Method and device for predicting flight landing time

Abstract

The invention relates to a method and device for predicting flight landing time. The method comprises the following steps: acquiring flight data of a target flight, wherein the flight data comprises a current distance between the target flight and a target airport at a current moment; based on the position where the target flight reaches the circumference of a reference circle with the target airport as the center, a sector where the target flight of the reference circle falls is determined, and the sector of the reference circle is divided based on distribution of the positions where historical flights reach the circumference of the reference circle along the circumference of the reference circle; determining a reference landing duration of the target flight at the current moment based on the radius of the reference circle, the current distance and the landing duration of the latest one or more historical flights passing through the determined sector before the current moment from reaching the circumference of the reference circle to landing at the target airport; inputting the flight data and the reference landing duration into a machine learning model trained for predicting the landing duration; and determining a predicted landing moment of the target flight based on the current moment and the predicted landing duration output by the machine learning model.

Description

Technical Field

[0001] This disclosure relates to the field of aviation management technology, and more specifically, to a method, apparatus, computer-readable storage medium, and computer program product for predicting flight landing times. Background Technology

[0002] Airport ground handling services are typically arranged based on flight arrival times, and preparations need to begin well in advance of the estimated arrival time. Traditionally, the estimated arrival time is derived from the flight plan. However, delays caused by factors such as weather along the route, air traffic control, or mechanical failures can lead to significant discrepancies between the estimated and actual arrival times. This results in longer waiting times for ground handling personnel and equipment on the tarmac, leading to a substantial waste of resources. Summary of the Invention

[0003] A brief overview of this disclosure is given below to provide a basic understanding of some aspects of it. However, it should be understood that this overview is not an exhaustive summary of this disclosure. It is not intended to identify key or essential parts of this disclosure, nor is it intended to limit the scope of this disclosure. Its purpose is merely to present certain concepts of this disclosure in a simplified form as a prelude to the more detailed description that follows.

[0004] According to a first aspect of this disclosure, a method for predicting flight landing time is provided, the method comprising: acquiring flight data of a target flight, the flight data including the current distance between the target flight and a destination airport at the current time; determining the sector of a reference circle into which the target flight falls based on the position of the target flight reaching the circumference of the reference circle centered on the destination airport, wherein the sectors of the reference circle are divided based on the distribution of positions of historical flights reaching the circumference of the reference circle along the circumference of the reference circle; determining a reference landing time of the target flight at the current time based on the radius of the reference circle, the current distance, and the landing time experienced by one or more recent historical flights that landed at the destination airport via the determined sectors before the current time, from reaching the circumference of the reference circle to landing at the destination airport; inputting the flight data and the reference landing time into a machine learning model, the machine learning model being trained to predict the landing time; and determining the predicted landing time of the target flight based on the current time and the predicted landing time output by the machine learning model.

[0005] In some embodiments, the radius of the reference circle is determined based on a first empirical value of the distance between the flight to be landed at the destination airport and the destination airport at the start of the landing procedure, or based on a second empirical value of the flight time from the start of the landing procedure to landing at the destination airport.

[0006] In some embodiments, the first empirical value is between 300 km and 500 km, and the second empirical value is between 30 minutes and 40 minutes.

[0007] In some embodiments, dividing the reference circle into sectors includes: representing the distribution of the positions of historical flights arriving at the circumference of the reference circle as a scatter plot; assigning adjacent regions where the scatter density is greater than a threshold density to the same sector; dividing sectors whose angle range is greater than a threshold angle into multiple sectors; and assigning regions where the scatter density is not greater than the threshold density to the sectors adjacent to the region.

[0008] In some embodiments, assigning a region where the scatter density is not greater than a threshold density to a sector adjacent to the region includes: dividing the region where the scatter density is not greater than the threshold density into two sub-regions, and assigning each sub-region in the two sub-regions to a sector adjacent to the sub-region.

[0009] In some embodiments, dividing a sector whose angle range is greater than a threshold angle into multiple sectors includes: dividing a sector whose angle range is greater than a threshold angle into multiple sectors such that the angle range of each of the multiple sectors is not greater than the threshold angle.

[0010] In some embodiments, the reference landing time of the target flight at the current moment is determined as the product of the ratio of the current distance to the radius of the reference circle and the average landing time experienced by the most recent one or more historical flights from the circumference of the reference circle to landing at the destination airport.

[0011] In some embodiments, flight data may also include the landing direction of the target flight relative to the destination airport at the current time.

[0012] In some embodiments, in response to the current distance being greater than a threshold distance, the landing direction of the target flight relative to the destination airport at the current time is determined to be the landing direction of the flight that recently landed at the destination airport; or in response to the current distance not being greater than a threshold distance, the landing direction of the target flight relative to the destination airport at the current time is determined based on the change of the target flight's current position at the current time relative to the past position of the target flight at a previous time.

[0013] In some embodiments, the current distance is determined as the maximum circular distance on Earth between the current location of the target flight and the location of the destination airport at the current moment.

[0014] In some embodiments, flight data may also include the target flight’s departure time, planned flight duration, aircraft type, and one or more of the target flight’s latitude and longitude coordinates, speed, and heading at the current time.

[0015] In some embodiments, the departure time of the target flight is represented as a combination of a day of the week and an hour of a 24-hour day when it is fed into a machine learning model.

[0016] In some embodiments, the machine learning model includes an input module, a first prediction model, a second prediction model, and an output module. The input module is configured to receive flight data and a reference landing duration. The first prediction model is coupled to the input module and configured to generate a first prediction value based on the flight data and the reference landing duration. The second prediction model is coupled to the input module and configured to generate a second prediction value based on the flight data and the reference landing duration. The output module is coupled to the first prediction model and the second prediction model and configured to determine a predicted landing duration based on the first prediction value and the second prediction value. The first prediction model is different from the second prediction model.

