A method for predicting the flight plans of approaching flights for terminal area traffic simulation

Abstract

A method for predicting flight plans for approaching flight flows in terminal area traffic simulation is proposed: The method extracts trackpoints within the terminal area from a database, stores these trackpoints as trajectories based on flight numbers, and performs trajectories trimming and arrival / departure trajectory separation to form an approach trajectory set Γ. Flight flow identification is then performed: a symmetrical unidirectional great circle distance measurement model between trajectories is established, and a trajectory similarity matrix is ​​constructed. Based on the approach trajectory set Γ, the proportion of altitude layers and aircraft types used by flights crossing the terminal area boundary is calculated. The center trajectory of the flight flow is identified based on the flight flow trajectories, and turning feature points are further extracted from the center trajectory to construct planned flight routes. The flight traffic volume for each flight flow is statistically analyzed for each hour within a 24-hour period. Future flight plans are then predicted. This invention improves the efficiency of simulation planning, enhances the realism of airspace traffic simulation, increases the reliability of simulation evaluation results, and improves the scientific nature of terminal area traffic planning.

Description

Technical Field

[0001] This invention relates to a method for predicting flight plans for terminal area flight flows. In particular, it relates to a method for predicting arrival flight plans for terminal area traffic simulation. Background Technology

[0002] Terminal areas are characterized by dense convergence of arrival and departure traffic, with arrival routes exhibiting unrestricted characteristics, making them the most complex component of the national airspace system. With the rapid recovery and continued growth of air traffic demand, terminal area airspace congestion and flight delays have become increasingly severe, posing a bottleneck to the national airspace system.

[0003] To improve the safety and efficiency of arrival traffic operations in terminal areas, air traffic management requires frequent optimization of the planning and design of terminal airspace and controlled sectors. Given the safety and cost-effectiveness of computer simulation, airspace planning and design schemes often require verification and evaluation using simulation methods before implementation. The general process for air traffic simulation evaluation involves establishing a digital twin airspace model, compiling the annual arrival flight plan, generating corresponding traffic scenarios under the drive of an air traffic simulation engine, and statistically analyzing the simulation results to ultimately achieve a simulation evaluation of air traffic planning and design schemes.

[0004] Foreign countries recognized the important role of simulation evaluation software in aviation system planning and operational decision-making earlier and have developed several simulation software systems, such as TAAM, AirTop, SIMMOD and RAMS.

[0005] These simulation and evaluation software programs primarily function to simulate air traffic operation logic, aircraft 4D flight trajectories, air traffic control rules, analyze simulation results data, and provide graphical interfaces. However, they rarely address the scientific methods for generating flight flows, which are the starting point and input for simulations. In air traffic simulation work, flight plans input into the simulation software are often manually entered based on the experience of air traffic control experts, or by adding flights to the existing actual flight plans. For a long time, the problem of unscientific flight plans leading to deviations in simulation results has not received sufficient attention.

[0006] In general, current flight flow plan generation technology has the following defects and shortcomings:

[0007] 1. The flight plans for arriving and departing flights do not fully reflect the dynamic development and evolution of these flows, affecting the accuracy and correctness of the simulation evaluation results. Inputting flight plans that do not conform to real-world flight flow patterns into the simulation software may lead to inaccurate or even erroneous conclusions.

[0008] 2. The flight plans for incoming flights lack consideration of random factors and lack reasonable modeling of random events. The simulation based on the deterministic flight plans is an incomplete simulation, and the simulation results lack credibility.

[0009] 3. Flight plans used for simulation evaluation are characterized by large scale and numerous data items. The current method, which is mainly based on manual compilation, is labor-intensive, inefficient, and prone to errors.

[0010] 4. Since the planned approach routes in the terminal area are unrestricted routes, the flight plans based on fixed routes in the past affected the accuracy of the spatial distribution of flight flow and reduced the realism of air traffic simulation.

[0011] To improve the realism of the terminal area approach traffic simulation process, enhance the credibility of the simulation evaluation results, and improve the scientific nature of terminal area traffic planning, developing a method for automatically generating large-scale flight flow plans that conforms to the changing patterns of flight flow has become a major challenge in the field of airspace simulation evaluation. It is also an important component in developing an aviation operation simulation evaluation software system with independent intellectual property rights in my country. Summary of the Invention

[0012] The technical problem to be solved by the present invention is to overcome the shortcomings of the prior art and provide a method for predicting the flight plan of approaching flights for terminal area traffic simulation that conforms to the spatiotemporal variation law of actual flight flow.

[0013] The technical solution adopted in this invention is: a method for predicting the flight plans of approaching flights for terminal area traffic simulation, comprising the following steps:

[0014] 1) Perform trajectory data preprocessing: Extract the track points within the terminal area from the database, store the track points as trajectories according to the flight number, and perform trajectories trimming and arrival / departure trajectory separation processing to form the arrival trajectory set Γ;

[0015] 2) Flight flow identification: Establish a symmetrical unidirectional great circle distance measurement model between trajectories, further establish a trajectory similarity matrix, and use spectral clustering to cluster trajectories, thereby realizing flight flow identification;

[0016] 3) Based on the approach trajectory set Γ, calculate the composition ratio of the altitude layer and the composition ratio of the aircraft type when the flight crosses the boundary of the terminal area;

[0017] 4) Based on the identified flight flow trajectory, the K-nearest neighbor method is used to identify the center trajectory of the flight flow, and then the turning feature points are extracted from the center trajectory to realize the construction of the planned flight route;

[0018] 5) Calculate the flight traffic volume for each flight in one hour throughout a 24-hour period;

[0019] 6) Predict future flight plans: Based on the composition ratio of aircraft types operating flights, the composition ratio of altitude layers at which flights cross the boundary of the terminal area, the planned flight routes, and the flight traffic during the 24 time periods of a day, the elements of the future simulated flight plan are generated using a random sampling method, and finally combined to generate the future flight plan.

