An airport time capacity evaluation method under resource occupation uncertainty

Abstract

The application discloses an airport time capacity evaluation method under resource occupation uncertainty, and relates to the field of civil aviation, which comprises the following steps: presetting an airport time flight plan of a current date; collecting and preprocessing other airspace user activity data and civil aviation flight activity data of the current date; importing a trained other airspace user activity prediction model and a civil aviation flight prediction model to generate a typical day other airspace user activity timetable of the current date and a typical day civil aviation flight plan timetable of the current date; obtaining a typical day civil aviation flight delay timetable of the current date; expanding the typical day civil aviation flight delay timetable of the current date by constructing a variational autoencoder model, and calculating a daily average delay time of the current date; comparing and analyzing the daily average delay time with a preset daily average delay time threshold; and determining that a typical day civil aviation flight actual operation timetable of the current date is the airport time capacity, thereby improving the operation efficiency of the airport.

Description

Technical Field

[0001] The present invention relates to the field of airport operation management, and more specifically, to an airport time capacity evaluation method under resource occupancy uncertainty. Background Art

[0002] Slot capacity assessment is a crucial task in airport operations management. However, traditional airport slot capacity assessment methods often fail to account for the uncertainty of resource occupancy, resulting in inaccurate and unreliable assessment results. Therefore, it is of great significance to design an airport slot capacity assessment method that addresses the uncertainty of resource occupancy.

[0003] The invention, with publication number CN115310366A, discloses a method for evaluating an airport's published capacity based on a stochastic optimization model, comprising: S1: performing feature analysis and extraction on flight data of the airport to be evaluated, and clustering the data using the k-means method to obtain typical airport operation scenarios and their occurrence probabilities; S2: constructing a stochastic optimization model, using the scenarios obtained by clustering in step S1 and the airport's operating capacity as input, and calculating the number of time slots that should be set for each time period at the airport as the airport's published capacity; S3: establishing a flight delay prediction model using a machine learning method, using the model to predict delays under the published capacity obtained in step S2, and evaluating the setting level of the published capacity. The invention considers the uncertainty of airport capacity under different scenarios, identifies airport operation scenarios through cluster analysis, establishes a stochastic optimization model to calculate the airport's published capacity, and uses a machine learning method to predict flight delays and evaluate the setting level of the published capacity. It does not consider that in airport operation management, in addition to civil aviation flights, other airspace user activities may also affect the scheduled time slots of civil aviation flights. These activities may increase the uncertainty of resource utilization, thereby affecting the airport's time slot capacity. Therefore, the invention provides an innovative design to address the above issues. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, the present invention provides the following technical solution: a method for evaluating airport time slot capacity under resource occupancy uncertainty, comprising:

[0005] Step S1: Preset the airport flight schedule for the current date; collect other airspace user activity data and civil aviation flight activity data for the current date and perform pre-processing;

[0006] Step S2: importing the pre-processed other airspace user activity data of the current date into the trained other airspace user activity prediction model to generate a typical day other airspace user activity schedule for the current date;

[0007] Step S3: importing the pre-processed civil aviation flight activity data of the current date into the trained civil aviation flight prediction model to generate a typical day civil aviation flight schedule for the current date;

[0008] Step S4: Construct a joint simulation delay model, jointly simulate the activity schedule of other airspace users on a typical day of the current date and the civil aviation flight schedule on a typical day of the current date, and obtain the daily delay schedule of civil aviation flights on a typical day of the current date;

[0009] Step S5: Expand the typical daily civil aviation flight delay schedule for the current date by constructing a variational autoencoder model, and calculate the average daily delay time for the current date;

[0010] Step S6: Compare and analyze the daily average delay time with the preset daily average delay time threshold;

[0011] If the daily average delay time is greater than the preset daily average delay time threshold, the airport schedule for the current date exceeds the airport schedule capacity and needs to be manually adjusted.

[0012] If the daily average delay time is less than or equal to the preset daily average delay time threshold, the airport schedule flight plan for the current date is within the airport schedule capacity. The actual operation schedule of civil aviation flights for the typical day of the current date is updated and generated based on the daily average delay time. At this time, it is determined that the actual operation schedule of civil aviation flights for the typical day of the current date is the airport schedule capacity.

[0013] Furthermore, the pre-processing method includes: cleaning other airspace user activity data and civil aviation flight activity data of the current date respectively by using the LOF algorithm;

[0014] Assume that dataset C represents the activity data of other airspace users and civil aviation flight activities on any current date. Use the LOF algorithm to clean abnormal data points in dataset C. The steps are as follows:

[0015] The dataset C includes c data points. Let p be any data point in the dataset C. Calculate the distance between the data point p and all other data points in the dataset C, and select the distance of the kth nearest neighbor as the k distance of the data point p, denoted as d k (p); the neighborhood is all data points within the distance range from data point p to its kth nearest neighbor;

[0016] Calculate the kth reachable distance between a data point p and any other data point o in the neighborhood: reach_dist k (p,o)=max{d k(p), d(p, o)}; where d(p, o) represents the distance between a data point p and a data point o in its neighborhood; if o is within the distance range of p's kth nearest neighbor, then the kth reachable distance from data point p to o is the kth distance of point p, otherwise it is the distance d(p, o);