[0017] In some embodiments, the first prediction model is an XGBoost model, and the second prediction model is a decision tree model.

[0018] In some embodiments, the output module is configured to perform the following calculation process:

[0019] ,

[0020] Among them, Y pre It is the predicted landing time, λ is the bias, and XGB pre The first prediction value generated by the XGBoost model, and Tree pre It is the second predicted value generated by the decision tree model.

[0021] In some embodiments, the bias λ is adjusted depending on the destination airport.

[0022] In some embodiments, while using a first version of the machine learning model to predict the landing time of a target flight, a second version of the machine learning model is trained, and wherein, in response to the completion of training of the second version of the machine learning model, the second version of the machine learning model is used instead of the first version to predict the landing time.

[0023] In some embodiments, a second version of the machine learning model is trained using the actual landing time of the target flight as a label after the target flight lands at the destination airport.

[0024] In some embodiments, during the process of a target flight from the circumference of the reference circle to landing at the destination airport, the method for predicting the landing time of the flight according to any embodiment of the first aspect of this disclosure is executed at intervals to update the predicted landing time of the target flight.

[0025] According to a second aspect of this disclosure, an apparatus for predicting flight landing time is provided. The apparatus includes an acquisition module, a determination module, and a prediction module. The acquisition module is configured to acquire flight data of a target flight, including the current distance between the target flight and the destination airport at the current time. The determination module is configured to determine the sector of the reference circle into which the target flight will land, based on the position of the target flight reaching the circumference of the reference circle centered on the destination airport. The sectors of the reference circle are divided based on the distribution of historical flights' arrival positions along the circumference of the reference circle. The determination module is configured to determine a reference landing time of the target flight at the current time, based on the radius of the reference circle, the current distance, and the landing time experienced by one or more recent historical flights that landed at the destination airport via the determined sectors before the current time, from reaching the circumference of the reference circle to landing at the destination airport. The prediction module is configured to input the flight data and the reference landing time into a machine learning model, which is trained to predict the landing time. The prediction module is configured to determine the predicted landing time of the target flight based on the current time and the predicted landing time output by the machine learning model.

[0026] According to a third aspect of this disclosure, an electronic device is provided, comprising a processor and a memory storing computer-executable instructions. When executed by the processor, the computer-executable instructions cause the processor to perform a method for predicting flight landing times according to any embodiment of the first aspect of this disclosure.

[0027] According to a fourth aspect of this disclosure, a computer-readable storage medium having computer program instructions stored thereon is provided. When executed by a computer, the computer program instructions cause the computer to perform a method for predicting flight landing times according to any embodiment of the first aspect of this disclosure.

[0028] According to a fifth aspect of this disclosure, a computer program product is provided, comprising program instructions. When executed by a processor, the program instructions implement the method for predicting flight landing times according to any embodiment of the first aspect of this disclosure. Attached Figure Description

[0029] The foregoing and other features and advantages of this disclosure will become clear from the following description of embodiments illustrated in conjunction with the accompanying drawings. The drawings, incorporated herein and forming a part of the specification, are further used to explain the principles of this disclosure and to enable those skilled in the art to make and use it. Wherein:

[0030] Figure 1 This is a flowchart illustrating a method for predicting flight landing times according to some embodiments of the present disclosure;

[0031] Figure 2 This is a schematic diagram illustrating sector partitioning according to some embodiments of the present disclosure;

[0032] Figure 3 This is a schematic block diagram illustrating a machine learning model for predicting flight landing times according to some embodiments of the present disclosure;

[0033] Figure 4 This is a schematic block diagram illustrating an apparatus for predicting flight landing times according to some embodiments of the present disclosure;

[0034] Figure 5 This is a schematic block diagram illustrating an electronic device according to some embodiments of the present disclosure;

[0035] Figure 6 This is a schematic block diagram illustrating a computer system in which embodiments of the present disclosure can be implemented.

[0036] Note that in the embodiments described below, the same reference numerals are sometimes used across different figures to denote the same parts or parts with the same function, and repeated descriptions are omitted. In some cases, similar reference numerals and letters are used to denote similar items, so once an item is defined in one figure, it does not need to be discussed further in subsequent figures.

[0037] For ease of understanding, the positions, dimensions, and extents of the structures shown in the accompanying drawings and other materials may not represent actual positions, dimensions, and extents. Therefore, this disclosure is not limited to the positions, dimensions, and extents disclosed in the accompanying drawings and other materials. Detailed Implementation

[0038] Various exemplary embodiments of the present disclosure will now be described in detail with reference to the accompanying drawings. It should be noted that, unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps set forth in these embodiments do not limit the scope of the present disclosure.

[0039] The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the scope of this disclosure or its application or use. That is, the structures and methods herein are shown in an exemplary manner to illustrate different embodiments of the structures and methods in this disclosure. However, those skilled in the art will understand that they merely illustrate exemplary ways that can be used to implement this disclosure, and not exhaustive ways. Furthermore, the drawings are not necessarily drawn to scale, and some features may be enlarged to show details of specific components.

[0040] In addition, techniques, methods and equipment known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods and equipment should be considered part of the specification.

[0041] In all examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values.

[0042] This disclosure provides a method for predicting flight landing times, which combines historical and real-time flight data and extracts valuable reference time information for predicting flight landing times by dividing the data into sectors, thereby assisting machine learning models to more accurately predict flight landing times.