[0020] This invention discloses a method for predicting arrival flight flow in terminal area traffic simulation. It is a rapid method for generating large-scale flight plans that conforms to the spatiotemporal variation patterns of actual flight flows and is geared towards terminal area planning and evaluation. This invention improves the efficiency of simulation plan preparation, enhances the realism of airspace traffic simulation, increases the reliability of simulation evaluation results, and improves the scientific rigor of terminal area traffic planning, laying the foundation for developing aviation simulation software with independent intellectual property rights. This invention has the following beneficial effects:

[0021] 1. This invention creates a rapid generation method and process for large-scale flight flow flight plans that conforms to the spatiotemporal variation and composition patterns of actual flight flows and is geared towards terminal area planning and evaluation. This invention can improve the efficiency of simulation plan preparation, enhance the realism of airspace traffic simulation, increase the reliability of simulation evaluation results, and improve the scientific nature of terminal airspace traffic planning, laying a technical foundation for the development of aviation simulation software with independent intellectual property rights.

[0022] 2. This invention proposes a novel approach route generation method that accurately preserves trajectory shape features based on the degree of heading change. A symmetrical unidirectional great circle distance calculation method between trajectories is proposed, and spectral clustering is used for flight flow identification. Based on flight flow identification, the KNN method is used to identify the center trajectory. Trajectory segmentation is performed using the heading change of the center trajectory to obtain typical approach routes. Typical approach routes represent the true spatial distribution characteristics of air traffic, improving the realism of the simulation. This invention proposes a trajectory data-driven method for analyzing the time-varying and compositional characteristics of flight flows, establishing a uniform distribution model of flight composition and a negative exponential distribution model of aircraft tail spacing, laying the foundation for flight flow flight plan prediction based on flight flow characteristics.

[0023] 3. Flight schedule generation method based on random sampling. Based on the flight flow distribution model, a random sampling method is used to generate flight schedules that conform to the spatiotemporal and compositional characteristics of flight flows. This method ensures the similarity between the characteristics of future and current flight flows, improving the simulation realism and the efficiency of flight schedule generation. Attached Figure Description

[0024] Figure 1 This is a flowchart of an approach flight flow prediction method for terminal area traffic simulation according to the present invention;

[0025] Figure 2This is a schematic diagram of the entry trajectory;

[0026] Figure 3a This is a schematic diagram of the incoming flight flow (F1).

[0027] Figure 3b This is a schematic diagram of the incoming flight flow F2;

[0028] Figure 3c This is a schematic diagram of the incoming flight flow F3;

[0029] Figure 3d This is a diagram of the incoming flight flow F4;

[0030] Figure 3e This is a schematic diagram of the incoming flight flow (F5).

[0031] Figure 3f This is a schematic diagram of the incoming flight flow F6;

[0032] Figure 4 This is a schematic diagram of the planned flight path R4 for flight flow F4;

[0033] Figure 5 This is a schematic diagram of the planned flight routes R1 to R6;

[0034] Figure 6 A diagram showing the flight flow of each incoming flight during the period from 14:00 to 15:00;

[0035] Figure 7 A diagram illustrating the flight flow of the current inbound flight flow F4 at different times;

[0036] Figure 8 A diagram illustrating the inbound flight flow (F4) at different times over the next 5 years. Detailed Implementation

[0037] The following describes in detail, with reference to embodiments and accompanying drawings, a method for predicting the flight plans of arriving flights for terminal area traffic simulation according to the present invention.

[0038] This invention relates to numerous aviation-related technical terms. For ease of understanding, the aviation-related technical terms used in this application will first be introduced.

[0039] Air traffic control departments use ADS-B surveillance equipment to acquire real-time navigation data and changes, such as position, altitude, speed, and heading, of aircraft flying within the terminal area. The air traffic control automation system receives, processes, and displays this surveillance data, while simultaneously recording all or part of the aircraft's flight process. In the field of air traffic control applications, the concepts of flight track and trajectory are somewhat vague; this invention will first distinguish between flight track and trajectory.

[0040] A track refers to the spatial position, velocity, and heading characteristics of an aircraft recorded by monitoring sensors at a given time t, represented as a four-dimensional vector: P = (λ, φ, h, t). Where λ, φ, h, t... h and h represent the longitude, latitude and altitude of the spatial location, respectively, and t represents the recording time of the track point P;

[0041] A trajectory is the historical trace left by an aircraft's flight over a period of time. While a trajectory is actually a continuous curve, due to considerations of sensor data processing cycles and controller cognitive engineering, the aircraft flight trajectory data recorded in the air traffic control system is not continuous. A trajectory T consists of a series of discrete trajectories arranged chronologically, forming a time series of trajectories. The formal definition of trajectory T is: The dataset T of the aircraft's flight trajectory's track points is defined as a vector set T = (P1, P2, ..., P...). j ,..,P n ), where: j∈[1,n] is the time sequence number of the track points, and n is the total number of track points in the flight trajectory T.

[0042] The mathematical description of the set of arrival trajectories of all approaching aircraft in the terminal area over a certain period is as follows: Let Γ = {T1, T2, ..., T} be the set of aircraft trajectories formed by all approaching aircraft in the terminal area during a certain time period. i ,..,T N}, where each T i This represents an approach flight path, and N represents the total number of paths.

[0043] Approaching flight flow: Multiple approaching aircraft that fly in succession with basically the same flight path.

[0044] Flight plans: Flight plans used for traffic simulation within the terminal area include flight identification number, aircraft type, planned flight route, altitude level at which the arriving flight enters the terminal area, and arrival time. The flight plans of multiple arriving flights over a period of time constitute the initial conditions for the terminal area traffic simulation.

[0045] This invention discloses a method for predicting arrival flight flow flight plans for terminal area traffic simulation. It collects actual aircraft flight trajectory data from the current terminal area and identifies arrival flight flows using trajectory spectrum clustering. Based on the identified arrival flight flows, it analyzes the traffic flow variation characteristics and representative routes of each flow; it analyzes the spatiotemporal characteristics and component parameters of the flight flows; and uses a random sampling method to predict various elements of the flight plan through a uniform distribution probability model of flight routes, altitude layers, aircraft types, and a negative exponential distribution model of nose-distance. Finally, it synthesizes and generates the arrival flight flow flight plan.

