following steps:

[0065] S1, Unobstructed taxiing time estimation: Obtain historical flight operation dataset, and based on the historical flight operation dataset, estimate the baseline taxiing time for the parking stand-runway combination under the assumptions of no queuing, no waiting, and no control conflicts to obtain the unobstructed taxiing time.

[0066] In this specific implementation, step S1 includes:

[0067] ① First, the historical flight operation data is preprocessed to remove missing or abnormal data, such as negative taxiing time, duplicate flight records, extreme taxiing duration, timestamp errors, or missing time fields;

[0068] ② Filter out flight data with fixed combinations of parking stands and runways, and extract the actual taxiing time (i.e., the actual takeoff time minus the actual pushback time) for each "parking stand-runway" combination.

[0069] ③ Take the lowest 20th percentile of the taxiing time distribution for each fixed combination of parking position and runway as the unobstructed taxiing time.

[0070] S2, Saturation Critical Inflection Point Identification: Filter out peak departure traffic flow and time period information for specific runways from historical flight operation data to obtain the corresponding runway departure peak time slice flight operation dataset, count the taxiing congestion level of each peak time slice, and determine the saturation critical inflection point when the flight operation set of the runway changes from a steady state to a congested state; traverse all runways of the airport to obtain the saturation critical inflection points of all runways of the airport.

[0071] In this specific implementation, step S2 includes:

[0072] ① Filter out peak departure traffic flow and time periods for specific runways from historical flight operation data, and divide the peak time slots into 1-minute intervals to obtain the runway data. r Peak departure time slots collection ;

[0073] ② Set up the statistics window Where n is the window length, For statistical window time offset;

[0074] ③ Statistical analysis of traffic congestion levels at each peak time point. The surface taxiing congestion level is defined as the number of departing flights taxiing at time t.

[0075] ④ Calculate the runway departure rate equal to =

Statistics Window

[0076] ⑤ Traverse all peak time slots and calculate the congestion level for each scene. Average runway departure rate ;

[0077] ⑥ Use the "forced zero-crossing hyperbola" to fit the curve. Find the x-coordinate (congestion level) of the fitted curve when its slope equals the preset saturation inflection point slope, and define this point as the runway. r saturation critical inflection point The horizontal axis represents the critical inflection point of saturation.

[0078] ⑦ Repeating the above process will yield the saturation critical inflection point for all runways within the airport.

[0079] S3, Departure Runway Service Mode Statistics: Based on the flight operation dataset of the departure peak time slice, the departure interval of consecutive departing flight pairs when they reach the departure point is statistically analyzed, and a probability distribution model of the flight departure interval is constructed to obtain the departure runway service mode.

[0080] In this specific implementation, step S3 includes:

[0081] ① Select peak time slots And must meet ;

[0082] ② Calculate each peak time slot sequentially. Corresponding statistics window Runway spacing for consecutive departing flight pairs on the same runway;

[0083] ③ To reduce the interference of outliers on the results, the highest 1% to 5% interval values ​​in the global data are removed;

[0084] ④ Based on the remaining interval data after removing anomalies, the probability quality function of the departure interval is calculated, which is the departure runway service mode for that runway.

[0085] S4, Quantitative calculation of taxiing time increment factor: Based on the historical flight operation dataset and the departure peak time slice flight operation dataset, combined with different scene congestion conditions, quantitatively evaluate the additional taxiing time caused by conflicts and taxiing interference, and identify the taxiing time increment factor under each congestion level.

[0086] In this specific implementation, step S4 includes:

[0087] ① Based on historical flight operation data, the unobstructed taxiing time is generated for each flight using the method in step S1, and the flight time is recorded. f Unobstructed gliding time ;

[0088] ② Based on experience, the congestion level is divided into low-congestion, medium-congestion, and high-congestion zones, each corresponding to a taxiing time increment factor. ,in This is a factor for increasing taxiing time in low-congestion areas. This is a factor for increasing coasting time in moderately congested areas. This is a factor for increasing taxiing time in highly congested areas;

[0089] ③ Construct the simulation function and execute the following procedures:

[0090] 1) Iterate through all flights f Calculate the estimated takeoff time:

[0091]

[0092] in, For flights f The estimated takeoff time For flights f The actual launch time This refers to the actual level of congestion at the time of flight pushback. The factor for the descent time increment under this congestion level (to be solved).