[0043] The methods for predicting flight landing times according to various embodiments of the present disclosure will now be described in detail with reference to the accompanying drawings. It will be understood that actual methods for predicting flight landing times may include other processes, but to avoid obscuring the essential points of the disclosure, these other processes will not be discussed herein and are not shown in the accompanying drawings.

[0044] In addition, for ease of explanation, in this article, the flight whose landing time is to be predicted can be referred to as the target flight, and the flight that will have already landed at the time of prediction can be referred to as the historical flight.

[0045] Figure 1 A flowchart illustrating a method 100 for predicting flight landing times according to some embodiments of the present disclosure is shown. Figure 1 As shown, method 100 includes steps S102 to S110.

[0046] In step S102, the flight data of the target flight is obtained, including the current distance between the target flight and the destination airport at the current time.

[0047] In some embodiments, the current distance is determined as the maximum circular distance on Earth between the current position of the target flight and the position of the destination airport at the current moment. For example, the current distance d can be calculated by d = R × arccos[cos(Y1) × cos(Y2) × cos(X1-X2) + sin(Y1) × sin(Y2)], where R is the Earth's radius, [X1, Y1] are the latitude and longitude coordinates of the destination airport, and [X2, Y2] are the latitude and longitude coordinates of the current position of the target flight. The latitude and longitude coordinates of the destination airport are known values, while the latitude and longitude coordinates of the current position of the target flight can be obtained through an Automatic Dependent Surveillance Broadcast (ADS-B) system.

[0048] In some embodiments, flight data may further include one or more of the following: the target flight's departure time, planned flight duration, aircraft type, and the target flight's current latitude and longitude coordinates, speed, heading, and landing direction relative to the destination airport. For example, the target flight's departure time, planned flight duration, and aircraft type can be obtained from flight plan data, while the target flight's current latitude and longitude coordinates, speed, and heading can be obtained from ADS-B data.

[0049] In some embodiments, the landing direction of a target flight relative to its destination airport at the current moment can be determined by: in response to the target flight's current distance from the destination airport being greater than a threshold distance, determining the target flight's landing direction relative to the destination airport at the current moment as the landing direction of the most recently landed flight at the destination airport; or in response to the target flight's current distance from the destination airport being no greater than a threshold distance, determining the target flight's landing direction relative to the destination airport at the current moment based on changes in the target flight's current position relative to its past position at previous times (e.g., changes in latitude and longitude). When the target flight is far from the destination airport, its real-time heading may not correspond to the final landing direction, while flights on the same day at similar times may have substantially the same landing direction because they face substantially the same airport conditions, weather conditions, etc. Therefore, in this case, it is meaningful to estimate the target flight's landing direction by referring to the landing direction of the most recently landed flight. When the target flight is close to the destination airport, estimating the target flight's landing direction by the real-time position changes of the target flight becomes meaningful and provides more accurate information compared to referring to the landing direction of the most recently landed flight. The threshold distance can be set based on experience, for example, it can be set to no more than 100 kilometers, or no more than 75 kilometers, or no more than 50 kilometers, or for example, 40 kilometers.

[0050] As a non-restrictive example, flight data can be transmitted to the data platform via a Kafka message queue by a background timer program. The data platform then performs preliminary parsing of the flight data and writes it to the database, enabling access to real-time flight data and storage of historical flight data. The stored flight data can be preprocessed. For example, ADS-B data and flight locations can be aligned at the granular level to ensure data accuracy and accelerate data processing. Second-level ADS-B data can also be processed into minute-level data to reduce the density of high-density data and improve operational efficiency. For text-based data such as aircraft type, numerical processing can be performed, for example, through one-hot encoding. In some examples, the departure time of a target flight can be represented as a combination of a day of the week and an hour of the day. For example, if the departure time is 3 PM on Monday, October 20, 2025, it can be represented as [1, 15]; if the departure time is 7 AM on Tuesday, October 21, 2025, it can be represented as [2, 7]. This representation method facilitates subsequent input into machine learning models for processing and effectively reflects the periodicity of flights, which is beneficial for machine learning models to learn and uncover potential patterns.

[0051] In step S104, based on the position of the target flight arriving at the circumference of the reference circle centered on the destination airport, the sector in which the target flight falls is determined. The sectors of the reference circle are divided based on the distribution of the positions of historical flights arriving at the circumference of the reference circle along the circumference of the reference circle.

[0052] In some embodiments, the circumference of the reference circle includes the positions of all historical flights that began performing a five-sided approach procedure from all directions, centered on the destination airport; that is, the positions of historical flights when they began the landing procedure. In some embodiments, the radius of the reference circle may be determined based on a first empirical value of the distance between the flight to be landed at the destination airport and the destination airport at the start of the landing procedure. As a non-limiting example, the first empirical value may be between 300 km and 500 km, for example, set to 300 km, 350 km, 400 km, 450 km, 500 km, etc. In some embodiments, the radius of the reference circle may be determined based on a second empirical value of the flight time from the start of the landing procedure to landing at the destination airport. As a non-limiting example, the second empirical value may be between 25 minutes and 45 minutes, or between 30 minutes and 40 minutes. The radius of the reference circle can then be determined based on the average flight speed of the flight during the landing process, using the second empirical value.

[0053] When determining a flight's position from ADS-B data, due to gaps or missing data in the ADS-B data, the position where the flight arrives at the circumference of a reference circle can be determined as the closest position to the circumference of the reference circle in the flight's ADS-B data. This arrival position may fall exactly on the circumference of the reference circle, or it may appear inside or outside the circumference.