[0017] Calculate the kth local reachability density of data point p Among them, N k (p) represents the kth neighborhood distance of data point p; |N k (p)| is the set of points within the kth neighborhood distance of the data point p;

[0018] Calculate the k-th local anomaly factor of data point p: Among them, lrd k (o) represents the local reachability density of data point o in the neighborhood of data point p; It represents the ratio of the local reachability density of data point o in the kth neighborhood of data point p to the local reachability density of data point p;

[0019] Calculate the kth local anomaly factor for each of the c data points in the dataset C, set the kth local anomaly factor value threshold, and treat the data points that exceed the kth local anomaly factor value threshold as abnormal data points. Output the abnormal data point set M, remove the abnormal data point set M from the dataset C, and output the final dataset after removing the abnormal data points; this dataset represents the preprocessed other airspace user activity data and civil aviation flight activity data on any current date.

[0020] Furthermore, the other airspace user activity data on the current date includes the total number of different flights in other airspaces, routes of different flights in other airspaces, take-off times of different flights in other airspaces, and landing times of different flights in other airspaces; the other airspace user activity data on the current date is obtained through a flight recorder.

[0021] Furthermore, other airspace user activity prediction model training methods include:

[0022] Pre-collect u groups of training data on other airspace user activities on historical dates and a typical daily other airspace user activity schedule on the corresponding historical dates as a first sample set, where u is an integer greater than 1;

[0023] Dividing the first sample set into a training set and a test set, constructing a classifier, using the training data of other airspace user activities on historical dates in the training set as input data, and the typical daily other airspace user activity schedules on historical dates in the training set as output data, training the classifier to obtain an initial classifier, and testing the initial classifier using the test set;

[0024] When the root mean square error between the predicted historical date typical day other airspace user activity timetable and the actual historical date typical day other airspace user activity timetable is within 0.01, the classifier meeting the preset accuracy is output as a trained other airspace user activity prediction model, and the other airspace user activity prediction model is one of a Naive Bayes algorithm, a random forest regression or a neural network regression model.

[0025] The root mean square error expression is

[0026] RMSE=sqrt((1 / f)*sum((y_pred-y_obs)^2))

[0027] In the formula, RMSE is the root mean square error, f is the number of historical date typical day other airspace user activity timetables, y_pred is the predicted historical date typical day other airspace user activity timetable, and y_obs is the actual historical date typical day other airspace user activity timetable.

[0028] Further, the current date civil aviation flight activity data includes civil aviation flight total times, civil aviation flight routes, civil aviation flight takeoff times and civil aviation flight landing times; and the current date civil aviation flight activity data is obtained through a flight recorder.

[0029] Further, the training method of the civil aviation flight prediction model comprises:

[0030] o groups of historical date civil aviation flight activity training data and corresponding historical date typical day civil aviation flight schedule timetables are pre-collected as a second sample set, and o is an integer greater than 1;

[0031] The second sample set is divided into a training set and a test set, a classifier is constructed, the historical date civil aviation flight activity training data in the training set is taken as input data, the historical date typical day civil aviation flight schedule timetable in the training set is taken as output data, the classifier is trained to obtain an initial classifier, and the initial classifier is tested by using the test set;

[0032] When the loss function error between the predicted historical date typical day civil aviation flight schedule timetable and the actual historical date typical day civil aviation flight schedule timetable is within 0.01, the classifier meeting the preset accuracy is output as a trained civil aviation flight prediction model, and the civil aviation flight prediction model is one of a decision tree algorithm, a random forest algorithm or a neural network regression model, and the loss function of the civil aviation flight prediction model is set as a huber loss function;

[0033] The huber loss function expression is: In the formula, y is the real target output. is a prediction result of the civil aviation flight prediction model; and is a control behavior parameter; and is used for controlling the behavior of the loss function;

[0034] By calculating the gradient information of the loss function, the value of the loss function is gradually reduced, the civil aviation flight prediction model is optimized, and the prediction output continuously approaches the value in the real training set. When the value of the loss function of the training set no longer decreases, the training is terminated, and the trained civil aviation flight prediction model is obtained.

[0035] Further, the method for creating the joint simulation delay model comprises:

[0036] The joint simulation delay model is expressed as a function of other airspace user activities and civil aviation flight scheduled times, and the expression is as follows:

[0037]

[0038] In the formula, D represents a typical daily civil aviation flight delay schedule of the current date, kf represents a typical daily civil aviation flight schedule of the current date, ym represents a typical daily other airspace user activity schedule of the current date, a represents an influence factor of the typical daily civil aviation flight schedule of the current date, and represents an influence factor of the typical daily other airspace user activity schedule of the current date.

[0039] The typical daily other airspace user activity schedule of the current date and the typical daily civil aviation flight schedule of the current date are jointly simulated by a computer to simulate the influence of the typical daily other airspace user activity schedule of the current date on the typical daily civil aviation flight schedule of the current date, and the typical daily civil aviation flight delay schedule of the current date is obtained.