[0046] like Figure 1As shown, the present invention provides a method for predicting the flight plans of approaching flights for terminal area traffic simulation, comprising the following steps:

[0047] 1) Trajectory data preprocessing: Although the acquired trajectory data has undergone correlation, noise reduction, and deduplication processing, it still contains a large amount of redundant and irrelevant trajectory information for analyzing incoming flight flow. Trajectory data preprocessing involves extracting trajectory points within the terminal area from the database, storing these points as trajectories based on flight numbers, and then trimming and separating the arrival and departure trajectories to form a trajectory set Γ. Specifically, this includes:

[0048] (1.1) Track clipping

[0049] The flight path data includes the entire flight process from takeoff to landing. First, the required flight path data is cropped. The polygon formed by the horizontal range of the airport terminal area is used as the flight path cropping area. Flight path data within the flight path cropping area is retained, and those flight paths outside the flight path cropping area are removed.

[0050] (1.2) Track cleaning and differentiation of arrival and departure flight tracks

[0051] The terminal area also includes flight path data for numerous overflying flights. Additionally, when general aviation flights are present in the terminal area, if the aircraft is equipped with an ADS-B transmitter, the monitoring system will also record the aircraft's flight path. This invention can selectively exclude overflying flights and general aviation flight paths based on whether the aircraft uses the airport runway. Furthermore, it can differentiate between arriving and departing flights using information provided by flight plan data, thereby distinguishing between arrival and departure paths and retaining the flight path data of arriving commercial aircraft within the terminal area, forming an arrival path set Γ.

[0052] 2) Flight flow identification: Establish a symmetric unidirectional great circle distance metric model between trajectories, further establish a trajectory similarity matrix, and use spectral clustering to cluster trajectories, thereby achieving flight flow identification; specifically including:

[0053] (2.1) Calculate the symmetrical one-way great circle distance between trajectories

[0054] Previous methods for measuring inter-track distances used in clustering have largely relied on Euclidean geometry. However, in aviation and maritime applications, the Earth's spherical nature must be considered. Traditional methods first employ Mercator map projection to convert the latitude and longitude coordinates of the track into map coordinates, then process them using Euclidean geometric principles. Straight lines on a Mercator-projected map are called isoazimuths, representing the compass heading. However, on a sphere, such isoazimuths do not represent the shortest distance between two points. Instead, the shortest distance between two points is the great circle distance. Therefore, this invention proposes a method for measuring inter-track distances based on great circle distance.

[0055] (2.1.1) Calculate the great circle distance between two points

[0056] Let the first waypoint O1(λ1,φ1) and the second waypoint O2(λ2,φ2) be defined. The shortest distance between the two waypoints is the great circle distance passing through the two points. Considering that errors are easily generated when the two points are close to each other, the Havesine function H(θ) with angle θ is constructed as shown in equation (1):

[0057]

[0058] Where λ1 and φ1 are the longitude and latitude of the first track point O1, and λ2 and φ2 are the longitude and latitude of the second track point O2, thus obtaining the great circle distance between the two points. As in equation (2):

[0059]

[0060] Where R0 is the average radius of the Earth.

[0061] (2.1.2) Calculate the spherical distance from the point to the trajectory segment.

[0062] Let the arc of the b-th segment of trajectory T0 be... Flight segment arc Composed of endpoints P1 and P2, the arc from the first waypoint O1 to the segment is... Great circle distance The calculation is as shown in equation (3):

[0063]

[0064] (2.1.3) Calculate the spherical distance from the point to the trajectory T0.

[0065] The distance d from the first waypoint O1 to the trajectory T0 pT (O1,T0) is the minimum distance from the first waypoint O1 to all the arc segments that make up the trajectory T0, as shown in equation (4):

[0066]

[0067] Where K2 represents the number of track points for T2;

[0068] (2.1.4) Calculate the symmetrical one-way great circle distance from trajectory T1 to trajectory T2.

[0069] Given trajectories T1 and T2, the one-way great circle distance from trajectory Y1 to trajectory T2. The calculation is as shown in equation (5):

[0070]

[0071] Where K1 represents the number of track points in trajectory T1. This represents the k-th point on trajectory T1. Furthermore, the symmetric one-way great circle distance from trajectory T1 to trajectory T2 is obtained. As shown in equation (6):

[0072]

[0073] Calculate the symmetric one-way great circle distance between each pair of trajectories in the set of approach trajectories Γ, and generate the distance matrix D between trajectories D = [d i,j ] N×N , where the element d in the i-th row and j-th column i,j As shown in equation (7):

[0074]

[0075] Where N is the number of trajectories contained in the set of entry trajectories Γ.

[0076] (2.2) Flight flow identification based on spectral clustering

[0077] Trajectory spectral clustering treats a trajectory as a point in a high-dimensional space, with the symmetric unidirectional great circle distance between trajectories constituting the weight of the edges between them. This high-dimensional space, containing points and weighted edges, forms a graph G, represented by a trajectory similarity matrix S. By segmenting graph G, the weights of edges between different subgraphs are minimized while the weights of edges within subgraphs are maximized, thus achieving the goal of clustering.

[0078] (2.2.1) Construct the similarity matrix W between trajectories = [w i,j ] N×N Calculate the trajectory similarity w in the i-th row and j-th column. i,j As shown in equation (8):

[0079]

[0080] Where σ is a parameter that controls the width of the neighborhood;

[0081] (2.2.2) Estimate the number of clusters in the set of incoming trajectories Γ by using the Laplace eigenvalues ​​of the similarity matrix W between trajectories; specifically, call the spectralcluster function in MATLAB 2022b or later, input a maximum cluster estimate k1, where k1 is an integer greater than 1, and the spectralcluster function outputs the eigenvalue matrix E to estimate the number of clusters;

[0082] In the eigenvalue matrix E, the number of the first few eigenvalues ​​equal to zero reflects the number of clusters in the trajectory data; assuming that the first k2 eigenvalues ​​in E are equal to 0, the number of clusters in the set of incoming trajectories is k2, where 0≤k2≤k1;

[0083] (2.2.3) Call the MATLAB function spectralcluster of the spectral clustering algorithm, and take the trajectory similarity matrix W and the number of clusters k2 as input to obtain an N×1 vector idx1 representing the classification of each trajectory;

[0084] Vector idx1 represents the classification of each trajectory in the set of approach trajectories Γ. The results are categorized into Z classes, meaning the trajectories in the set of approach trajectories Γ are divided into F1, F2, ..., F... z ,..,F Z By dividing the flight flow into Z subsets, flight flow identification is achieved.

