Data processing method and device for obtaining departure flight taxiing time and medium

Abstract

The invention relates to the technical field of data processing, in particular to a data processing method, device and medium for acquiring the taxiing time of departure flights, the method predicts the taxiing time by fusing taxiing trajectory data and taxiing associated data, can cover influence factors in the aspects of geographic space, time dynamics, environment interaction and the like, and can improve the taxiing time of departure flights. And the predicted sliding time is more accurate. Besides, by combining the trajectory data and the sliding associated data, spatial-temporal feature extraction is performed on the local path, and an attention mechanism is introduced to dynamically identify the key local path, so that the prediction precision of the sliding time can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of aviation data processing, and particularly to a data processing method, device, and medium for obtaining the taxiing time of departing flights. Background Art

[0002] The arrival and departure processes of flights are highly dynamic, random, and uncertain. Therefore, it is particularly important to accurately and scientifically predict and analyze the ground taxiing time. For example, accurately predicting the departure taxiing time can help airports optimize the flight push-back sequence, thereby improving airport operation efficiency and flight punctuality rate. Currently, the research methods for flight taxiing time mainly focus on the following two categories. One is the prediction based on traditional simulation methods. These methods rely on factors such as airport site characteristics and flight dynamics for modeling. Although they can consider multiple factors, the complexity and computational volume of the model are relatively large, and there are certain limitations in dealing with non-linear complex relationships. The other is the prediction method based on machine learning, such as using methods like random forest, support vector machine, neural network, etc. to improve the prediction accuracy of taxiing time. Compared with simulation methods, machine learning methods can better handle non-linear relationships and adapt to complex dynamic environments. However, existing machine learning methods mainly focus on the modeling of macroscopic features, such as taxiing distance and airport traffic flow, etc., ignoring the fine-grained spatio-temporal features of the taxiing trajectory, and these features are crucial for accurate prediction in complex environments. In addition, the existing models have poor robustness under the interaction of multiple factors. Especially when facing complex operating environments and dynamic changes, they often cannot maintain stable prediction accuracy. Traditional machine learning models usually have difficulty effectively identifying and extracting key local spatio-temporal features during the taxiing process, resulting in weak generalization ability of the model in complex environments. Therefore, how to extract spatio-temporal features from fine-grained taxiing trajectory data and simultaneously consider the influence of dynamic external factors has become the key to improving the prediction accuracy and precision of departure taxiing time. Summary of the Invention

[0003] In view of the above technical problems, the technical solution adopted by the present invention is as follows: According to the first aspect of the present invention, there is provided a data processing method for obtaining the taxiing time of departing flights, the method comprising the following steps: S100, obtaining a dataset to be processed; the dataset to be processed includes the simulated taxiing trajectory data and taxiing correlation data of the target flight.

[0004] S200, respectively performing encoding processing on the simulated taxiing trajectory data and the taxiing correlation data to obtain the feature vector of the simulated taxiing trajectory data and the feature vector of the taxiing correlation data.

[0005] S300, perform a one-dimensional convolution operation on the feature vectors of the simulated taxiing trajectory data using a convolution kernel with a width of k to extract m local features, constituting the spatial features of m local paths; m = n - k + 1, where n is the number of position points in the simulated taxiing trajectory data.

[0006] S400, obtain the geometric distance of each local path, perform feature concatenation on the spatial feature of each local path and the corresponding geometric distance, and use the concatenation result as the enhanced spatial feature of the local path.

[0007] S500, perform fusion processing on the enhanced spatial feature of each local path and the feature vector of the taxiing association data to obtain the multi-modal feature of each local path.

[0008] S600, extract the temporal dependence relationship in the multi-modal feature of each local path to obtain the spatio-temporal feature of each local path.

[0009] S700, use the attention mechanism to perform weighted processing on all spatio-temporal features to obtain the global spatio-temporal feature.

[0010] S800, process the global spatio-temporal feature to obtain the predicted value of the taxiing time of the target flight.

[0011] According to a second aspect of the present invention, there is provided an electronic device, including a processor and a memory; the processor is configured to execute the steps of the method according to the first aspect of the present invention by calling a program or instruction stored in the memory.

[0012] According to a third aspect of the present invention, there is provided a computer-readable storage medium storing a program or instruction, and the program or instruction causes a computer to execute the steps of the method according to the first aspect of the present invention.