[0040] Further, the method for training the variational autoencoder model comprises:

[0041] The civil aviation flight daily delay schedule of the typical day of the historical date and the corresponding expanded civil aviation flight daily delay schedule of the typical day of the v historical dates are pre-acquired;

[0042] The civil aviation flight daily delay schedule of the typical day of the historical date is used as an input feature, and the corresponding expanded civil aviation flight daily delay schedule of the typical day of the v historical dates is used as a target output; a convolutional neural network architecture is used as an encoder and a decoder, the encoder maps the input feature to a latent vector in a latent space, and the last layer of the encoder outputs the mean and standard deviation of the latent vector;

[0043] Using the mean and standard deviation of the encoder output, a latent vector is sampled from the latent space. The decoder maps the sampled latent vector back to the reconstructed data space, generating an output similar to the original data. The final layer of the decoder outputs the reconstructed data. A typical daily civil aviation flight delay schedule is used as a training dataset, and the model is optimized by minimizing a loss function consisting of a reconstruction loss and a KL divergence.

[0044] The reconstruction loss expression is: loss = r_loss + β·kl_d; where r_loss is the reconstruction loss, kl_d is the KL divergence, and β is the weight factor of the KL divergence;

[0045] The KL divergence expression is: kl_d = -0.5 sum(1 + log(s^2) - mu^2 - s^2);

[0046] Where mu is the mean of the latent vector output by the encoder, and s is the standard deviation of the latent vector output by the encoder.

[0047] Furthermore, the airport time capacity assessment method is:

[0048] If the daily average delay time is greater than the preset daily average delay time threshold, the airport schedule for the current date exceeds the airport schedule capacity and needs to be manually adjusted.

[0049] If the daily average delay time is less than or equal to the preset daily average delay time threshold, the airport schedule flight plan for the current date is within the airport schedule capacity. The typical day civil aviation flight schedule for the current date is added to the daily average delay time to update the current date's typical day civil aviation flight actual operation schedule. The typical day civil aviation flight actual operation schedule for the current date is determined to be the airport schedule capacity.

[0050] An airport slot capacity assessment system for conditions of resource occupancy uncertainty is implemented based on the aforementioned airport slot capacity assessment method for conditions of resource occupancy uncertainty, comprising:

[0051] The data collection module presets the airport flight schedule for the current date; collects other airspace user activity data and civil aviation flight activity data for the current date, and performs pre-processing;

[0052] A module for generating a typical day other airspace user activity schedule for the current date imports the pre-processed other airspace user activity data for the current date into the trained other airspace user activity prediction model to generate a typical day other airspace user activity schedule for the current date;

[0053] A typical day civil aviation flight schedule generation module for the current date imports the pre-processed civil aviation flight activity data for the current date into the trained civil aviation flight prediction model to generate a typical day civil aviation flight schedule for the current date;

[0054] The module generates the typical day civil aviation flight delay schedule for the current date, builds a joint simulation delay model, and jointly simulates the typical day other airspace user activity schedule for the current date with the typical day civil aviation flight plan schedule for the current date to obtain the typical day civil aviation flight delay schedule for the current date;

[0055] The module calculates the average daily delay time of the current date. It expands the typical daily civil aviation flight delay schedule of the current date by building a variational autoencoder model, obtains the typical daily civil aviation flight delay schedule of v historical dates, and calculates the average daily delay time of the current date.

[0056] The airport time capacity assessment module compares and analyzes the statistical daily average delay time with the preset daily average delay time threshold to determine whether the airport time flight plan for the current date exceeds the airport time capacity. It also updates and generates the actual operation schedule of civil aviation flights on a typical day for the current date based on the daily average delay time, and determines that the actual operation schedule of civil aviation flights on a typical day for the current date is the airport time capacity.

[0057] The technical effects and advantages of the airport time slot capacity assessment method under resource occupancy uncertainty of the present invention are as follows:

[0058] The present invention presets the airport flight schedule for the current date; collects and pre-processes the activity data of other airspace users and civil aviation flight activities on the current date, thereby obtaining more authentic and comprehensive data and improving the availability and reliability of the data. By generating a typical day's other airspace user activity schedule and civil aviation flight schedule, the present invention helps air traffic control departments and airlines to better arrange flight plans and resource allocation, thereby improving the efficiency and safety of flight operations.

[0059] The constructed joint simulation delay model jointly simulates the activity schedule of other airspace users on a typical day on the current date with the scheduled civil aviation flight schedule on a typical day on the current date to obtain the daily delay schedule of civil aviation flights on the typical day on the current date. This model takes into account the impact of the activities of other airspace users on the typical day on the scheduled civil aviation flight schedule on a typical day and comprehensively considers the usage of the entire airspace.

[0060] By building a variational autoencoder model to expand the typical daily civil aviation flight delay schedule for the current date, we obtain v historical typical daily civil aviation flight delay schedules and calculate the average daily delay time for the current date. By expanding the typical daily delay schedule, we can generate more daily delay time data, which helps increase the diversity and quantity of data and more accurately estimate delay time.