[0085] 3) Based on the approach trajectory set Γ, calculate the composition ratio of altitude layers and aircraft types when flights cross the boundary of the terminal area.

[0086] (3.1) Determine the proportion of high-rise floors used.

[0087] (3.1.1) Set the height layer set A = {a q}, q=1,2,...,a max There are a total of max There are several altitude layers, where q is the altitude layer number;

[0088] (3.1.2) For the z-th flight flow F z Choose any trajectory T r Calculate trajectory T r The altitude h of the first waypoint start With each height layer a in the height layer set A q The distance is taken as the height layer corresponding to the minimum distance, and the trajectory T is selected. r The altitude layer used when flying into the terminal area;

[0089] (3.1.3) Traverse the z-th flight flow F z For all trajectories in the dataset, obtain the height layer used for all trajectories and calculate the height layer 'a'.q Number of trajectories This indicates that in flight flow F z The number of trajectories using the q-th height layer;

[0090] (3.1.4) Statistical analysis of the flow of flight F of the z-th flight z The number of trajectories using the q-th altitude layer accounts for a certain percentage of the number of flight flows in the z-th flight flow. z The proportion of all trajectories is used to obtain the flight flow F of the z-th flight. z Using the proportion distribution r1(z,q) of the q-th height layer Where, N z Indicates the flow of flight z. z The total number of all trajectories;

[0091] (3.2) Determine the usage ratio of different machine models

[0092] (3.2.1) Count all aircraft types in the arrival trajectory set Γ, and set the aircraft type set C = {c m}, m=1,2,...,c max ;

[0093] (3.2.2) Count the number N trajectories using the m-th model. m ;

[0094] (3.2.3) Calculate the usage ratio r2(m) of the m-th model.

[0095] Where N represents the total number of aircraft in the approach trajectory set Γ.

[0096] 4) Based on the identified flight flow trajectory, the K-nearest neighbor method is used to identify the center trajectory of the flight flow, and then the turning feature points are extracted from the center trajectory to realize the construction of the planned flight route;

[0097] (4.1) Identify the trajectory of the flight flow center

[0098] (4.1.1) Select the z-th flight flow F z The number of trajectories is N z Set the number of nearest neighbors of the trajectory involved in the calculation to k3;

[0099] (4.1.2) Calculate the flow F of the z-th flight according to formula (7). z Trajectory T in r With Flight Flow F z Other trajectories T v distance d r,v And generate a distance matrix.

[0100] (4.1.3) Calculate the trajectory Tr The average distance to its k3 nearest neighbors

[0101] Sort all elements of the distance matrix D1 in ascending order, and calculate the average of the first k3 numbers, denoted as . Represents trajectory T r The average distance to its k3 nearest neighbors.

[0102] (4.1.4) Traverse the z-th flight flow F z For all trajectories, calculate the average distance of the k3 nearest neighbors for each trajectory, and generate the z-th flight flow F. z The average distance of the k3 nearest neighbors of all trajectories vector

[0103] (4.1.5) Using MATLAB's K-centroid clustering function kmedoids, the z-th flight flow F z Trajectories are divided into two categories: mainstream trajectories and abnormal trajectories, as shown in equation (9):

[0104]

[0105] idx2 represents N of the two types of trajectories. z A vector of size 1, N z It is the zth flight flow F z The number of included trajectories; Q is two numbers, the first number Q(1) represents the z-th flight flow F z The average distance of the k3 nearest neighbors of the center trajectory of the mainstream trajectory The second number Q(2) represents the flow of flight F of the z-th flight. z The average distance of the k3 nearest neighbors of the center trajectory of the anomalous trajectory Through the vector in S4.1.4 Found The corresponding trajectory number is c1, with T c1 As the central trajectory of the mainstream trajectory, and also as the z-th flight flow F z The central trajectory.

[0106] (4.1.6) Repeat steps (4.1.1) to (S4.1.5) to obtain the center trajectory of all incoming flight flows;

[0107] (4.2) Based on the center trajectory T c1 Generate planned flight routes

[0108] The central trajectory is formed by a long period of straight flight and a short period of turning flight, resulting in a large amount of data redundancy, which cannot be directly used in the navigation database. This invention proposes a new approach route generation method that accurately preserves the trajectory shape characteristics based on the degree of heading change.

[0109] (4.2.1) Obtain the flow F of the z-th flight. z Heading change value of the center track

[0110] Let the center trajectory T c1 There are two adjacent waypoints P. k-1 and P k Their heading changes are shown in equation (10):

[0111] Δφ k =φ k -φ k-1 (10)

[0112] Where, φ k This indicates that the aircraft is at track point P. k The heading at that time, φ k-1 This indicates that the aircraft is at track point P. k-1 The heading at time, Δφ k This indicates that the aircraft is at track point P. k-1 Fly to waypoint P k The change in course that occurs over time.

[0113] In navigation, the range of track angle variation is typically set to φ∈(0, 360], and is taken as an integer. Aircraft turns are categorized into left turns and right turns. When turning right, the track angle increases; when turning left, the track angle decreases. Since heading in the due north direction is defined as 360 degrees, when an aircraft crosses 360 degrees and turns right, although the change in heading is continuous in physical space, headings greater than 360 degrees are still represented in navigation. Interval. Δφ needs to be considered. k Make the correction as shown in equation (11):

[0114]

[0115] If the cumulative heading change occurs within Δk consecutive waypoints... As in equation (12):

[0116]

[0117] (4.2.2) Save the characteristic track points of the center trajectory

[0118] If Δφ k or If the value exceeds a given threshold, the center trajectory is considered to have turned, and the track point corresponding to each turn is recorded as a feature track point P. xFurthermore, the first and last track points of the central trajectory are also used as feature track points, and they are arranged and saved in order according to the order of each feature track point on the central trajectory, while the remaining track points are removed.