[0013] The data processing method for obtaining the taxiing time of a departing flight provided by the embodiments of the present invention predicts the taxiing time by fusing taxiing trajectory data and taxiing association data, and can cover influencing factors in aspects such as geographical space, time dynamics, and environmental interaction, making the predicted taxiing time more accurate. In addition, by combining trajectory data and taxiing association data, spatio-temporal features of local paths are extracted, and an attention mechanism is introduced to dynamically identify key local paths, which can improve the prediction accuracy of the departing taxiing time and provide data support for the airport to optimize the flight push sequence and resource allocation decision.

[0014] It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the present invention, nor is it used to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. Description of the Drawings

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0016] Figure 1 It is a flowchart of a data processing method for obtaining the taxiing time of departing flights provided by an embodiment of the present invention; Figure 2 It is a structural block diagram of a taxiing time prediction model provided by an embodiment of the present invention; Figure 3 It is a schematic diagram showing the effect of the taxiing time prediction model provided by an embodiment of the present invention. Specific Embodiments

[0017] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs. The terms used in the description of the present invention in this specification are only for the purpose of describing specific embodiments, and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0019] It should be noted that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe the steps as sequential processes, many of the steps can be implemented in parallel, concurrently, or simultaneously. In addition, the order of the steps can be rearranged. The process can be terminated when its operation is completed, but there can also be additional steps not included in the drawings. The process can correspond to a method, function, procedure, subroutine, subprogram, etc.

[0020] An embodiment of the present invention provides a data processing method for obtaining the taxiing time of departing flights, as Figure 1 shown, the method includes the following steps: S100, obtain a dataset to be processed.

[0021] In an embodiment of the present invention, the dataset to be processed includes the simulated taxiing trajectory data of the target flight and taxiing correlation data. The taxiing correlation data includes flight attribute information, operating period identification value, environmental data, and flight interaction data.

[0022] In an embodiment of the present invention, the simulated taxiing trajectory data in the dataset to be processed is the taxiing trajectory data obtained by simulation, and the specific simulation method can be the prior art.

[0023] In an embodiment of the present invention, the target flight of the dataset to be processed is the departure flight corresponding to the dataset currently to be processed, that is, the dataset to be processed is the dataset corresponding to the target flight, and is the data subject corresponding to the dataset. In an embodiment of the present invention, one departure flight corresponds to one aircraft. In an embodiment of the present invention, the flight attribute information may include: minute timestamp based on the planned departure time, aircraft type, aircraft ID, taxiing distance, etc. The taxiing distance refers to the distance between the specified starting position and the specified ending position. In an embodiment of the present invention, the specified starting position can be a parking position or a boarding gate, and the specified ending position is the take-off departure point of the aircraft.

[0024] In an embodiment of the present invention, the minute timestamp based on the planned departure time refers to the minute timestamp corresponding to the planned departure time. For example, if the format of the planned departure time is "2025-5-8 8:42", the minute timestamp based on the planned departure time is: 522 = (8×60 + 42). The aircraft ID refers to the unique identifier of the aircraft, and can be, for example, the aircraft registration number.

[0025] In an embodiment of the present invention, the operating period identification value refers to the time period identification value when the flight departs and taxis. The traffic flow of the airport changes with time, and the flight departure and taxiing time is different in different time periods.

[0026] In an embodiment of the present invention, the operating period can be determined according to the average number of flights per hour. In a schematic embodiment, the present invention divides the whole day into three types of operating periods by actually referring to the actual operation of an airport according to the average number of flights per hour, namely: the first operating period to the third operating period. Among them, the time period of the first operating period is [0:00 - 8:00], the time period of the second operating period is [8:00 - 10:00, 22:00 - 24:00], and the time period of the third operating period is [10:00 - 22:00]. Different operating periods can be represented by different operating period identification values.

[0027] In an embodiment of the present invention, the environmental data can be the weather data of the airport, and can include visibility, wind speed, abnormal weather flag, etc. Among them, the abnormal weather flag can include a thunderstorm flag representing thunderstorm weather and a precipitation flag representing whether there is precipitation, etc.

[0028] In the embodiments of the present invention, the flight interaction data corresponding to the target flight is the number of flights moving on the airport surface during the taxiing process of the target flight, that is, the number of flights in the taxiing state on the airport surface. Specifically, it may include the number of flights in the taxiing state on the runway, taxiway, and apron during the taxiing process of the target flight, including the number of inbound flights and outbound flights. In a busy airport, inbound and outbound flights may share some taxiing routes, which may lead to traffic congestion and thus extend the taxiing time of outbound flights. The impact of interaction between flights on taxiing is a complex process. Especially when analyzing the taxiing time of outbound flights, the impact of interactions between different types of flights needs to be fully considered.