[0061] The statistical daily average delay time is compared and analyzed with the preset daily average delay time threshold to determine whether the airport flight schedule for the current date exceeds the airport schedule capacity. The actual operation schedule of civil aviation flights for a typical day on the current date is updated and generated based on the daily average delay time. It is determined that the actual operation schedule of civil aviation flights for a typical day on the current date is the airport schedule capacity. The present invention can provide a reference for the actual operation capacity of the airport under the activities of other airspace users, thereby improving the airport operation efficiency while ensuring operation safety and efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 Schematic diagram of an airport time slot capacity assessment method under resource occupancy uncertainty according to the present invention;

[0063] Figure 2 The figure is a schematic diagram of an airport time slot capacity assessment system under resource occupancy uncertainty according to the present invention. DETAILED DESCRIPTION

[0064] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0065] Example 1

[0066] See also Figure 1 As shown, the airport time slot capacity assessment method under resource occupancy uncertainty described in this embodiment includes:

[0067] Step S1: Preset the airport flight schedule for the current date; collect other airspace user activity data and civil aviation flight activity data for the current date and perform pre-processing;

[0068] Step S2: importing the pre-processed other airspace user activity data of the current date into the trained other airspace user activity prediction model to generate a typical day other airspace user activity schedule for the current date;

[0069] Step S3: importing the pre-processed civil aviation flight activity data of the current date into the trained civil aviation flight prediction model to generate a typical day civil aviation flight schedule for the current date;

[0070] Step S4: Construct a joint simulation delay model, jointly simulate the activity schedule of other airspace users on a typical day of the current date and the civil aviation flight schedule on a typical day of the current date, and obtain the daily delay schedule of civil aviation flights on a typical day of the current date;

[0071] Step S5: Expand the typical daily civil aviation flight delay schedule for the current date by constructing a variational autoencoder model, and calculate the average daily delay time for the current date;

[0072] Step S6: Compare and analyze the daily average delay time with the preset daily average delay time threshold;

[0073] If the daily average delay time is greater than the preset daily average delay time threshold, the airport schedule for the current date exceeds the airport schedule capacity and needs to be manually adjusted.

[0074] If the daily average delay time is less than or equal to the preset daily average delay time threshold, the airport schedule flight plan for the current date is within the airport schedule capacity. The actual operation schedule of civil aviation flights for the typical day of the current date is updated and generated based on the daily average delay time. At this time, it is determined that the actual operation schedule of civil aviation flights for the typical day of the current date is the airport schedule capacity.

[0075] The preprocessing method includes: cleaning other airspace user activity data and civil aviation flight activity data of the current date respectively by using the LOF algorithm;

[0076] Assume that dataset C represents the activity data of other airspace users and civil aviation flight activities on any current date. Use the LOF algorithm to clean abnormal data points in dataset C. The steps are as follows:

[0077] The dataset C includes c data points. Let p be any data point in the dataset C. Calculate the distance between the data point p and all other data points in the dataset C, and select the distance of the kth nearest neighbor as the k distance of the data point p, denoted as d k (p); the neighborhood is all data points within the distance range from data point p to its kth nearest neighbor;

[0078] Calculate the kth reachable distance between a data point p and any other data point o in the neighborhood: reach_dist k (p,o)=max{d k(p), d(p, o)}; where d(p, o) represents the distance between a data point p and a data point o in its neighborhood; if o is within the distance range of p's kth nearest neighbor, then the kth reachable distance from data point p to o is the kth distance of point p, otherwise it is the distance d(p, o);

[0079] Calculate the kth local reachability density of data point p Among them, N k (p) represents the kth neighborhood distance of data point p; |N k (p)| is the set of points within the kth neighborhood distance of the data point p;

[0080] Calculate the k-th local anomaly factor of data point p: Among them, lrd k (o) represents the local reachability density of data point o in the neighborhood of data point p; It represents the ratio of the local reachability density of data point o in the kth neighborhood of data point p to the local reachability density of data point p;

[0081] Calculate the kth local anomaly factor for each of the c data points in the dataset C, set the kth local anomaly factor value threshold, and treat the data points that exceed the kth local anomaly factor value threshold as abnormal data points. Output the abnormal data point set M, remove the abnormal data point set M from the dataset C, and output the final dataset after removing the abnormal data points; this dataset represents the preprocessed other airspace user activity data and civil aviation flight activity data on any current date.

[0082] For example, dataset C includes 20 data points. Let p be any data point in dataset C. Calculate the distance between data point p and all other data points in dataset C, and select the distance of the fifth nearest neighbor as the k distance of data point p, denoted as d k (p); d k The value of (p) is 10m; the neighborhood is all data points within the distance range from data point p to its fifth nearest neighbor;

[0083] Calculate the kth reachable distance between a data point p and any other data point o in the neighborhood: reach_dist k (p,o)=max{d k (p),d(p,o)}; reach_dist k The reachable distance of (p,o) is 10m, and the kth local reachable density lrd of data point p is calculated. k (p) and the local reachable density lrd of the data point o k (o), and then calculate the k-th local anomaly factor of data point p;

[0084] The k-th local anomaly factor of each of the 20 data points in the data set C is calculated, a threshold value of the k-th local anomaly factor is set, data points exceeding the threshold value of the k-th local anomaly factor are taken as abnormal data points, an abnormal data point set M is output, and the abnormal data point set M is removed from the data set C to output a data set after removing abnormal data points;

[0085] The threshold value of the k-th local anomaly factor is set by a worker, and the average value of the k-th local anomaly factors of the plurality of data points is taken as the threshold value of the k-th local anomaly factor through calculation;

[0086] The other airspace user activity data of the current date includes total takeoff times of different flights in other airspace, flight routes of different flights in other airspace, takeoff times of different flights in other airspace, and landing times of different flights in other airspace; and the other airspace user activity data of the current date is obtained through a flight recorder.