[0054] Since flights that begin landing from similar directions usually have similar landing times, dividing the landing into sectors by combining the historical landing trajectories of flights can help identify which flights begin landing from similar directions.

[0055] In some embodiments, dividing the reference circle into sectors includes: representing the distribution of historical flight arrivals along the circumference of the reference circle as a scatter plot; assigning adjacent regions with scatter point densities greater than a threshold density to the same sector; dividing sectors with angle ranges greater than a threshold angle into multiple sectors; and assigning regions with scatter point densities not greater than the threshold density to sectors adjacent to that region. For example, scatter point density can be represented as the number of scatter points per unit angle range. The threshold density and threshold angle can be specifically set according to actual needs, and are not particularly limited here.

[0056] In some embodiments, dividing a sector whose angle range is greater than a threshold angle into multiple sectors includes: dividing the sector whose angle range is greater than the threshold angle into multiple sectors, such that the angle range of each of the multiple sectors is not greater than the threshold angle. Further, the multiple sectors may have the same angle range. This helps to improve the granularity of sector division, thereby improving the accuracy of subsequent calculations of reference landing time.

[0057] In some embodiments, assigning a region where the scatter density is not greater than the threshold density to an adjacent sector includes: dividing the region where the scatter density is not greater than the threshold density into two sub-regions, and assigning each sub-region to an adjacent sector. Further, the two sub-regions may have the same angular range. In an alternative, the region where the scatter density is not greater than the threshold density can be directly assigned to an adjacent sector without dividing it into sub-regions. Compared to this alternative, the sub-regional allocation method is beneficial for improving the accuracy of subsequent reference landing time calculations.

[0058] Figure 2 A schematic diagram of sector partitioning according to some embodiments of the present disclosure is shown. For example... Figure 2As shown, the center O of reference circle 200 represents the destination airport of the target flight, and each point represents the position of a corresponding historical flight arriving at the circumference L of reference circle 200. Some points in the figure may be far from the circumference L, which may be due to missing ADS-B data, etc., and these far points can be ignored when dividing the sector.

[0059] from Figure 2 As can be seen, some points are densely distributed between the regions [275°, 350°], and others are densely distributed between the regions [50°, 220°]. Therefore, we can initially define the region [275°, 350°] as one sector and the region [50°, 220°] as another sector. Assuming the threshold angle is 80°, the angle range of region [275°, 350°] (75°) is no greater than the threshold angle of 80°, so no further division is needed. The angle range of region [50°, 220°] (170°) is greater than the threshold angle of 80°, so further division is required. For example, it can be equally divided into three sectors: [50°, 106.7°], [106.7°, 163.3°], and [163.3°, 220°], with the angle range of each sector not exceeding the threshold angle of 80°.

[0060] Next, for the region [220°, 275°] where there is no point distribution, it can be split into two sub-regions [220°, 247.5°] and [247.5°, 275°]. Then, the sub-region [247.5°, 275°] is merged into its adjacent sector [275°, 350°], and the sub-region [220°, 247.5°] is merged into the expected adjacent sector [163.3°, 220°]. Similarly, for a sparsely distributed region [350°, 360°; 0°, 50°] (where 360° and 0° overlap), it can be split into two sub-regions [350°, 360°; 0°, 20°] and [20°, 50°]. Then, the sub-region [350°, 360°; 0°, 20°] is merged into its adjacent sector [275°, 350°], and the sub-region [20°, 50°] is merged into its expected adjacent sector [50°, 106.7°].

[0061] After the above process, the reference circle 200 can be divided into four sectors: sector A [20°, 106.7°], sector B [106.7°, 163.3°], sector C [163.3°, 247.5°], and sector D [247.5°, 360°; 0°, 20°].

[0062] In step S106, the reference landing time of the target flight at the current moment is determined based on the radius of the reference circle, the current distance, and the landing time of the latest one or more historical flights that landed at the destination airport via the sector determined in step S104 before the current moment, from the time of arrival at the circumference of the reference circle to landing at the destination airport.

[0063] In some embodiments, the reference landing time of the target flight at the current moment is determined as the product of the ratio of the current distance to the radius of the reference circle and the landing time of the most recent historical flight from its arrival at the circumference of the reference circle to its landing at the destination airport. That is, the reference landing time of the target flight at the current moment can be obtained by multiplying the landing time of the most recent historical flight belonging to the same sector as the target flight by the ratio of the current distance between the target flight and the destination airport to the radius of the reference circle. For example, assuming the radius of the reference circle is 400 kilometers and the landing time of the most recent historical flight from its arrival at the circumference of the reference circle to its landing at the destination airport is 30 minutes, then when the current distance between the target flight and the destination airport at the current moment is 300 kilometers, the formula for calculating the reference landing time t of the target flight at the current moment is: (minutes). Since the latest historical flight and the target flight entered the same sector at the start of their descent, they can be considered to have started their descent from similar directions and therefore are likely to have similar descent durations. In this case, the reference descent duration determined for the target flight based on the descent duration of the latest historical flight is of great significance for predicting the landing time of the target flight. Considering that flights that are closer in time are more likely to be affected by factors such as en-route weather and air traffic control, the reference descent duration can be determined directly based on the descent duration of the latest historical flight. In some embodiments, the reference descent duration can also be determined based on the average descent duration of the most recent multiple historical flights (e.g., but not limited to, the most recent two or three, etc.) to reduce the impact of outlier data. In some embodiments, obvious outliers in the descent durations of the most recent multiple historical flights can also be removed, and the average descent duration after removing outliers can be used to determine the reference descent duration.

[0064] In step S108, flight data and reference landing time are input into the machine learning model, which is then trained to predict landing time.