[0029] In the embodiments of the present invention, during the taxiing process of the target flight, it may include the number of flights in 12 taxiing states, that is, the flight interaction data may include 12 types of interaction data, and one type of interaction data corresponds to the number of flights in one taxiing state. In a schematic embodiment, the 12 types of interaction data may specifically include the number of five inbound flights during the taxiing process of the target flight and the number of seven outbound flights during the taxiing process of the target flight.

[0030] Furthermore, in the embodiments of the present invention, the number of inbound flights during the taxiing process of the target flight may include the following five types of flight numbers: (1) the number of inbound flights whose landing time is earlier than the wheel chock release time of the target flight and whose wheel chock time is during the taxiing process of the target flight; (2) the number of inbound flights whose landing time is earlier than the wheel chock release time of the target flight and whose wheel chock time is later than the takeoff and lift-off time of the target flight; (3) the number of inbound flights whose landing time is later than the wheel chock release time of the target flight and whose wheel chock time is earlier than the takeoff and lift-off time of the target flight; (4) the number of inbound flights whose landing time is during the taxiing process of the target flight and whose wheel chock time is later than the takeoff and lift-off time of the target flight; (5) the number of inbound flights in the inbound taxiing state during the taxiing process of the target flight, specifically, the number of inbound flights whose landing time is earlier than the takeoff and lift-off time of the target flight and whose wheel chock time is later than the wheel chock release time of the target flight.

[0031] In an embodiment of the present invention, the number of departing flights during the taxiing process of a target flight may include the following seven types of flight numbers: (1) the number of departing flights whose chocks-off time is earlier than that of the target flight and whose take-off time is during the taxiing process of the target flight; (2) the number of departing flights whose chocks-off time is earlier than that of the target flight and whose take-off time is later than the take-off time of the target flight; (3) the number of departing flights whose chocks-off time is later than that of the target flight and whose take-off time is earlier than the take-off time of the target flight; (4) the number of departing flights whose chocks-off time is during the taxiing process of the target flight and whose take-off time is later than the take-off and leaving-the-ground time of the target flight; (5) during the departure process of the target flight, the number of flights in the departure taxiing, specifically, the number of departing flights whose chocks-off time is earlier than the take-off and leaving-the-ground time of the target flight and whose take-off and leaving-the-ground time is later than the chocks-off time of the target flight; (6) at the chocks-off time of the target flight's departure, the number of flights in the departure taxiing process, specifically, the number of flights whose chocks-off time is earlier than the chocks-off time of the target flight and whose take-off and leaving-the-ground time is later than the chocks-off time of the target flight; (7) during the departure taxiing process of the target flight, the number of flights in the take-off and leaving-the-ground state, specifically, the number of flights whose take-off and leaving-the-ground time is between the chocks-off time and the take-off and leaving-the-ground time of the target flight.

[0032] Those skilled in the art should understand that the number of arriving flights and departing flights that interact with the target flight during the taxiing process can be initially screened based on the estimated times in the flight schedule and calibrated by combining real-time monitoring data (such as radar, ADS-B data).

[0033] S200. Respectively perform encoding processing on the simulated taxiing trajectory data and the taxiing correlation data to obtain the feature vector of the simulated taxiing trajectory data and the feature vector of the taxiing correlation data.

[0034] In an embodiment of the present invention, the simulated taxiing trajectory data may include the position information of n position points, and the position information may be longitude and latitude coordinates.

[0035] In an embodiment of the present invention, the feature vector of the simulated taxiing trajectory data can be obtained by performing non-linear mapping processing on the position information of each position point in the simulated taxiing trajectory data. Specifically, the feature vector of the simulated taxiing trajectory data satisfies the following conditions: FL={Local1, Local2, ……, Local u , ……, Local n}; where FL is the feature vector of the simulated taxiing trajectory data, and Local u is the mapping feature obtained by performing non-linear mapping processing on the longitude and latitude coordinates of the u-th position point, and the value of u ranges from 1 to n.

[0036] In an embodiment of the present invention, Local u satisfies the following condition: Local u =tanh(W local •concat(Lon u , Lat u ), where Lon u is the longitude coordinate of the u-th position point, Lat u is the latitude coordinate of the u-th position point, concat represents the concatenation operation, W local is a learnable weight matrix, and tanh() is the tanh activation function.

[0037] In an embodiment of the present invention, the feature vector of the taxiing correlation data is obtained through the following steps: S201, preprocess the taxiing correlation data to obtain initial feature vectors corresponding to flight attribute information, operation period identification values, environmental data, and flight interaction data.

[0038] In an embodiment of the present invention, the preprocessing of the taxiing correlation data may specifically include: (1) For discrete variables in the taxiing correlation data, such as category values of minute timestamps, aircraft types, aircraft IDs, abnormal weather flags, etc., use an embedding layer to convert the features of each category attribute into low-dimensional vectors to obtain initial feature values for each discrete variable. The specific embedding method may be prior art.