[0087] The obtaining method includes installing a flight recorder device on an aircraft and correctly configuring and calibrating the flight recorder device, and the flight recorder device regularly uploads the other airspace user activity data to a ground flight management device for storage.

[0088] The other airspace user activity prediction model training method includes:

[0089] A first sample set is obtained by pre-collecting u sets of historical date other airspace user activity training data and corresponding historical date typical day other airspace user activity schedules, and u is an integer greater than 1;

[0090] The first sample set is divided into a training set and a test set, a classifier is constructed, the historical date other airspace user activity training data in the training set is taken as input data, the historical date typical day other airspace user activity schedule in the training set is taken as output data, the classifier is trained to obtain an initial classifier, and the initial classifier is tested by using the test set;

[0091] When the root mean square error between the predicted historical date typical day other airspace user activity schedule and the actual historical date typical day other airspace user activity schedule is within 0.01, the classifier satisfying the preset accuracy is output as a trained other airspace user activity prediction model, and the other airspace user activity prediction model is one of a Naive Bayes algorithm, a random forest regression, or a neural network regression model;

[0092] The root mean square error expression is

[0093] RMSE = sqrt((1 / f) * sum((y_pred-y_obs)^2))

[0094] Where RMSE is the root mean square error, f is the number of other airspace user activity schedules on a typical day on a historical date, y_pred is the predicted other airspace user activity schedule on a typical day on a historical date, and y_obs is the actual other airspace user activity schedule on a typical day on a historical date.

[0095] The civil aviation flight activity data of the current date includes the total number of civil aviation flights, civil aviation flight routes, civil aviation flight take-off times and civil aviation flight landing times; the civil aviation flight activity data of the current date is obtained through the flight recorder, and the acquisition method is consistent with the acquisition method of other airspace user activity data of the current date.

[0096] The training method of the civil aviation flight prediction model includes:

[0097] Pre-collect o groups of civil aviation flight activity training data of historical dates and typical daily civil aviation flight schedules of corresponding historical dates as the second sample set, where o is an integer greater than 1;

[0098] Dividing the second sample set into a training set and a test set, constructing a classifier, using the civil aviation flight activity training data of historical dates in the training set as input data, and the typical daily civil aviation flight schedule of historical dates in the training set as output data, training the classifier to obtain an initial classifier, and testing the initial classifier using the test set;

[0099] When the loss function error between the predicted historical typical day civil aviation flight schedule and the actual historical typical day civil aviation flight schedule is within 0.01, output a classifier that meets the preset accuracy as a trained civil aviation flight prediction model, the civil aviation flight prediction model is one of a decision tree algorithm, a random forest algorithm, or a neural network regression model, and the loss function of the civil aviation flight prediction model is set to a Huber loss function;

[0100] The huber loss function expression is: Where y is the true target output; is the prediction result of the civil aviation flight prediction model; δ is the control behavior parameter; it is used to control the behavior of the loss function;

[0101] By calculating the gradient information of the loss function, gradually reducing the value of the loss function, optimizing the civil aviation flight prediction model, and making the predicted output continuously approach the value in the real training set, the training is terminated when the value of the loss function of the training set no longer decreases, and a trained civil aviation flight prediction model is obtained.

[0102] The method for creating the joint simulation delay model includes:

[0103] The joint simulation delay model is expressed as a function of other airspace user activities and civil aviation flight schedule time, and the expression is as follows:

[0104]

[0105] In the formula, D represents the typical daily civil aviation flight delay schedule of the current date, kf represents the typical daily civil aviation flight schedule of the current date, ym represents the typical daily other airspace user activity schedule of the current date, alpha represents the influence factor of the typical daily civil aviation flight schedule of the current date, and theta represents the influence factor of the typical daily other airspace user activity schedule of the current date.

[0106] The typical daily other airspace user activity schedule of the current date and the typical daily civil aviation flight schedule of the current date are simulated by a computer to simulate the influence of the typical daily other airspace user activity schedule of the current date on the typical daily civil aviation flight schedule of the current date, and the typical daily civil aviation flight delay schedule of the current date is obtained.

[0107] The training method of the variational autoencoder model comprises:

[0108] The typical daily civil aviation flight delay schedule of the historical date and the corresponding expanded v typical daily civil aviation flight delay schedule of the historical date are collected in advance.