[0065] In some embodiments, the machine learning model includes an input module, a first prediction model, a second prediction model, and an output module. The input module is configured to receive flight data and a reference landing time. The first prediction model is coupled to the input module and configured to generate a first predicted value based on the flight data and the reference landing time. The second prediction model is coupled to the input module and configured to generate a second predicted value based on the flight data and the reference landing time. The output module is coupled to the first and second prediction models and configured to determine the predicted landing time based on the first and second predicted values. The first prediction model may differ from the second prediction model, thus configuring the machine learning model as a multi-model architecture. Integrating the predictions from multiple prediction models can improve the model's prediction accuracy. For example, different destination airports may have their own data patterns, resulting in different fits to different models. By adjusting the integration method of the multi-model predictions, better prediction of flight landing times at different airports can be achieved.

[0066] Figure 3 A schematic block diagram of a machine learning model 300 for predicting flight landing times according to some embodiments of the present disclosure is shown.

[0067] like Figure 3 As shown, the machine learning model 300 includes an input module 302, an XGBoost model 304, a decision tree model 306, and an output module 308. The input module 302 is configured to receive flight data and a reference landing duration. The XGBoost model 304 is coupled to the input module 302 and configured to generate a first predicted value based on the flight data and the reference landing duration. The decision tree model 306 is coupled to the input module 302 and configured to generate a second predicted value based on the flight data and the reference landing duration. The output module 308 is coupled to the XGBoost model 304 and the decision tree model 306 and configured to determine the predicted landing duration based on the first and second predicted values.

[0068] Decision trees are models that use a tree-like data structure to display decision rules and classification results, offering good interpretability and ease of use. XGBoost (eXtreme Gradient Boosting) is an efficient machine learning algorithm, an improved implementation of Gradient Boosting Decision Trees (GBDT), which builds a predictive model by integrating multiple weak learners (usually decision trees). While decision trees may excel at capturing some local, simple linear boundaries, XGBoost is better at modeling complex nonlinear relationships. By integrating the predictions of XGBoost model 304 and decision tree model 306, predictive performance and robustness can be improved, overfitting risk can be reduced, different patterns can be captured, and bias can be decreased.

[0069] For example, before practical application, the XGBoost model 304 and decision tree model 306 need to have their parameters optimized. In some embodiments, historical flight data and reference landing times can be used as samples, with the actual landing times of historical flights as labels, to train the XGBoost model 304 and decision tree model 306 respectively. During the hyperparameter tuning process, Bayesian optimization can first be used to process a large number of hyperparameters to determine the range of optimal solutions. Based on this, a grid search (GridSearchCV) method can be used to traverse the search grid formed by a small set of hyperparameters to find the optimal hyperparameters for each model. Cross-validation can also be used to improve the model's generalization ability and prevent overfitting, thereby optimizing the hyperparameters. After obtaining the optimized hyperparameters, they can be substituted into the corresponding XGBoost model 304 and decision tree model 306 to complete the model construction. To improve prediction speed, the completed XGBoost model 304 and decision tree model 306 can be saved separately, so that the built machine learning model can be directly called when predicting flight landing time, without having to rebuild the model for each prediction. This can speed up the operation and improve efficiency under high-density data.

[0070] In some embodiments, the prediction results of the XGBoost and decision tree dual-model can be integrated. For example, output module 308 can be configured to perform the following calculation process:

[0071] ,

[0072] Among them, Y pre It is the predicted landing time, λ is the bias, and XGB pre The first prediction is generated by XGBoost model 304, and Tree pre It is the second predicted value generated by decision tree model 306.

[0073] In some embodiments, the bias λ is adjusted depending on the destination airport. In some embodiments, the bias λ can be adjusted according to the prediction accuracy of the dual models applied to different airports to ensure the robustness of the prediction results. In addition, the dual-model setup can also provide comparison and reference of prediction results, improving the accuracy of predictions.

[0074] In some embodiments, the bias λ can be used as a hyperparameter of the machine learning model 300. For example, the bias λ can be determined by jointly training the XGBoost model 304 and the decision tree model 306 using historical flight data associated with the destination airport. Before joint training, the XGBoost model 304 and the decision tree model 306 may have been trained independently, for example, through the aforementioned process.

[0075] To enable dynamic updates to the machine learning model, in some embodiments, a second version of the machine learning model can be trained while simultaneously using a first version to predict the landing duration of a target flight. Upon completion of training of the second version, it can be used to replace the first version in predicting the landing duration. This locking mechanism allows for parallel execution of model training and application. The model can be automatically and periodically updated whenever a new version becomes available.

[0076] In step S110, the predicted landing time of the target flight is determined based on the current time and the predicted landing duration output by the machine learning model.

[0077] For example, after obtaining the predicted landing time output by the machine learning model, the predicted landing time of the target flight can be obtained by adding the predicted landing time to the current time.

[0078] In some embodiments, method 100 may be executed at regular intervals during the process of the target flight from its arrival at the circumference of the reference circle to its landing at the destination airport to update the predicted landing time of the target flight. For example, method 100 may be executed every 5 minutes to improve the accuracy of the prediction of the landing time of the target flight as it approaches the destination airport.

[0079] Figure 4 A schematic block diagram of an apparatus 400 for predicting flight landing times according to some embodiments of the present disclosure is shown.