[0039] (2) For non-discrete variables in the taxiing correlation data, such as visibility, wind speed, flight interaction data, etc., perform normalization processing to obtain initial feature values corresponding to each non-discrete variable.

[0040] In an embodiment of the present invention, the initial feature value of each non-discrete variable is obtained through Z-score normalization, that is, the value obtained by dividing the difference between the value of the non-discrete variable and the corresponding mean by the corresponding standard deviation. Among them, the mean and standard deviation corresponding to each non-discrete variable are calculated based on historical data statistics, for example, obtained according to the training set.

[0041] S202, splice the initial feature vectors corresponding to the operation period identification value and the environmental data, and use the obtained splicing result as the fused environmental feature vector.

[0042] S203, based on the fused environmental feature vector, obtain a weight vector corresponding to the initial feature vector of the flight interaction data, and based on the obtained weight vector, perform weighted processing on the initial feature vector of the flight interaction data to obtain a corresponding weighted processing result as the weighted interaction feature vector.

[0043] In an embodiment of the present invention, the weighted interaction feature vector satisfies the following conditions: FW = (λ1f1, λ2f2, ……, λ r f r , ……, λ Q f Q ).

[0044] Wherein, FW is the weighted interaction feature vector, f r is the initial feature vector corresponding to the r-th interaction data, and λ r is the weight corresponding to the r-th interaction data. The value of λ r ranges from 0 to 1, the value of r ranges from 1 to Q, and Q is the number of interaction data in the aircraft interaction data. In an embodiment of the present invention, Q = 12.

[0045] In an embodiment of the present invention, λ r can be determined based on the fused environmental feature vector and f r . Further, λ r satisfies the following conditions: λ r = σ(W e ·E env + W f ·f r + b); Wherein, W e is the environmental feature weight matrix, E env is the fused environmental feature vector, W f is the weight matrix corresponding to f r , b is the bias term, σ( ) is the Sigmoid activation function, and · represents dot product.

[0046] S204. Concatenate the initial feature vector corresponding to the flight attribute information, the fused environmental feature vector, and the weighted interaction feature vector, and use the obtained concatenation result as the feature vector of the taxiing correlation data.

[0047] In an embodiment of the present invention, since the feature vector of the taxiing correlation data considers the weighted interaction feature obtained based on the environmental data, that is, the weighted interaction feature vector is obtained through the dynamic weighting mechanism of environmental perception, it can achieve scene-adaptive feature screening and fusion, improve the prediction accuracy in busy or harsh scenes, reduce the computational complexity, and enhance the cooperation efficiency with the subsequent time series modeling module.

[0048] S300. Use a convolution kernel with a width of k to perform a one-dimensional convolution operation on the feature vector of the simulated taxiing trajectory data to extract m local features and form the spatial features of m local paths.

[0049] In an embodiment of the present invention, a one-dimensional convolutional neural network can be used to perform a convolution operation on the feature vector local path of the simulated taxiing trajectory data. The purpose of using the convolutional neural network to extract features from the feature vector of the local path of the simulated taxiing trajectory data is to perform local feature extraction on the position sequence through a one-dimensional sliding window, helping the model identify turning patterns, path features, and other complex geographical laws in the path.

[0050] As is known to those skilled in the art, the number m of local paths is m = n - k + 1. In an embodiment of the present invention, each local path is composed of k consecutive position points. The spatial features corresponding to each local path satisfy the following conditions: LP i conv =σ conv (W conv ·LP i +b conv )。

[0051] Wherein, LP i conv is the spatial feature of the i-th local path obtained by the convolutional neural network, W conv is the learnable weight matrix of the convolutional neural network, LP i is the feature vector corresponding to the i-th local path, that is, the feature vector formed by the mapping features from the i-th position point to the i - k + 1-th position point, and b conv is the bias term of the convolutional neural network.

[0052] S400. Obtain the geometric distance of each local path, and perform feature splicing on the spatial feature of each local path and the corresponding geometric distance, and use the splicing result as the enhanced spatial feature of the local path.

[0053] In an embodiment of the present invention, the geometric distance of the i-th local path , where the value of i ranges from 1 to m. Among them, is the geometric distance between the g - 1-th position point and the g-th position point on the i-th local path, and the value of g ranges from 2 to k.

[0054] S500. Perform a fusion process on the enhanced spatial feature of each local path and the feature vector of the taxiing correlation data to obtain the multi-modal feature of each local path.