[0109] The typical daily civil aviation flight delay schedule of the historical date is used as an input feature, and the corresponding expanded v typical daily civil aviation flight delay schedule of the historical date is used as a target output; a convolutional neural network architecture is used as an encoder and a decoder, the encoder maps the input feature to a latent vector in a latent space, and the last layer of the encoder outputs the mean and standard deviation of the latent vector;

[0110] A latent vector is sampled from the latent space using the mean and standard deviation output by the encoder, the decoder maps the sampled latent vector back to a reconstructed data space to generate an output similar to the original data, and the last layer of the decoder outputs the reconstructed data; the typical daily civil aviation flight delay schedule is used as a training data set, and the model is optimized by minimizing the loss function; the loss function is composed of two parts of reconstruction loss and KL divergence;

[0111] The reconstruction loss expression is: loss = r_loss + beta * kl_d; in the formula, r_loss is the reconstruction loss, kl_d is the KL divergence, and beta is the weight factor of the KL divergence;

[0112] The KL divergence expression is: kl_d = -0.5 * sum(1 + log(s^2) - mu^2 - s^2);

[0113] Where mu is the mean of the latent vector output by the encoder, and s is the standard deviation of the latent vector output by the encoder.

[0114] The airport slot capacity assessment method is:

[0115] If the daily average delay time is greater than the preset daily average delay time threshold, the airport schedule for the current date exceeds the airport schedule capacity and needs to be manually adjusted.

[0116] If the daily average delay time is less than or equal to the preset daily average delay time threshold, the airport schedule flight plan for the current date is within the airport schedule capacity. The typical day civil aviation flight schedule for the current date is added to the daily average delay time to update the current date's typical day civil aviation flight actual operation schedule. The typical day civil aviation flight actual operation schedule for the current date is determined to be the airport schedule capacity.

[0117] The daily average delay time threshold is set by the staff. The variational autoencoder model is used to expand the typical daily civil aviation flight delay schedule for the current date to obtain v typical daily civil aviation flight delay schedules for the current date. The mean of the daily delay time is taken as the daily average delay time threshold.

[0118] In this embodiment, by presetting the airport flight schedule for the current date, collecting other airspace user activity data and civil aviation flight activity data for the current date, and performing preprocessing, more realistic and comprehensive data can be obtained, thereby improving data availability and reliability. By generating a typical day's other airspace user activity schedule and civil aviation flight schedule, air traffic control departments and airlines can better arrange flight plans and resource allocation, thereby improving the efficiency and safety of flight operations.

[0119] The constructed joint simulation delay model jointly simulates the activity schedule of other airspace users on a typical day on the current date with the scheduled civil aviation flight schedule on a typical day on the current date to obtain the daily delay schedule of civil aviation flights on the typical day on the current date. This model takes into account the impact of the activities of other airspace users on the typical day on the scheduled civil aviation flight schedule on a typical day and comprehensively considers the usage of the entire airspace.

[0120] By building a variational autoencoder model to expand the typical daily civil aviation flight delay schedule for the current date, we obtain v historical typical daily civil aviation flight delay schedules and calculate the average daily delay time for the current date. By expanding the typical daily delay schedule, we can generate more daily delay time data, which helps increase the diversity and quantity of data and more accurately estimate delay time.

[0121] The statistical daily average delay time is compared and analyzed with the preset daily average delay time threshold to determine whether the airport flight schedule for the current date exceeds the airport schedule capacity. The actual operation schedule of civil aviation flights for a typical day on the current date is updated and generated based on the daily average delay time. It is determined that the actual operation schedule of civil aviation flights for a typical day on the current date is the airport schedule capacity. The present invention can provide a reference for the actual operation capacity of the airport under the activities of other airspace users, thereby improving the airport operation efficiency while ensuring operation safety and efficiency.

[0122] Example 2

[0123] See also Figure 2 As shown, for parts not described in detail in this embodiment, please refer to the description of Example 1. A system for evaluating airport slot capacity under resource occupancy uncertainty is provided, including:

[0124] The data collection module presets the airport flight schedule for the current date; collects other airspace user activity data and civil aviation flight activity data for the current date, and performs pre-processing;

[0125] A module for generating a typical day other airspace user activity schedule for the current date imports the pre-processed other airspace user activity data for the current date into the trained other airspace user activity prediction model to generate a typical day other airspace user activity schedule for the current date;

[0126] A typical day civil aviation flight schedule generation module for the current date imports the pre-processed civil aviation flight activity data for the current date into the trained civil aviation flight prediction model to generate a typical day civil aviation flight schedule for the current date;

[0127] The module generates the typical day civil aviation flight delay schedule for the current date, builds a joint simulation delay model, and jointly simulates the typical day other airspace user activity schedule for the current date with the typical day civil aviation flight plan schedule for the current date to obtain the typical day civil aviation flight delay schedule for the current date;

[0128] The module calculates the average daily delay time of the current date. It expands the typical daily civil aviation flight delay schedule of the current date by building a variational autoencoder model, obtains the typical daily civil aviation flight delay schedule of v historical dates, and calculates the average daily delay time of the current date.