[0119] (4.2.3) The sequentially arranged characteristic waypoints form the z-th flight flow F. z Planned flight route R z =(P1,P2,..,P x ,,..,P X X represents the number of feature track points;

[0120] 5) Calculate the flight traffic volume for each flight in one hour throughout a 24-hour period;

[0121] (5.1) Let the statistical time period be 1 hour, denoted as S, S = [t start ,t end ], t start Let t be the starting time. end This is the termination time.

[0122] (5.2) Obtain the flow of the z-th flight F z Trajectory T in r time series of flight paths S time =[t1,t2,..t k ,..t K ], where K is the trajectory T r Number of track points included;

[0123] (S5.3) Determine S time Does it contain at least one track time t? k ∈[t start ,t end If it contains, then the trajectory T is considered to be... r Within the time range S; if not included, the trajectory T is considered to be... r Not within the time range S;

[0124] (5.4) Traverse the z-th flight flow F z The total number of trajectories within the time range S will be used as the flight flow F for the z-th flight in the S-th hour. z Flight traffic

[0125] (S5.5) Obtain the flow of the z-th flight respectively. z Flight traffic in China during 24 time periods.

[0126] 6) Predicting Future Flight Plans: Based on the aircraft type composition ratio, the altitude layer composition ratio of flights crossing the terminal area boundary, planned flight routes, and flight traffic during the 24 time periods of the day, random sampling is used to generate elements for future simulated flight plans, which are then combined to generate the final future flight plan; including:

[0127] (6.1) Hourly traffic flow forecast for air routes

[0128] (6.1.1) Let the average annual growth rate of flight volume be α, and the forecast year be the next y years, where y is an integer;

[0129] (6.1.2) Planned flight route R z Flight traffic λz during the Sth time period of a 24-hour day in the future year y. S With the current z-th flight flow F z Flight traffic during the corresponding time period The relationship between them is shown in equation (13):

[0130]

[0131] (6.1.3) Repeat step (6.1.2) to predict the flight flow for the z-th flight in the next y-year period. z Flight traffic at different times of the day 24 hours

[0132] (6.2) Constructing flight plans

[0133] (6.2.1) Take the flow F of the z-th flight. z Planned flight route R z Flight traffic in the Sth time period create There are several simulated flight plans, where the p-th simulated flight plan is denoted as FlightPlan(p). The elements of a simulated flight plan (FlightPlan(p)) include the identification number (FlightPlan(p).Id), the planned flight route (FlightPlan(p).Route), the aircraft type (FlightPlan(p).Type), the altitude (FlightPlan(p).Attitude), and the time of occurrence (FlightPlan(p).Time).

[0134] (6.2.2) Let p be the identifier of the simulated flight plan FlightPlan(p), and let the z-th flight flow F z Planned flight route R z The planned flight path for the simulated flight plan FlightPlan(p);

[0135] (6.2.3) Using random sampling, generate the altitude layers for each simulated flight plan (FlightPlan(p)); including:

[0136] (a) Generation A random number between 0 and 1, following a uniform distribution.

[0137] (b) If The simulated flight plan FlightPlan(p) uses the l-th altitude layer (FlightPlan(p).Altitude); where r1(z,q) represents the z-th flight flow. z Use the proportional distribution of the qth height layer;

[0138] (c) Iterate through all random numbers Generate the corresponding usage altitude layer for each simulated flight plan FlightPlan(p);

[0139] (6.2.4) Using random sampling, generate the aircraft type for each simulated flight plan FlightPlan(p), including:

[0140] (a) Generation A random number between 0 and 1, following a uniform distribution.

[0141] (b) If Then, the aircraft type FlightPlan(p).Type in the simulated flight plan FlightPlan(p) is the g-th aircraft type; where r2(m) represents the proportion of the m-th aircraft type used;

[0142] (c) Iterate through all random numbers Generate the corresponding aircraft type for each simulated flight plan FlightPlan(p);

[0143] (6.2.5) Using a random sampling method, generate the occurrence time for each simulated flight plan FlightPlan(p);

[0144] (a) Generation A random number between 0 and 1, following a uniform distribution.

[0145] (b)Δt p Let Δt be the time interval between two consecutively flying aircraft, following a negative exponential distribution. p Calculate as shown in equation (14):

[0146]

[0147] Where τ is the set time interval, representing the minimum safe time interval between two aircraft flying consecutively in the flight flow.

[0148] (c) Set the occurrence time of the first simulated flight plan FlightPlan(1) to 0 seconds, and calculate the occurrence time of all flights in sequence according to formula (15):

[0149] FlightPlan(p+1).Time=FlightPlan(p).Time+Δt p (15)

[0150] The arrival time of an incoming flight indicates the moment it enters the terminal area boundary;

[0151] (6.2.6) Sort all the flights generated in the Sth time period according to their appearance time and combine them into the arrival and departure flight operation simulation plan for the Sth time period.

[0152] The following are specific examples:

[0153] A specific example of the method for predicting arrival flight plans for terminal area traffic simulation according to the present invention includes six steps: trajectory data preprocessing, flight flow identification, aircraft type composition ratio calculation, flight altitude layer composition ratio calculation for crossing the terminal area boundary, generation of planned flight routes, statistical analysis of hourly flight flow, and simulation flight plan prediction. Figure 1 As shown.

[0154] 1) Trajectory data acquisition and preprocessing

[0155] (1.1) Obtaining trajectory data

[0156] This embodiment acquired ADS-B flight track data of the area surrounding Chengdu Shuangliu International Airport on a certain day. The flight track data was recorded in chronological order at 10-second intervals, representing the flight track data of all aircraft within the receiving range, which can be represented as P. k =(λ k ,φ k ,h k ,t k ), where λ k , and h k Representing track P respectively k The longitude, latitude, and altitude of t k Indicates track P k The recorded moment.

[0157] The flight path data of the corresponding flight is extracted according to the flight number and arranged in chronological order to form the flight path T.

[0158] (1.2) Trajectory clipping

[0159] The polygon formed by the horizontal range of the Chengdu terminal area is used as the clipping area. Tracks within the clipping area are retained, while tracks outside the clipping area are removed.