[0055] In an embodiment of the present invention, performing a fusion process on the enhanced spatial feature of each local path and the corresponding feature vector of the taxiing correlation data means performing feature splicing on the enhanced spatial feature of each local path and the corresponding feature vector of the taxiing correlation data.

[0056] S600 extracts the temporal dependence relationships in the multimodal features of each local path to obtain corresponding spatio-temporal features.

[0057] In an embodiment of the present invention, the temporal dependence relationships in the multimodal features of each local path can be extracted through two cascaded gated recurrent units. Specifically, the temporal dependence relationships in the multimodal features of each local path can be extracted through a first gated recurrent unit and a second gated recurrent unit connected in sequence.

[0058] In an embodiment of the present invention, the first gated recurrent unit is used to perform preliminary temporal modeling on the input features to capture short-term and long-term dependence information in the sequence, and the second gated recurrent unit is used to further model the features processed by the first gated recurrent unit to extract more complex temporal patterns for more accurately describing the variation of the taxiing time.

[0059] Those skilled in the art know that the specific structure and working principle of the gated recurrent unit can be prior art.

[0060] S700 uses an attention mechanism to weight all the spatio-temporal features to obtain global spatio-temporal features.

[0061] In an embodiment of the present invention, the global spatio-temporal features can be obtained through the following steps: S701 performs a linear transformation on the feature vector of the taxiing correlation data to obtain a corresponding query vector.

[0062] In an embodiment of the present invention, the query vector satisfies the following condition: q = σ att (W q •F c + b q ).

[0063] Wherein, q is the query vector, F c is the feature vector of the taxiing correlation data, W q is the linear transformation weight matrix, b q is the bias term. σ att ( )is a non-linear activation function. In a schematic embodiment, σ att ( )can be the tanh activation function.

[0064] S702 obtains the interaction intensity between the spatio-temporal features of each local path and the query vector.

[0065] In an embodiment of the present invention, the interaction intensity z(i) corresponding to the i-th local path satisfies the following condition: z(i) = (q T •h i ) / (dk ) 1 / 2 。

[0066] Among them, h i is the spatio-temporal feature of the i-th local path, q T is the transpose of q, and d k is the dimension of q;.

[0067] S703. Normalize the m interaction intensities to obtain the attention weights corresponding to the spatio-temporal features of each local path.

[0068] In the embodiment of the present invention, the attention weight α i corresponding to the spatio-temporal feature of the i-th local path satisfies the following conditions: , e is the natural constant, and j takes values from 1 to m.

[0069] S704. Perform weighted summation on the spatio-temporal features of all local paths and the corresponding attention weights to obtain the global spatio-temporal feature.

[0070] In the embodiment of the present invention, the global spatio-temporal feature Hatt satisfies the following conditions: 。

[0071] S800. Process the global spatio-temporal feature to obtain the predicted value of the taxiing time of the target flight.

[0072] In the embodiment of the present invention, the global spatio-temporal feature can be processed by a residual fully connected network to obtain the predicted value of the taxiing time corresponding to the current dataset to be processed. Those skilled in the art know that the specific structure and working principle of the residual fully connected network can be the prior art.

[0073] The data processing method for obtaining the taxiing time of a departing flight provided by the embodiment of the present invention predicts the taxiing time by fusing taxiing trajectory data and taxiing associated data, and can cover influencing factors such as geographical space, time dynamics, and environmental interaction, making the predicted taxiing time more accurate. In addition, by combining trajectory data and taxiing associated data, extracting spatio-temporal features of local paths, and dynamically identifying key local paths, the prediction accuracy of the taxiing time can be improved, and data support can be provided for the airport to optimize the flight departure sequence and resource allocation decision.

[0074] In the embodiment of the present invention, the data processing method for obtaining the taxiing time of a departing flight provided by the embodiment of the present invention can be implemented based on a trained taxiing time prediction model.

[0075] Such as Figure 2As shown, the taxiing time prediction model may include: a feature encoding module 1, a spatial feature extraction module 2, a feature enhancement module 3, a multi-modal feature fusion module 4, a temporal modeling module 5, an attention weighting module 6, and a prediction output module 7. Among them, the spatial feature extraction module 2, the feature enhancement module 3, the multi-modal feature fusion module 4, the temporal modeling module 5, the attention weighting module 6, and the prediction output module 7 are connected in sequence, and the data processing module 1 is respectively connected to the spatial feature extraction module, the multi-modal feature fusion module, and the attention weighting module.

[0076] In an embodiment of the present invention, the feature encoding module 1 is used to obtain the feature vector of the taxiing trajectory data and the feature vector of the taxiing association data.