[0080] like Figure 4 As shown, the prediction device 400 includes an acquisition module 402, a determination module 404, and a prediction module 406. The acquisition module 402 is configured to acquire flight data of the target flight. The flight data includes the current distance between the target flight and the destination airport at the current time. The determination module 404 is configured to determine the sector of the reference circle into which the target flight falls, based on the position of the target flight reaching the circumference of the reference circle centered on the destination airport. The sectors of the reference circle are divided based on the distribution of historical flights' arrival positions along the circumference of the reference circle. The determination module 404 is also configured to determine the reference landing time of the target flight at the current time, based on the radius of the reference circle, the current distance, and the landing time experienced by one or more recent historical flights that landed at the destination airport via the determined sectors before the current time, from reaching the circumference of the reference circle to landing at the destination airport. The prediction module 406 is configured to input the flight data and the reference landing time into a machine learning model. The machine learning model is trained to predict the landing time. The prediction module 406 is also configured to determine the predicted landing time of the target flight based on the current time and the predicted landing duration output by the machine learning model.

[0081] Various embodiments of the apparatus 400 can be similarly referred to the various embodiments described above with respect to method 100, and will not be repeated here.

[0082] The method and apparatus for predicting flight landing times provided in this disclosure can accurately predict flight landing times and update the prediction results in real time, which is conducive to the rational allocation of airport support resources and improves airport operational efficiency.

[0083] Figure 5 An electronic device 500 according to some embodiments of the present disclosure is shown. For example... Figure 5 As shown, electronic device 500 includes one or more processors 502 and a memory 504 storing computer-executable instructions that, when executed by one or more processors 502, cause the processors 502 to perform the method 100 for predicting flight landing times according to any of the foregoing embodiments. Processor 502 may be, for example, a central processing unit (CPU) of device 500. Processor 502 may be any type of general-purpose processor or may be a processor specifically designed for predicting flight landing times, such as an application-specific integrated circuit (“ASIC”). Memory 504 may be coupled to processor 502 and may include various computer-readable media accessible by processor 502. In various embodiments, memory 504 described herein may include volatile and non-volatile media, removable and non-removable media. For example, memory 504 may include any combination of random access memory (“RAM”), dynamic RAM (“DRAM”), static RAM (“SRAM”), read-only memory (“ROM”), flash memory, cache memory, and / or any other type of non-transient computer-readable media. The memory 504 may store instructions that, when executed by the processor 502, cause the processor 502 to execute the method 100 according to any of the foregoing embodiments of the present disclosure.

[0084] This disclosure also provides a computer-readable storage medium having computer program instructions stored thereon, which, when executed by a computer, cause the computer to perform a method for predicting flight landing times according to any of the foregoing embodiments of this disclosure.

[0085] This disclosure also provides a computer program product that may include instructions that, when executed by a processor, implement the method for predicting flight landing times according to any of the foregoing embodiments of this disclosure. The instructions may be any set of instructions that will be executed directly by a processor, such as machine code, or any set of instructions that will be executed indirectly, such as a script. The instructions may be stored in an object code format for direct processing by a processor, or stored in any other computer language, including scripts or sets of independent source code modules that are interpreted on demand or compiled in advance.

[0086] Figure 6A schematic block diagram of a computer system 600 on which embodiments of the present disclosure may be implemented is shown. The computer system 600 includes a bus 602 or other communication mechanism for transmitting information, and a processing means 604 coupled to the bus 602 for processing information. The computer system 600 also includes a memory 606 coupled to the bus 602 for storing instructions to be executed by the processing means 604; the memory 606 may be random access memory (RAM) or other dynamic storage device. The memory 606 may also be used to store temporary variables or other intermediate information during the execution of instructions to be executed by the processing means 604. The computer system 600 also includes a read-only memory (ROM) 608 or other static storage device coupled to the bus 602 for storing static information and instructions for the processing means 604. A storage device 610, such as a magnetic disk or optical disk, is provided and coupled to the bus 602 for storing information and instructions. Computer system 600 may be coupled via bus 602 to output device 612 for providing output to a user, such as, but not limited to, a display (e.g., a cathode ray tube (CRT) or liquid crystal display (LCD)), speakers, etc. Input device 614, such as a keyboard, mouse, microphone, gaze tracker, etc., is coupled to bus 602 for transmitting information and command selections to processing device 604. In some implementations, input device 614 and output device 612 may be integrated together, such as a touch display panel. Computer system 600 may perform embodiments of this disclosure. Consistent with some implementations of this disclosure, results are provided by computer system 600 in response to processing device 604 executing one or more sequences of one or more instructions contained in memory 606. Such instructions may be read into memory 606 from another computer-readable medium, such as storage device 610. Execution of the sequence of instructions contained in memory 606 causes processing device 604 to perform the methods described herein. Alternatively, hardwired circuitry may be used in place of or in combination with software instructions to implement the teachings. Therefore, implementations of this disclosure are not limited to any particular combination of hardware circuitry and software. In various embodiments, computer system 600 may be connected across a network to one or more other computer systems, such as computer system 600, to form a networked system via network interface 616. This network may include a private network or a public network such as the Internet. In a networked system, one or more computer systems may store data and supply data to other computer systems. As used herein, the term "computer-readable medium" refers to any medium that participates in providing instructions to processing device 604 for execution. Such media may take many forms, including but not limited to non-volatile media, volatile media, and transmission media. Non-volatile media include, for example, optical discs or magnetic disks such as storage device 610.Volatile media include dynamic memory such as memory 606. Transmission media include coaxial cable, copper wire, and optical fiber, including wiring containing bus 602. Common forms of computer-readable media or computer program products include, for example, floppy disks, flexible disks, hard disks, magnetic tape, or any other magnetic media, CD-ROMs, digital video discs (DVDs), Blu-ray discs, any other optical media, thumb drives, memory cards, RAM, PROMs and EPROMs, fast EPROMs, any other memory chips or cartridges, or any other tangible media from which a computer can read. Various forms of computer-readable media may be involved when carrying one or more sequences of one or more instructions to processing device 604 for execution. For example, instructions may initially be carried on a disk of a remote computer. The remote computer may load the instructions into its dynamic memory and transmit the instructions over a telephone line using a modem. A modem local to computer system 600 may receive data over a telephone line and convert the data into an infrared signal using an infrared transmitter. An infrared detector coupled to bus 602 may receive the data carried in the infrared signal and place the data on bus 602. Bus 602 carries data to memory 606, and processing device 604 retrieves instructions from memory 606 and executes the instructions. Optionally, instructions received by memory 606 may be stored on storage device 610 before or after execution by processing device 604.