[0160] (1.3) Track cleaning and separation of arrival and departure flight tracks

[0161] This invention eliminates overflight flights and general aviation trajectory data based on whether the aircraft uses the airport runway. Simultaneously, it distinguishes between arriving and departing flights using information provided by flight plan data, thereby achieving separation of arrival and departure trajectories.

[0162] After the above operations, this embodiment has successfully retained the flight trajectory data of approaching commercial aircraft within the airspace surrounding Shuangliu Airport, obtaining N = 407 approach flight trajectory data, which are saved as an approach trajectory set Γ, Γ = {T1, T2, ..., T...} i ,..,T 407}

[0163] Processed arrival trajectories from ADS-B around a certain airport on a certain day, as shown below. Figure 2 As shown.

[0164] 2) Conduct flight flow identification

[0165] (2.1) Calculation of symmetrical unidirectional great circle distance between trajectories

[0166] Taking the approach trajectory dataset Γ as an example, calculate the symmetrical unidirectional great circle distance between the trajectories. For example, extract two trajectories from Γ. and T5={P1,P 2, ,...,P n}, Let represent the k-th point on T1. m represents the number of track points contained in trajectory T1 (m = |T1| = 130), and n represents the number of track points contained in trajectory T5 (n = |T5| = 135). The calculations are demonstrated using steps (2.1.1) to (2.1.4) of the "Approach Flight Flow Flight Plan Prediction Method for Terminal Area Traffic Simulation" of this invention, combining the data from trajectories T1 and T5.

[0167] The point is calculated using equation (3) in the method according to the present invention. The spherical distance to any segment on trajectory T5 is obtained by using equation (4) in the method of this invention to obtain the track points on T1. The distance to trajectory T5 is calculated by traversing all waypoints on trajectory T1. The distances to trajectory T5 are summed and averaged to obtain the great circle distance from T1 to T5 according to equation (5) in the method of the present invention. Great circle distance from T5 to T1 Further, according to equation (6) of the method of the present invention, the symmetrical unidirectional great circle distance from trajectory T1 to trajectory T5 is obtained.

[0168] According to equation (7) of the method of the present invention, the symmetric one-way great circle distance between any two trajectories in the trajectory set Γ is calculated, and the distance matrix between trajectories D = [d i,j ] 407×407 Where i = 1, 2, ..., 407; j = 1, 2, ..., 407. The distance matrix between the trajectories of arriving flights in Γ is obtained through calculation. Part of matrix D is:

[0169]

[0170] (2.2) Flight flow identification based on spectral clustering

[0171] Taking the distance matrix D between approach trajectories obtained in the previous step as an example, we perform flight flow identification calculation based on spectral clustering.

[0172] (1) Construct the similarity matrix between trajectories W = [w i,j ] 407×407 The similarity w between different trajectories is calculated using equation (8) in the method of the present invention. i,j The parameter σ, which controls the neighborhood width, is 8.8645.

[0173] The calculated similarity matrix W between the approach trajectories is:

[0174]

[0175] (2) Estimating the number of clusters. The number of possible clusters in the trajectory set Γ is estimated using the Laplace eigenvalues ​​of the similarity matrix W between trajectories. The spectralcluster function in MATLAB 2022b or later is called, with a maximum cluster estimate k1 = 10 as input. The spectralcluster function outputs the eigenvalue matrix E that estimates the number of clusters. The first 6 eigenvalues ​​in E are equal to 0, so the estimated number of clusters in the trajectory set is k2 = 6.

[0176] (3) Call the function spectralcluster in MATLAB 2022b or later versions of the spectral clustering algorithm, and take the trajectory similarity matrix W and the estimated number of clusters k2 = 6 as inputs to obtain the 407×1 vector idx1 of each trajectory classification.

[0177] (4) Vector idx1 gives the classification of each trajectory in the set of approach trajectories Γ, resulting in 6 clusters of approach flight flows F. i ={F1,F2,F3,F4,F5,F6}, thereby achieving the identification of incoming flight flow, such as Figure 3a , Figure 3b , Figure 3c , Figure 3d , Figure 3e , Figure 3f As shown.

[0178] 3) Based on the approach trajectory set Γ, calculate the composition ratio of aircraft types and the composition ratio of altitude layers when flying over the terminal area boundary.

[0179] (3.1) Determine the proportion of flight flow altitude layers used.

[0180] Taking the incoming flight flow F4 as an example, determine the proportion of altitude layers used when the flight crosses the boundary of the terminal area.

[0181] (1) Arrival altitude layer set in Shuangliu Airport terminal area

[0182] A={3600,3900,4200,4500,4800,5100,5400,5700,6000}

[0183] (2) Take any trajectory T in the incoming flight flow F4 r Calculate T r The altitude of the first waypoint and the distance between each altitude layer in A are used to determine the altitude layer corresponding to the minimum distance when the trajectory crosses the boundary of the terminal area.

[0184] (3) Traverse all trajectories of flight flow F4, obtain the altitude layers of all trajectories that cross the terminal area boundary, and count the number of aircraft using each altitude layer. For example, the count shows that 61 aircraft in flight flow F4 use altitude layer 3600.

[0185] (4) Calculate the proportion of each altitude layer used to the total number of trajectories in flight flow F4 to obtain the altitude layer usage ratio. For example, the proportion of trajectories using the 3600 altitude layer to the total number of trajectories in flight flow F4 is 0.5495. The altitude layers used and the proportion of each altitude layer for all trajectories in arriving flight flow F4 are shown in Table 1.

[0186] Table 1. Air Levels Used by Flight F4 and Percentage of Each Air Level in Arrival Flight Flow.

[0187] Height layer / m Quantity used / units Proportion 3600 61 0.5495 3900 32 0.2883 4200 13 0.1171 4500 4 0.0360 4800 1 0.0090 5100 0 0 5400 0 0 5700 0 0 6000 0 0

[0188] Iterate through all flight flows to obtain the usage ratio of each altitude layer for different flight flows.

[0189] (3.2) Determine the usage ratio of different machine models

[0190] The aircraft types in the approach trajectory set Γ are counted to obtain the aircraft type set C = {"A319","A320","A321","A330","B737","B747","B773"}. The number of aircraft of each type used is counted and the usage ratio of each type is calculated, as shown in Table 2.