[0093] 2) Based on the probability mass function of the departure interval obtained in step S3, randomly generate the time interval between consecutive departing flight pairs; and adjust the estimated departure time of all flights according to the first-come, first-served (FCFS) principle to obtain the calculated departure time of all flights. .

[0094] 3) Recalculate the distribution of flight congestion levels within the taxiing window. Takeoff rate index This is to reflect the simulation operation characteristics under different congestion conditions.

[0095] 4) Design the following evaluation criteria to evaluate the true values ​​of the gliding data. (From step S2) and simulation values The frequency distribution was used to construct the chi-square statistic as an evaluation index.

[0096] ④ Iterative optimization step ③ is a typical "black box" optimization problem, and the particle swarm optimization algorithm (but not limited to it) can be used to update the conflict increment factor. The algorithm repeats the simulation process in each iteration, traversing the preset parameter range and recording the corresponding chi-square statistics results.

[0097] ⑤ Select the conflict increment factor corresponding to the minimum value from the set of chi-square statistics. That is, as the optimal conflict increment factor .

[0098] S5, Flight Taxiing Time Prediction Model Construction: A flight taxiing time prediction model is constructed using unobstructed taxiing time and taxiing time increment factors.

[0099] In practice, a real-time flight operation plan data set is extracted from the flight flight plan, and combined with the optimal conflict increment factor obtained in step S4. A flight taxiing time series prediction model is constructed, and its expression is:

[0100]

[0101] in, For flights f The estimated takeoff time For flights f The actual launch time This refers to the actual level of congestion at the time of flight pushback. This is the optimal conflict increment factor.

[0102] As can be seen from step S4 above, this model can predict flight taxiing time. Simultaneously, based on the probability mass function of the departure intervals of consecutive flight pairs on the runway, the time intervals between consecutive departing flight pairs are randomly generated, and the estimated departure times of all flights are corrected according to the First-Come, First-Served (FCFS) principle to obtain the calculated departure time of the flight. .

[0103] S6, Saturation Critical Inflection Point Control: By predicting the taxiing sequence of planned flights, and combining the flight taxiing sequence prediction model with the characteristics of the saturation critical inflection point, the departure flight traffic is subject to coordinated control of measurement and pushback.

[0104] In this specific implementation, step S6 includes:

[0105] ① The real-time flight schedule information will be adjusted according to the estimated wheel chock time. Arrange the data in ascending order to generate an initial virtual queue;

[0106] ② Determine if the virtual queue is empty. If so, proceed to step ⑦. If not empty, calculate the surface taxiing congestion level for the first flight in the departure virtual queue based on the set of planned flights allocated to departure slots (i.e., the planned execution queue). This planned execution queue is the set of planned flights allocated to departure slots.

[0107] ③ Determine whether the taxiing congestion level is below the runway saturation critical inflection point. If yes, continue; otherwise, skip to ⑥.

[0108] ④ Call the flight taxiing timing prediction model established in step 5 to predict the taxiing time of the first flight in the virtual queue, and calculate the expected takeoff time and release order of the flight accordingly;

[0109] ⑤ Add the flight to the flight execution queue set, remove it from the virtual queue, and return to ②;

[0110] ⑥ Find the flight with the earliest departure time in the flight execution queue, and use the sum of the departure time and the predefined delay time of the flight as the estimated departure time of the first flight in the virtual queue, and return to ②; For example, find a flight with the earliest flight time among the multiple flights in the execution queue (let's call this time A), and call the delay time of the flight C. Then, A+C can be used as the estimated departure time of the first flight in the virtual queue.

[0111] ⑦ Output the final execution queue.

[0112] The advantages of implementing the technical solution provided in the embodiments of the present invention are as follows:

[0113] (1) This method achieves “precise measurement” of ground traffic without increasing runway capacity by accurately predicting flight taxiing time, scientifically controlling pushback timing, and optimizing takeoff release sequences. This effectively alleviates taxiing congestion, balances the flow relationship between taxiways and runways, and improves the overall operational efficiency and safety margin of complex hub airports.

[0114] (2) Strong technical interpretability and clear physical meaning: Compared with traditional "black box" flight taxiing time prediction methods such as neural networks, this scheme is based on the traffic characteristics and operating mechanism of the aircraft taxiing process for modeling, and each step is based on quantifiable physical parameters. The model structure is clear and logically traceable, and the prediction results are verifiable and physically interpretable, which makes it easier for researchers and operations managers to understand the prediction mechanism and the source of the results, and is more in line with the requirements of model transparency and reliability in high-safety operation fields such as air traffic control.