[0087] According to various embodiments, instructions configured to be executed by a processing device to perform a method are stored on a computer-readable medium. The computer-readable medium may be a device for storing digital information. For example, a computer-readable medium includes a compact disc read-only memory (CD-ROM) as known in the art for storing software. The computer-readable medium is accessed by a processor adapted to execute the instructions configured to be executed.

[0088] The foregoing has described one or more exemplary embodiments of this disclosure. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0089] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or physical entities, or by products with certain functions. A typical implementation device is a server system. Of course, this disclosure does not exclude the possibility that, with the future development of computer technology, the computer implementing the functions of the above embodiments can be, for example, a personal computer, a laptop computer, an in-vehicle human-machine interaction device, a cellular phone, a camera phone, a smartphone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or any combination of these devices.

[0090] While one or more embodiments of this disclosure provide the method operation steps as described in the embodiments or flowcharts, more or fewer operation steps may be included based on conventional or non-inventive means. The order of steps listed in the embodiments is merely one possible execution order among many and does not represent the only execution order. In actual device or terminal product execution, the methods shown in the embodiments or drawings can be executed sequentially or in parallel (e.g., in a parallel processor or multi-threaded processing environment, or even a distributed data processing environment).

[0091] The terms "comprising," "including," or any other variations thereof are intended to cover a non-exclusive inclusion, such that a process, method, product, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, product, or apparatus. Without further limitation, the presence of other identical or equivalent elements in the process, method, product, or apparatus that includes said elements is not excluded. For example, the use of terms such as "first" or "second" to denote names does not indicate any particular order.

[0092] For ease of description, the above devices are described in terms of function, divided into various modules. Of course, when implementing one or more embodiments of this disclosure, the functions of each module can be implemented in one or more software and / or hardware, or a module that performs the same function can be implemented by a combination of multiple sub-modules or sub-units. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0093] This disclosure is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in one or more blocks of the flowchart illustrations and / or one or more blocks of the block diagrams.

[0094] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means that implement the functions specified in one or more flowcharts and / or one or more block diagrams. These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, such that the instructions, which execute on the computer or other programmable apparatus, provide steps for implementing the functions specified in one or more flowcharts and / or one or more block diagrams.

[0095] Those skilled in the art will understand that one or more embodiments of this disclosure may take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, one or more embodiments of this disclosure may take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0096] One or more embodiments of this disclosure can be described in the general context of computer-executable instructions, such as program modules, that are executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform a particular task or implement a particular abstract data type. One or more embodiments of this disclosure can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In a distributed computing environment, program modules can reside in local and remote computer storage media, including storage devices.

[0097] The same or similar parts between the various embodiments of this disclosure can be referred to mutually, and each embodiment focuses on describing the differences from other embodiments. In particular, for the apparatus embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and relevant parts can be referred to the description of the method embodiments. In the description of this disclosure, the descriptions of terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., mean that the specific feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of this disclosure. In this disclosure, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in a suitable manner in any one or more embodiments or examples. Furthermore, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this disclosure and the features of the different embodiments or examples.

[0098] Additionally, when used in this disclosure, the terms “here,” “above,” “below,” “below,” “in the following,” “overall,” and similar terms should refer to the entirety of this disclosure and not any particular part thereof. Furthermore, unless expressly stated otherwise or otherwise understood in the context in which they are used, conditional language used herein, such as “may,” “possibly,” “for example,” “like,” etc., is generally intended to express that certain embodiments include, while other embodiments do not, certain features, elements, and / or states. Therefore, such conditional language is not generally intended to imply that one or more embodiments require features, elements, and / or states in any way, or whether such features, elements, and / or states are included or performed in any particular embodiment.

[0099] The above description is merely an embodiment of one or more embodiments of this disclosure and is not intended to limit the scope of the one or more embodiments of this disclosure. Various modifications and variations can be made to the one or more embodiments of this disclosure by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this disclosure should be included within the scope of the claims.

Claims

1. A method for predicting flight arrival times, comprising: Obtain flight data for the target flight, including the current distance between the target flight and the destination airport at the current time; Based on the position of the target flight arriving at the circumference of a reference circle centered on the destination airport, the sector in which the target flight falls within the reference circle is determined, wherein the sector of the reference circle is divided based on the distribution of the positions of historical flights arriving at the circumference of the reference circle along the circumference of the reference circle. The reference landing time of the target flight at the current time is determined based on the radius of the reference circle, the current distance, and the landing time of the most recent one or more historical flights that landed at the destination airport via the determined sector before the current time, from the time of arrival at the circumference of the reference circle to the time of landing at the destination airport. The flight data and the reference landing duration are input into a machine learning model, which is then trained to predict landing duration; and The predicted landing time of the target flight is determined based on the current time and the predicted landing duration output by the machine learning model.