[0129] The airport time capacity assessment module compares and analyzes the statistical daily average delay time with the preset daily average delay time threshold to determine whether the airport time flight plan for the current date exceeds the airport time capacity. It also updates and generates the actual operation schedule of civil aviation flights on a typical day for the current date based on the daily average delay time, and determines that the actual operation schedule of civil aviation flights on a typical day for the current date is the airport time capacity.

[0130] Since the electronic device introduced in this embodiment is an electronic device used to implement a method for assessing airport time slot capacity under uncertainty in resource occupancy in the embodiment of this application, those skilled in the art will be able to understand the specific implementation of the electronic device of this embodiment and its various variations based on the method for assessing airport time slot capacity under uncertainty in resource occupancy introduced in the embodiment of this application. Therefore, how the electronic device implements the method in the embodiment of this application will not be described in detail here. As long as those skilled in the art implement the electronic device used in the method for assessing airport time slot capacity under uncertainty in resource occupancy in the embodiment of this application, it falls within the scope of protection of this application.

[0131] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters and thresholds in the formulas are set by technicians in this field according to actual conditions.

[0132] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiment. All technical solutions based on the principles of the present invention are within the scope of protection of the present invention. It should be noted that for users of ordinary skill in the art, various improvements and modifications that do not depart from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A method for airport slot capacity assessment under resource occupancy uncertainty, characterized by: The method for evaluating airport slot capacity under resource occupancy uncertainty comprises the following steps: Step S1: Preset the airport flight schedule for the current date; collect other airspace user activity data and civil aviation flight activity data for the current date and perform pre-processing; The preprocessing method includes: cleaning other airspace user activity data and civil aviation flight activity data of the current date respectively by using the LOF algorithm; Assume that dataset C represents the activity data of other airspace users and civil aviation flight activities on any current date. Use the LOF algorithm to clean abnormal data points in dataset C. The steps are as follows: The dataset C includes c data points. Let p be any data point in the dataset C. Calculate the distance between the data point p and all other data points in the dataset C, and select the distance of the kth nearest neighbor as the k distance of the data point p, denoted as d k (p); the neighborhood is all data points within the distance range from data point p to its kth nearest neighbor; Calculate the kth reachable distance between a data point p and any other data point o in the neighborhood: reach_dist k (p,o)=max{d k (p), d(p, o)}; where d(p, o) represents the distance between a data point p and a data point o in its neighborhood; if o is within the distance range of p's kth nearest neighbor, then the kth reachable distance from data point p to o is the kth distance of point p, otherwise it is the distance d(p, o); Calculate the kth local reachability density of data point p Among them, N k (p) represents the kth neighborhood distance of data point p; |N k (p)| is the set of points within the kth neighborhood distance of the data point p; Calculate the k-th local anomaly factor of data point p: Among them, lrd k (o) represents the local reachability density of data point o in the neighborhood of data point p; Represents the ratio of the local reachable density of data point o in the kth neighborhood of data point p to the local reachable density of data point p; calculate the kth local anomaly factor for each of the c data points in the data set C, set the kth local anomaly factor value threshold, and regard the data points exceeding the kth local anomaly factor value threshold as abnormal data points, output the abnormal data point set M, remove the abnormal data point set M from the data set C, and output the final data set after removing the abnormal data points; Step S2: importing the pre-processed other airspace user activity data of the current date into the trained other airspace user activity prediction model to generate a typical day other airspace user activity schedule for the current date; Other airspace user activity prediction model training methods include: Pre-collect u groups of training data on other airspace user activities on historical dates and a typical daily other airspace user activity schedule on the corresponding historical dates as a first sample set, where u is an integer greater than 1; Dividing the first sample set into a training set and a test set, constructing a classifier, using the training data of other airspace user activities on historical dates in the training set as input data, and the typical daily other airspace user activity schedules on historical dates in the training set as output data, training the classifier to obtain an initial classifier, and testing the initial classifier using the test set; When the root mean square error between the predicted other airspace user activity schedule for a typical day on a historical date and the actual other airspace user activity schedule for a typical day on a historical date is within 0.01, a classifier that meets the preset accuracy is output as the trained other airspace user activity prediction model, where the other airspace user activity prediction model is one of a naive Bayes algorithm, a random forest regression model, or a neural network regression model; The root mean square error expression is: RMSE = sqrt((1 / f)*sum((y_pred-y_obs)^2)) where RMSE is the root mean square error, f is the number of other airspace user activity schedules on a typical day on a historical date, y_pred is the predicted other airspace user activity schedule on a typical day on a historical date, and y_obs is the actual other airspace user activity schedule on a typical day on a historical date. Step S3: importing the pre-processed civil aviation flight activity data of the current date into the trained civil aviation flight prediction model to generate a typical day civil aviation flight schedule for the current date; Step S4: Construct a joint simulation delay model, jointly simulate the activity schedule of other airspace users on a typical day of the current date and the civil aviation flight schedule on a typical day of the current date, and obtain the daily delay schedule of civil aviation flights on a typical day of the current date; Step S5: Expand the typical daily civil aviation flight delay schedule for the current date by constructing a variational autoencoder model, and calculate the average daily delay time for the current date; Step S6: Compare and analyze the daily average delay time with the preset daily average delay time threshold; If the daily average delay time is greater than the preset daily average delay time threshold, the airport schedule for the current date exceeds the airport schedule capacity and needs to be manually adjusted. If the daily average delay time is less than or equal to the preset daily average delay time threshold, the airport schedule flight plan for the current date is within the airport schedule capacity. The actual operation schedule of civil aviation flights for the typical day of the current date is updated and generated based on the daily average delay time. At this time, it is determined that the actual operation schedule of civil aviation flights for the typical day of the current date is the airport schedule capacity.