[0115] (3) It can realize a closed loop of "prediction-metering-control" for departing flights, solving the problem of the separation between departure pushback control and taxiing prediction: Traditional methods often only focus on taxiing time prediction or pushback timing optimization, lacking a linkage mechanism between the two, resulting in the prediction results not being directly used for control execution. This patented technology introduces a saturation critical inflection point control mechanism, directly using the output of the prediction model (such as taxiing time, congestion level, waiting time, etc.) for the metering control of flight pushback and departure traffic. This forms a closed loop system from "inflection point identification → taxiing status prediction → pushback timing control → traffic metering feedback", realizing the deep integration of departure flight prediction and control.

[0116] (4) The algorithm has a simple structure and excellent dynamic response and real-time operation capabilities: Compared with deep learning-based artificial intelligence models that rely on massive training samples and high-performance computing resources, this method has stronger deployability and practicality, and can be directly embedded into the operation control system to realize online dynamic decision-making. In the whole scheme, each core step (such as unobstructed gliding time estimation, gliding time increment factor calculation, etc.) adopts a statistical modeling method based on historical operation data. The model structure is clear and the computational complexity is low. It can quickly output prediction results under the condition of real-time data updates, and meet the high-frequency decision-making needs in complex operation environments.

[0117] Furthermore, the particle swarm optimization algorithm introduced in step 4 has advantages such as strong global optimization capability, fast convergence speed, and simple parameter settings. This algorithm only needs to be trained or calibrated once using historical data in the initial stage, without retraining the model in each prediction. It can efficiently complete the optimization solution of coasting timing and push-out control during operation, significantly improving the algorithm's dynamic adaptability and real-time computing performance.

[0118] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for metering control and time-series prediction of departing flights based on critical inflection points, characterized in that, include: S1, Unobstructed taxiing time estimation: Obtain historical flight operation dataset, and based on the historical flight operation dataset, estimate the baseline taxiing time for the parking stand-runway combination under the assumptions of no queuing, no waiting, and no control conflicts to obtain the unobstructed taxiing time; S2, Saturation Critical Inflection Point Identification: Filter out peak departure traffic flow and time period information for specific runways from historical flight operation data to obtain the corresponding runway departure peak time slice flight operation dataset, calculate the taxiing congestion level for each peak time slice, and determine the saturation critical inflection point where the flight operation set of the runway changes from a steady state to a congested state; traverse all runways of the airport to obtain the saturation critical inflection points of all runways in the airport. S3, Departure Runway Service Mode Statistics: Based on the flight operation dataset of the departure peak time slice, the departure interval of consecutive departing flight pairs when departing from the runway is statistically analyzed, and the probability quality function of the departure interval is statistically analyzed to obtain the departure runway service mode. S4, Taxiing time increment factor calculation: Based on the historical flight operation dataset and the departure peak time slice flight operation dataset, combined with different scene congestion conditions, the additional taxiing time caused by conflicts and taxiing interference is quantitatively evaluated, and the taxiing time increment factor under each congestion level is identified. S5, Flight Taxiing Time Prediction Model Construction: A flight taxiing time prediction model is constructed by using unobstructed taxiing time, departure runway service mode, and taxiing time increment factor; S6 combines real-time operational data with a flight taxiing timing prediction model to predict flight taxiing duration, calculates surface congestion, and combines the saturation critical inflection point to implement metering and pushback coordinated control of departing flight traffic.

2. The method as described in claim 1, characterized in that, Step S1 is as follows: Obtain historical flight operation datasets, preprocess the historical flight operation datasets, and filter out parking stand-runway combination data; Extract the actual taxiing time of flights corresponding to each parking stand-runway combination; For each group of parking stand-runway combination data, the lowest 20th percentile of the actual taxiing time distribution of flights is taken as the unobstructed taxiing time.

3. The method as described in claim 2, characterized in that, The historical flight operation dataset is preprocessed by removing missing or abnormal data, including negative taxiing time, duplicate flight records, extreme taxiing time, timestamp errors, or missing time fields.