[0191] Table 2. Aircraft Types and Percentage of Each Type on the Arrival Trajectory

[0192] model Quantity used / units Proportion A319 41 0.1007 A320 115 0.2826 A321 71 0.1744 A330 107 0.2629 B737 13 0.0319 B747 57 0.1400 B773 3 0.0074

[0193] 4) Based on the identified flight flow trajectories, the K-nearest neighbor method is used to identify the center trajectory of the flight flow, and then the turning feature points are extracted from the center trajectory to realize the construction of the planned flight route;

[0194] Taking the arrival flight flow as an example, we conduct a spatial distribution analysis of the flight flow and obtain the planned flight routes of the flight flow.

[0195] (4.1) Identify the trajectory of the flight flow center

[0196] (1) Select the incoming flight flow F4, which contains N6 = 298 trajectories; set the number of nearest neighbors of the trajectories participating in the calculation to k3 = 5;

[0197] (2) Following steps (4.1.2) to (4.1.5) of the method of this invention, the center trajectory number c1 = 111 is obtained, that is, the center trajectory of flight flow F4 is T. 111 .

[0198] (3) Traverse all incoming flight flows to obtain the center trajectory of all incoming flight flows; the center trajectory numbers of different incoming flight flows and the number of trajectories they contain are shown in Table 3:

[0199] Table 3 Center Trajectory of Arrival Flight Flow

[0200] Flight Flow Categories 1 2 3 4 5 6 Center trajectory number 401 399 259 298 129 59 Trajectory Count 28 76 61 111 49 82

[0201] (4.2) Generate the expected flight path based on the center trajectory

[0202] (1) The center trajectory T of the 4th flight flow F4 111 For example, calculate T 111 Change in heading of adjacent tracks Δφ k and the cumulative heading change value of 10 consecutive waypoints If Δφ k >1 or Then the central trajectory T is considered to be 111 When a turn occurs, the corresponding waypoint at each turn is recorded as a characteristic waypoint P. xFurthermore, the first and last track points of the central trajectory are also used as feature track points, and they are arranged and saved in order according to the order of each feature track point on the central trajectory. The remaining track points are removed, and a total of 13 feature track points are retained.

[0203] (2) The 13 characteristic waypoints arranged in sequence constitute the planned flight route R4 of the fourth flight flow F4 = (P1, P2, ..., P x ,,..,P 13 ),like Figure 4 As shown.

[0204] (3) Traverse the trajectory of each incoming flight flow center to obtain the planned flight routes R1, R2, R3, R5, and R6, such as Figure 5 As shown.

[0205] 5) Calculate the flight traffic volume for each flight in one hour throughout a 24-hour period;

[0206] (1) Taking 14:00-15:00 as an example. Taking 00:00 as the origin and seconds as the unit, the time statistics range is S0 = [50400, 54000];

[0207] (2) Taking the incoming flight flow F4 as an example, for trajectory T in F4 r Obtain its trajectory time series S time =[t1,t2,..t k ,t K In the image, K represents the trajectory T. r The number of track points included, where k represents the k-th track;

[0208] (3) Determine S time Does it contain at least one track time t? k ∈[50400,54000]. If it contains, then the trajectory T is considered to be... r Within the time range S0; if it does not exist, then the trajectory T is considered to be within the time range S0. r Not within the time range S0;

[0209] (4) Traverse all trajectories in flight flow F4, count the number of trajectories within the S0 time period as 6, and use this as the flight flow of the 4th flight flow F4 in the 14th hour.

[0210] (5) Calculate the flow of each arriving flight during the S0 time period, such as Figure 6 As shown. Statistics are compiled for all time periods to obtain flight traffic for flight flow F4 at different times, as follows. Figure 7 As shown.

[0211] 6) Predicting Future Flight Plans: Based on the aircraft type composition ratio, the altitude layer composition ratio of flights crossing the terminal area boundary, planned flight routes, and flight traffic during the 24 time periods of the day, random sampling is used to generate elements for future simulated flight plans, which are then combined to generate the final future flight plan; including:

[0212] (6.1) Hourly traffic flow forecast for air routes

[0213] (1) Assume that the average annual growth rate of flight volume in the terminal area is α = 5%, and the target year is the flight volume y = 5 years from now.

[0214] (2) Take flight path R4, time period S0 = [50400, 54000]. The traffic flow for flight path R4 during the time period 14:00-15:00. The flow rate of the future route R4 during the corresponding time period is calculated using equation (13) in the method of the present invention.

[0215] (3) Repeat (2) to predict the flight flow of F4 for the next 5 years, for each time period of the 24 hours of the day, such as Figure 8 As shown.

[0216] (6.2) Constructing a simulated flight plan

[0217] Take route R4, time period S0 = [50400, 54000], and flight traffic. Eight simulated flight plans are created, and each simulated flight plan is assigned an identification number. The flight path of the simulated flight plan is R4. According to step (6.2.3) of the method of the present invention, the altitude layer to be used for each simulated flight plan is generated. According to step (6.2.4) of the method of the present invention, the aircraft type to be operated for each simulated flight plan is generated. According to step (6.2.5) of the method of the present invention, the time interval τ = 60 seconds is set, and the plan occurrence time is generated for each simulated flight plan.

[0218] For time period S0 = [50400, 54000], the future flight schedule for route R4 is shown in Table 4:

[0219] Table 4 Flight plan for route R4 during the 14:00-15:00 time period

[0220]

[0221] Iterate through the remaining planned flight routes R1, R2, R3, R5, and R6 to generate simulated flight plans for the corresponding routes. Sort all the flights generated within this time period according to their occurrence time and combine them to form the future simulated arrival flight plan for this time period, as shown in Table 5.