[0077] The spatial feature extraction module 2 is used to perform a convolution operation on the feature vector corresponding to the taxiing trajectory data to obtain the spatial feature corresponding to each local path. The spatial feature extraction module can be a one-dimensional convolutional neural network.

[0078] The feature enhancement module 3 is used to perform feature splicing on the spatial feature corresponding to each local path and the corresponding geometric distance, and use the splicing result as the enhanced spatial feature of the local path.

[0079] The multi-modal feature fusion module 4 is used to perform fusion processing on the enhanced spatial feature of each local path and the feature vector of the taxiing association data to obtain the multi-modal feature of each local path.

[0080] The temporal modeling module 5 is used to extract the temporal dependence relationship in the multi-modal feature of each local path to obtain the spatio-temporal feature of each local path. The temporal modeling module 5 may include a first gated recurrent unit and a second gated recurrent unit.

[0081] The attention weighting module 6 is used to perform weighting processing on all spatio-temporal features by using the attention mechanism to obtain the global spatio-temporal feature.

[0082] The prediction output module 7 processes the global spatio-temporal feature to obtain the predicted value of the taxiing time of the target flight. The prediction output module 7 can be a residual fully connected network.

[0083] In a specific embodiment of the present invention, the convolution kernel size of the one-dimensional convolutional neural network is 3, and the number of convolution kernels is 32. The size of the hidden layer of the gated recurrent unit is 128. The number of residual fully connected layers is 4, and the size is 128. The optimization used in training is SGD, the learning rate of SGD is 0.001, the batch size of training is 32, and the epoch is 500.

[0084] In an embodiment of the present invention, the trained taxiing time prediction model can be obtained through the following steps: S10. Obtain the original dataset and divide the original dataset into a training set, a validation set, and a test set.

[0085] In an embodiment of the present invention, the ratios of the training set, the validation set, and the test set can be set based on actual needs.

[0086] In an embodiment of the present invention, the original dataset can be obtained through the Automatic Dependent Surveillance - Broadcast (ADS - B) of aircraft, the Airport Collaborative Decision - Making (A - CDM) system, and the Meteorological Aerodrome Report (METAR). The METAR report is generated every 30 to 60 minutes. The departure time of the aircraft is associated with the latest report, and the weather conditions are matched with the flight attributes, the operation period identification values, and the flight interaction data.

[0087] Among them, each piece of data in the original dataset may include the taxiing trajectory data, the taxiing time, and the taxiing correlation data of the corresponding departing flight. In addition, some features are obtained through the calculation and processing of the original data. For example, the operation period identification value of the flight in the taxiing correlation data and the features corresponding to the flight interaction data are both obtained through calculation.

[0088] S20. Preprocess the original dataset to obtain the preprocessed original dataset.

[0089] In an embodiment of the present invention, the preprocessing of the original dataset may include: (1) Standardize the original dataset using the mean and standard deviation.

[0090] (2) Fill in the missing values and remove the extreme outliers.

[0091] In an embodiment of the present invention, the 3σ principle can be used to remove outliers. If the data follows a normal distribution, outliers are defined as the set of values that deviate from the mean by more than three times the standard deviation, and will be regarded as outliers and deleted from the dataset.

[0092] In an embodiment of the present invention, the mean and standard deviation of each non - discrete variable in the dataset to be processed in the foregoing content are the mean and standard deviation of the corresponding non - discrete variable in the original dataset.

[0093] S30. Use the training set and the validation set to train the initial taxiing time prediction model to obtain the trained taxiing time prediction model, and evaluate the trained taxiing time prediction model through the test set.

[0094] In an embodiment of the present invention, using the training set and the validation set to train the initial taxiing time prediction model to obtain the trained taxiing time prediction model may specifically include the following steps: S31, set the iteration counter C = 1; set the initial value of the validation loss in the validation loss record set to infinity, and set the storage path for the optimal model parameters.

[0095] S32, if C ≤ C0, execute S33; otherwise, execute S50; C0 is a preset iteration number threshold and can be an empirical value.

[0096] S33, input the training data of the current batch into the current feature encoding module for processing to obtain the feature vectors of the sliding trajectory data and the feature vectors of the sliding correlation data corresponding to each training data, and send them to the spatial feature extraction module and the multi-modal feature fusion module of the current sliding time prediction model.

[0097] In the embodiment of the present invention, the acquisition methods of the feature vectors of the sliding trajectory data and the feature vectors of the sliding correlation data corresponding to each training data can be referred to the foregoing content. The initial value of the current sliding time prediction model is the initialized sliding time prediction model.