2. The method for airport time slot capacity assessment under resource occupancy uncertainty according to claim 1 is characterized in that: The other airspace user activity data includes the total number of different flights in other airspaces, different flight routes in other airspaces, take-off times of different flights in other airspaces, and landing times of different flights in other airspaces; the other airspace user activity data is obtained through flight recorders.

3. The method for airport time slot capacity assessment under resource occupancy uncertainty according to claim 1 is characterized in that: Civil aviation flight activity data includes the total number of civil aviation flights, civil aviation flight routes, civil aviation flight take-off times and civil aviation flight landing times; civil aviation flight activity data is obtained through flight recorders.

4. The method for airport time slot capacity assessment under resource occupancy uncertainty according to claim 1 is characterized in that: The training method of the civil aviation flight prediction model includes: Pre-collect o groups of civil aviation flight activity training data of historical dates and typical daily civil aviation flight schedules of corresponding historical dates as the second sample set, where o is an integer greater than 1; Dividing the second sample set into a training set and a test set, constructing a classifier, using the civil aviation flight activity training data of historical dates in the training set as input data, and the typical daily civil aviation flight schedule of historical dates in the training set as output data, training the classifier to obtain an initial classifier, and testing the initial classifier using the test set; When the loss function error between the predicted historical typical day civil aviation flight schedule and the actual historical typical day civil aviation flight schedule is within 0.01, output a classifier that meets the preset accuracy as a trained civil aviation flight prediction model, the civil aviation flight prediction model is one of a decision tree algorithm, a random forest algorithm, or a neural network regression model, and the loss function of the civil aviation flight prediction model is set to a Huber loss function; The huber loss function expression is: Where y is the true target output; is the prediction result of the civil aviation flight prediction model; δ is the control behavior parameter; it is used to control the behavior of the loss function; By calculating the gradient information of the loss function, gradually reducing the value of the loss function, optimizing the civil aviation flight prediction model, and making the predicted output continuously approach the value in the real training set, the training is terminated when the value of the loss function of the training set no longer decreases, and a trained civil aviation flight prediction model is obtained.

5. According to the method for airport time slot capacity assessment under resource occupancy uncertainty of claim 1, the method for constructing the joint simulation delay model comprises: The joint simulation delay model is expressed as a function of other airspace user activities and civil aviation flight schedules, as shown in the following expression: Down: Where D represents the typical day civil aviation flight delay schedule of the current date, kf represents the typical day civil aviation flight schedule of the current date, ym represents the typical day other airspace user activity schedule of the current date, α represents the influence factor of the typical day civil aviation flight schedule of the current date, and θ represents the influence factor of the typical day other airspace user activity schedule of the current date. The typical day other airspace user activity schedule of the current date and the typical day civil aviation flight plan schedule of the current date are jointly simulated by computer to obtain the typical day civil aviation flight delay schedule of the current date.

6. The method for airport time slot capacity assessment under resource occupancy uncertainty according to claim 1, characterized in that: The training method of the variational autoencoder model includes: Pre-collect the daily delay schedule of civil aviation flights on typical days of historical dates and the corresponding expanded daily delay schedule of civil aviation flights on typical days of historical dates; The typical day civil aviation flight delay schedule for historical dates is used as input features, and the corresponding expanded v typical day civil aviation flight delay schedules for historical dates are used as target outputs. A convolutional neural network architecture is used as the encoder and decoder. The encoder maps the input features to a latent vector in the latent space, and the last layer of the encoder outputs the mean and standard deviation of the latent vector. Using the mean and standard deviation of the encoder output, a latent vector is sampled from the latent space. The decoder maps the sampled latent vector back to the reconstructed data space, generating an output similar to the original data. The final layer of the decoder outputs the reconstructed data. A typical daily civil aviation flight delay schedule is used as a training dataset, and the model is optimized by minimizing a loss function consisting of a reconstruction loss and a KL divergence. The reconstruction loss expression is: loss = r_loss + β·kl_d; where r_loss is the reconstruction loss, kl_d is the KL divergence, and β is the weight factor of the KL divergence; The KL divergence expression is: kl_d = -0.5 sum(1 + log(s^2) - mu^2 - s^2); Where mu is the mean of the latent vector output by the encoder, and s is the standard deviation of the latent vector output by the encoder.

7. The method for airport time slot capacity assessment under resource occupancy uncertainty according to claim 1, characterized in that: The airport slot capacity assessment method is: The actual operation schedule of the typical day civil aviation flight on the current date is updated and generated by adding the typical day civil aviation flight schedule on the current date to the daily average delay time.

Images