2. The method according to claim 1, wherein, The radius of the reference circle is determined based on a first empirical value of the distance between the flight to be landed at the destination airport and the destination airport at the start of the landing procedure, or based on a second empirical value of the flight time from the start of the landing procedure to landing at the destination airport.

3. The method according to claim 2, wherein, The first experience value is between 300 km and 500 km, and the second experience value is between 30 minutes and 40 minutes.

4. The method according to claim 1, wherein, The sectors that divide the reference circle include: The distribution of the positions of historical flights arriving at the circumference of the reference circle along the circumference of the reference circle is represented as a scatter plot. Adjacent regions with a scatter point density greater than a threshold density are assigned to the same sector. Sectors whose angle range exceeds a threshold angle are divided into multiple sectors; and The regions with a scatter density not greater than the threshold density are assigned to the sectors adjacent to that region.

5. The method according to claim 4, wherein, Assigning regions with a scatter density not greater than the threshold density to adjacent sectors includes: The region with a scatter density not greater than the threshold density is divided into two sub-regions, and each sub-region is assigned to a sector adjacent to that sub-region.

6. The method according to claim 4, wherein, Sectors whose angle range is greater than a threshold angle are divided into multiple sectors, including: The sector whose angle range is greater than a threshold angle is divided into multiple sectors, such that the angle range of each of the multiple sectors is not greater than the threshold angle.

7. The method according to claim 1, wherein, The reference landing time of the target flight at the current moment is determined as the product of the ratio of the current distance to the radius of the reference circle and the average landing time experienced by the latest one or more historical flights from the circumference of the reference circle to landing at the destination airport.

8. The method according to claim 1, wherein, The flight data also includes the landing direction of the target flight relative to the destination airport at the current time.

9. The method according to claim 8, wherein: In response to the current distance being greater than a threshold distance, the landing direction of the target flight relative to the destination airport at the current time is determined to be the landing direction of the flight that recently landed at the destination airport; or In response to the current distance not being greater than the threshold distance, the landing direction of the target flight relative to the destination airport at the current time is determined based on the change of the target flight's current position at the current time relative to the past position of the target flight at a previous time.

10. The method according to claim 1, wherein, The current distance is determined as the maximum circular distance on Earth between the current position of the target flight and the position of the destination airport at the current moment.

11. The method according to claim 1 or 8, wherein, The flight data also includes the target flight's departure time, planned flight duration, aircraft type, and one or more of the target flight's latitude and longitude coordinates, speed, and heading at the current time.

12. The method according to claim 11, wherein, The departure time of the target flight is represented as a combination of a day of the week and an hour of a 24-hour day when it is input into the machine learning model.

13. The method according to claim 1, wherein, The machine learning model includes an input module, a first prediction model, a second prediction model, and an output module. The input module is configured to receive the flight data and the reference landing time. The first prediction model is coupled to the input module and configured to generate a first prediction value based on the flight data and the reference landing time. The second prediction model is coupled to the input module and configured to generate a second prediction value based on the flight data and the reference landing duration. The output module is coupled to the first prediction model and the second prediction model and is configured to determine the predicted landing duration based on the first prediction value and the second prediction value. The first prediction model is different from the second prediction model.

14. The method according to claim 13, wherein, The first prediction model is an XGBoost model, and the second prediction model is a decision tree model.

15. The method according to claim 14, wherein, The output module is configured to perform the following calculation process: ， Among them, Y pre It predicts the landing time. It's the bias value, XGB pre The first predicted value is generated by the XGBoost model, and Tree pre It is the second predicted value generated by the decision tree model.

16. The method according to claim 15, wherein, The bias It depends on the destination airport.

17. The method according to claim 1, wherein, While using a first version of the machine learning model to predict the landing time of the target flight, a second version of the machine learning model is trained, and wherein, in response to the completion of training of the second version of the machine learning model, the second version of the machine learning model is used to replace the first version to predict the landing time.

18. The method according to claim 17, wherein, After the target flight lands at the destination airport, the actual landing time of the target flight is used as a label to train the second version of the machine learning model.

19. The method according to claim 1, wherein, During the process of the target flight from arriving at the circumference of the reference circle to landing at the destination airport, the method for predicting the landing time of the flight according to any one of claims 1 to 18 is executed at regular intervals to update the predicted landing time of the target flight.

20. An apparatus for predicting flight landing time, comprising: The acquisition module is configured to acquire flight data of the target flight, the flight data including the current distance between the target flight and the destination airport at the current time; The module is defined as follows: Based on the position of the target flight upon arrival at the circumference of a reference circle centered on the destination airport, the sector within which the target flight falls on the reference circle is determined. The sectors of the reference circle are divided based on the distribution of historical flight arrival positions along the circumference of the reference circle. The reference landing time of the target flight at the current time is determined based on the radius of the reference circle, the current distance, and the landing time of the most recent one or more historical flights that landed at the destination airport via the determined sector before the current time, from the time of arrival at the circumference of the reference circle to the time of landing at the destination airport. as well as The prediction module is configured as follows: The flight data and the reference landing duration are input into a machine learning model, which is then trained to predict landing duration. The predicted landing time of the target flight is determined based on the current time and the predicted landing duration output by the machine learning model.

21. An electronic device, comprising: processor; as well as A memory storing computer-executable instructions, which, when executed by the processor, cause the processor to perform the method for predicting flight landing times according to any one of claims 1 to 19.

22. A computer-readable storage medium having stored thereon computer program instructions, which, when executed by a computer, cause the computer to perform a method for predicting flight landing times according to any one of claims 1 to 19.

23. A computer program product comprising instructions that, when executed by a processor, implement the method for predicting flight landing times according to any one of claims 1 to 19.

Images