[0098] S34, use the current spatial feature extraction module to perform a convolution operation on the feature vectors of the sliding trajectory data to obtain the spatial features of each local path, and send them to the feature enhancement module; and use the feature enhancement module to perform feature splicing on the spatial features of each local path and the corresponding geometric distance to obtain the enhanced spatial features of each local path, and send them to the multi-modal feature fusion module.

[0099] S35, use the multi-modal feature fusion module to fuse the enhanced spatial features of each local path and the feature vectors of the sliding correlation data to obtain the multi-modal features of each local path, and send them to the time series modeling module.

[0100] S36, use the time series modeling module to extract features from the multi-modal features of each local path to obtain the corresponding spatio-temporal features, and send them to the attention weighting module.

[0101] S37, use the attention weighting module to perform a weighting process on all the spatio-temporal features to obtain the global spatio-temporal features, and send them to the prediction output module.

[0102] S38, use the prediction output module to process the global spatio-temporal features to obtain the predicted values of the sliding time corresponding to the training data of the current batch.

[0103] S39. Obtain the loss of the current taxiing time prediction model based on the predicted value and the corresponding true value of the taxiing time corresponding to the training data of the current batch as the current loss, and update the parameters of the current taxiing time prediction model based on the current loss; if the training data of the current batch is the last batch of training data in the training set, execute S40, otherwise, use the next batch of training data as the training data of the current batch and execute S33.

[0104] In the embodiment of the present invention, the current loss can be obtained based on a preset loss function. For example, the mean absolute error. The parameters of the current taxiing time prediction model can be updated by the stochastic gradient descent method.

[0105] S40. Obtain the average validation loss of the current taxiing time prediction model based on the validation set as the current validation loss, and compare the current validation loss with the minimum validation loss in the current validation loss record set. If the current validation loss is less than the minimum validation loss in the current validation loss record, update the minimum validation loss in the current validation loss record set, and save the model parameters of the current taxiing time prediction model to the optimal model parameter storage path; set C = C + 1 and execute S32.

[0106] S50. Use the taxiing time prediction model corresponding to the minimum validation loss in the current validation loss record set as the trained taxiing time prediction model.

[0107] In the embodiment of the present invention, an independent test set is used to evaluate the trained taxiing time prediction model. The mean absolute error (MAE), root mean square error (RMSE), mean absolute percentage error (MAPE), coefficient of determination (R²), and different tolerance levels can be used to evaluate the accuracy of the model prediction.

[0108] Figure 3 Shows the comparison chart of the predicted taxiing time and the true taxiing time of the prediction model used in the method provided by the present invention and the error distribution chart. Figure 3The upper part of the figure shows the fitting degree between the actual taxiing time and the predicted taxiing time of the aircraft. It can be seen that the predicted taxiing time curve of the model has a similar trend to the actual taxiing curve, indicating that the model can capture the change pattern of the taxiing time to a certain extent. In summary, the data processing method for obtaining the taxiing time of departing flights provided by the embodiments of the present invention can solve the technical bottlenecks of traditional taxiing time prediction methods in fine-grained feature extraction and dynamic environment adaptability through multi-modal data deep fusion and deep learning architecture. Specifically, the present invention constructs a collaborative analysis framework for ADS-B trajectory data and multi-dimensional operation data of the A-CDM system, integrating multi-dimensional data such as trajectory points of aircraft taxiing, operation period identification values, flight attributes, and flight interaction data. The present invention uses a one-dimensional convolutional neural network to combine the great circle geometric distance to obtain an enhanced spatial feature map of the local path, realizing the refined modeling of the spatial features of key local paths such as airport taxiing path intersections and bottleneck areas; at the same time, a double-layer gated recurrent unit network is used to dynamically capture the temporal evolution law of the taxiing process through the gating mechanism. In addition, the present invention introduces an attention mechanism to automatically identify and strengthen the taxiing section features that have a significant impact on the prediction result according to external influencing factors, thereby improving the model's attention to key sections. The depth mining of global spatio-temporal features is realized through a residual fully connected network, and the predicted taxiing time value is output, so that the predicted taxiing time is more accurate.

[0109] An embodiment of the present invention also provides an electronic device, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein, the memory stores instructions executable by the at least one processor, and the instructions are configured to execute the method described in the embodiments of the present invention.

[0110] An embodiment of the present invention also provides a computer-readable storage medium storing computer-executable instructions for executing the method described in the embodiments of the present invention.

[0111] It should be understood that various forms of the processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in the present invention can be executed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in the present invention can be achieved, and no limitations are imposed herein.