4. The method as described in claim 1, characterized in that, Step S2 is as follows: The peak departure traffic flow and time period information for specific runways are filtered from the historical flight operation dataset, and peak time slots are divided using a preset time as the smallest unit to obtain the runway data. r Peak departure time slots collection ; Set up statistics window Where n is the window length, For statistical window time offset; Statistics on the level of traffic congestion at each peak time. Among them, the surface taxiing congestion level is defined as the number of departing flights taxiing at time t; Calculate runway departure rate =[Statistics Window] [Number of departing flights within the time interval] 60 / 2n; Iterate through all peak time slots and calculate the congestion level for each scene. Average runway departure rate ; Fitted curve Find the x-coordinate of the fitted curve when the slope equals the preset saturation inflection point slope, and define this point as the runway. r saturation critical inflection point The horizontal axis represents the critical inflection point of saturation. Repeat the above process to obtain the saturation critical inflection point of all runways in the airport.

5. The method as described in claim 4, characterized in that, Step S3 is as follows: Select peak time slices And must meet ; Calculate each peak time slice sequentially Corresponding statistics window For runway intervals of consecutive departing flight pairs on the same runway, the highest 1% to 5% interval values ​​in the global data are removed. Based on the remaining interval data after removing anomalies, the probability quality function of the departure interval is calculated, which is the departure runway service mode for that runway.

6. The method as described in claim 5, characterized in that, Step S4 is as follows: Based on the historical flight operation dataset, an unobstructed taxiing time is generated for each flight, and the flight is recorded. f Unobstructed gliding time ; Based on experience, traffic congestion levels are categorized into low-congestion, medium-congestion, and high-congestion zones, each corresponding to a taxiing time increment factor. ,in This is a factor for increasing taxiing time in low-congestion areas. This is a factor for increasing coasting time in moderately congested areas. This is a factor for increasing taxiing time in highly congested areas; Construct simulation functions and execute the simulation process; Algorithm selected to update conflict increment factor In this process, the particle swarm optimization algorithm is used to repeat the simulation process in each iteration, traversing the preset parameter range and recording the corresponding chi-square statistics results. Select the conflict increment factor corresponding to the minimum value from the set of chi-square statistics. That is, as the optimal conflict increment factor .

7. The method as described in claim 6, characterized in that, The specific steps for constructing the simulation function and executing the simulation process are as follows: (1) Traverse all flights f Calculate the estimated takeoff time: in, For flights f The estimated takeoff time For flights f The actual launch time This refers to the actual level of congestion at the time of flight pushback. This is the descent time increment factor for the congestion level to be solved; (2) Based on the probability mass function of the departure interval obtained in step S3, the time interval between consecutive departing flight pairs is randomly generated; and the estimated departure time of all flights is corrected according to the first-come, first-served principle to obtain the calculated departure time of all flights. ; (3) Re-analyze the distribution of flight congestion levels within the taxiing window. Takeoff rate index To reflect the simulation operation characteristics under different congestion conditions; (4) Calculate the true value of the gliding data Compared with simulation values The frequency distribution was used to construct the chi-square statistic as an evaluation index.

8. The method as described in claim 1, characterized in that, The expression for the flight taxiing timing prediction model in step S5 is: in, For flights f The estimated takeoff time For flights f The actual launch time This refers to the actual level of congestion at the time of flight pushback. This is the optimal conflict increment factor.

9. The method as described in claim 1, characterized in that, Step S6 is as follows: (1) The real-time flight schedule information is adjusted according to the expected wheel-shifting time. Arrange the data in ascending order to generate an initial virtual queue; (2) Determine whether the virtual queue is an empty set. If it is, proceed to step (7). If it is not empty, calculate the surface taxiing congestion level corresponding to the first flight in the departure virtual queue based on the planned execution queue. The planned execution queue is the set of planned execution flights with allocated departure time slots. (3) Determine whether the surface taxiing congestion level is lower than the runway saturation critical inflection point. If so, proceed to step (4); otherwise, proceed to step (6). (4) Call the flight taxiing timing prediction model to predict the taxiing time of the first flight in the virtual queue, and calculate the expected take-off time and release order of the flight accordingly; (5) Add the flight to the flight execution queue set, remove it from the virtual queue, and return (2); (6) Find the flight with the earliest departure time in the flight execution queue, take the sum of the departure time of the flight and the predefined delay time as the estimated departure time of the first flight in the virtual queue, and return (2). (7) Output the final plan execution queue.

Images