[0112] The above specific embodiments do not constitute a limitation to the protection scope of the present invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. A data processing method for obtaining the taxiing time of departing flights, characterized in that, The method includes the following steps: S100. Obtain a dataset to be processed; the dataset to be processed includes simulation taxiing trajectory data and taxiing correlation data of a target flight; S200. Respectively perform encoding processing on the simulation taxiing trajectory data and the taxiing correlation data to obtain a feature vector of the simulation taxiing trajectory data and a feature vector of the taxiing correlation data; S300. Use a convolutional kernel with a width of k to perform a one-dimensional convolutional operation on the feature vector of the simulation taxiing trajectory data to extract m local features and form spatial features of m local paths; m = n - k + 1, where n is the number of position points in the simulation taxiing trajectory data; S400. Obtain the geometric distance of each local path, perform feature splicing on the spatial feature of each local path and the corresponding geometric distance, and use the splicing result as the enhanced spatial feature of the local path; S500. Perform fusion processing on the enhanced spatial feature of each local path and the feature vector of the taxiing correlation data to obtain the multi-modal feature of each local path; S600. Extract the temporal dependence relationship in the multi-modal feature of each local path to obtain the spatio-temporal feature of each local path; S700. Use an attention mechanism to perform weighted processing on all spatio-temporal features to obtain global spatio-temporal features; S800. Process the global spatio-temporal features to obtain a predicted value of the taxiing time of the target flight.

2. The method according to claim 1, wherein The taxiing correlation data includes flight attribute information, operation period identification values, environmental data, and flight interaction data; In S200, the feature vector of the taxiing correlation data is obtained through the following steps: S201. Preprocess the taxiing correlation data to obtain initial feature vectors corresponding to flight attribute information, operation period identification values, environmental data, and flight interaction data; S202. Perform feature splicing on the initial feature vectors corresponding to the operation period identification values and the environmental data, and use the obtained splicing result as the fused environmental feature vector; S203. Based on the fused environmental feature vector, obtain a weight vector corresponding to the initial feature vector of the flight interaction data, and based on the obtained weight vector, perform weighted processing on the initial feature vector of the flight interaction data to obtain a corresponding weighted processing result as the weighted interaction feature vector; S204. Perform feature splicing on the initial feature vector corresponding to the flight attribute information, the fused environmental feature vector, and the weighted interaction feature vector, and use the obtained splicing result as the feature vector of the taxiing correlation data.

3. The method according to claim 2, wherein The flight interaction data includes the number of flights moving on the airport surface during the taxiing of the target flight.

4. The method according to claim 3, characterized in that, The weighted interaction feature vector satisfies the following conditions: FW = (λ1f1, λ2f2, ……, λ r f r ,……, λ Q f Q ); Among them, FW is the weighted interaction feature vector, and f r is the initial feature vector corresponding to the r-th interaction data, and λ r is the weight corresponding to the r-th interaction data. λ r is determined based on the fused environmental feature vector and f r The value range of r is from 1 to Q, where Q is the number of interaction data in the flight interaction data.

5. The method according to claim 4, characterized in that, λ r satisfies the following conditions: λ r =σ(W e ·E env +W f ·f r +b); Among them, W e is the environmental feature weight matrix, E env is the fused environmental feature vector, W f is the weight matrix corresponding to f r b is the bias term, σ( ) is the Sigmoid activation function, and · represents dot product.

6. The method according to claim 1, characterized in that, The global spatio-temporal features are obtained through the following steps: S701. Perform a linear transformation on the feature vector of the taxiing correlation data to obtain a corresponding query vector; S702. Obtain the interaction intensity between the spatio-temporal feature of each local path and the query vector to obtain m interaction intensities; S703. Perform normalization processing on the m interaction intensities to obtain the attention weight corresponding to the spatio-temporal feature of each local path; At S704, a weighted sum of the spatio-temporal features of all local paths and the corresponding attention weights is performed to obtain the global spatio-temporal features.

7. The method according to claim 6, wherein Among them, the interaction intensity z(i) corresponding to the i-th local path satisfies the following conditions: z(i)=(q T •h i ) / (d k ) 1 / 2 , h i is the spatio-temporal feature of the i-th local path, q is the query vector, d k is the dimension of q; q T is the transpose of q, and i ranges from 1 to m.

8. The method according to claim 1, characterized in that, In S600, two cascaded gated recurrent units are used to extract the temporal dependence relationship in the multi-modal features of each local path.

9. An electronic device, characterized in that, It includes a processor and a memory; The processor is configured to execute the steps of the method according to any one of claims 1 to 8 by calling the program or instructions stored in the memory.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium is used to store a program or instructions, and the program or instructions cause the computer to execute the steps of the method according to any one of claims 1 to 8.

Images