[0222] Table 5 Flight schedule for flights arriving at the airport between 14:00 and 15:00

[0223]

[0224]

Claims

1. A method for predicting the flight plans of arriving flights for terminal area traffic simulation, characterized in that, Includes the following steps: 1) Perform trajectory data preprocessing: Extract waypoints within the terminal area from the database, store the waypoints as trajectories according to the flight number, and perform trajectories trimming and arrival / departure trajectory separation to form an arrival trajectory set. ; 2) Flight flow identification: Establish a symmetrical unidirectional great circle distance measurement model between trajectories, further establish a trajectory similarity matrix, and use spectral clustering to cluster trajectories, thereby realizing flight flow identification; 3) Based on the set of entry trajectories Calculate the composition ratio of the altitude layers and the composition ratio of the aircraft types when the flight crosses the boundary of the terminal area; 4) Based on the identified flight flow trajectory, the K-nearest neighbor method is used to identify the center trajectory of the flight flow. Further, turning feature points are extracted from the center trajectory to construct the planned flight route; including: (4.1) Identify the trajectory of the flight flow center (4.1.1) Select the z-th flight flow The number of trajectories included is Set the number of nearest neighbors of the trajectory involved in the calculation to k3; (4.1.2) Calculate the flow of the z-th flight according to formula (7). Trajectory in With flight flow Other trajectories distance And generate a distance matrix. ; (4.1.3) Calculate the trajectory The average distance to its k3 nearest neighbors For distance matrix Sort all elements in the array in ascending order, and calculate the average of the first k3 numbers, denoted as . , Representing the trajectory The average distance to its k3 nearest neighbors; (4.1.4) Traverse the z-th flight flow For all trajectories, calculate the average distance of the k3 nearest neighbors for each trajectory, and generate the z-th flight flow. The average distance of the k3 nearest neighbors of all trajectories vector ; (4.1.5) Use MATLAB's K-centroid clustering function kmedoids to cluster the z-th flight flow. Trajectories are divided into two categories: mainstream trajectories and abnormal trajectories, as shown in equation (9): [idx2, Q]=kmedoids( ,2) (9); idx2 represents the two types of trajectories The vector, N z It is the zth flight. The number of included trajectories; Q is two numbers, the first number Q(1) represents the z-th flight flow. The average distance of the k3 nearest neighbors of the center trajectory of the mainstream trajectory The second number Q(2) represents the flow of the z-th flight. The average distance of the k3 nearest neighbors of the center trajectory of the anomalous trajectory Through the vector in S4.1.4 Found The corresponding trajectory number is c1, with As the central trajectory of the mainstream trajectory, and also as the z-th flight flow The central trajectory; (4.1.6) Repeat steps (4.1.1) to (S4.1.5) to obtain the center trajectory of all arriving and departing flight flows; (4.2) Based on the center trajectory Generate planned flight routes (4.2.1) Obtain the flow of the z-th flight Heading change value of the center track Let the center trajectory T c1 There are two adjacent waypoints P. k-1 and P k Their heading changes are shown in equation (10): （10）； in, This indicates that the aircraft is at track point P. k The course at that time This indicates that the aircraft is at track point P. k-1 The course at that time This indicates that the aircraft is at track point P. k-1 Fly to waypoint P k The change in course that occurs at that time; right Make the correction as shown in equation (11): （11）； If in Cumulative heading change within a series of consecutive waypoints As shown in equation (12): （12）； (4.2.2) Preserve the characteristic track points of the center trajectory if or If the value exceeds a given threshold, the center trajectory is considered to have turned, and the track point corresponding to each turn is recorded as a feature track point. Furthermore, the first and last track points of the central trajectory are also used as feature track points, and they are arranged and saved in order according to the order of each feature track point on the central trajectory, while the remaining track points are removed. (4.2.3) The sequentially arranged characteristic waypoints form the z-th flight flow. Planned flight routes X represents the number of feature track points; 5) Calculate the flight traffic volume for each flight in one hour throughout a 24-hour period; 6) Predict future flight plans: Based on the composition ratio of aircraft types operating flights, the composition ratio of altitude layers at which flights cross the boundary of the terminal area, the planned flight routes, and the flight traffic during the 24 time periods of a day, the elements of the future simulated flight plan are generated using a random sampling method, and finally combined to generate the future flight plan.

2. The method for predicting arrival flight plans for terminal area traffic simulation according to claim 1, characterized in that, Step 1) includes: (1.1) Trajectory clipping The flight path data includes the entire flight process from takeoff to landing. First, the required flight path data is cropped. The polygon formed by the horizontal range of the airport terminal area is used as the flight path cropping area. Flight path data within the flight path cropping area is retained, and those flight paths outside the flight path cropping area are removed. (1.2) Track cleaning and differentiation of arrival and departure flight tracks By selectively eliminating overflying flights and general aviation flight paths based on whether the aircraft uses the airport runway, and by using information provided by flight plan data to differentiate between arriving and departing flights, the flight path data of arriving commercial aircraft within the terminal area is retained to form an arrival path set. .

3. The method for predicting arrival flight plans for terminal area traffic simulation according to claim 1, characterized in that, Step 2) includes: (2.1) Calculate the symmetrical unidirectional great circle distance between the trajectories: (2.1.1) Calculate the great circle distance between two points. Set the first waypoint Second track point The shortest distance between two waypoints is the great circle distance passing through the two points; considering that errors can easily occur when the two points are close to each other, the angle is constructed as follows: Havesine function As shown in equation (1): (1)； in, and First waypoint Longitude and latitude and Second waypoint The longitude and latitude are used to obtain the great circle distance between the two points. As shown in equation (2): (2)； in, 0 represents the average radius of the Earth; (2.1.2) Calculate the spherical distance from the point to the trajectory segment. Let the trajectory The b-th segment arc Flight segment arc From the endpoint 1 and If it consists of 2, then the first waypoint to flight segment arc Great circle distance The calculation is as shown in equation (3): (3)； (2.1.3) Calculate the point to the trajectory spherical distance First track point To trajectory distance It is the first track point To form a trajectory The minimum distance among all arc segments, as shown in equation (4): (4)； Where K2 represents the number of track points in T2; (2.1.4) Calculate the symmetrical one-way great circle distance from trajectory T1 to trajectory T2. Set trajectory 1 and 2. From the trajectory 1 to trajectory 2 One-way great circle distance The calculation is as shown in equation (5): (5)； Where K1 represents the number of track points in trajectory T1. Representing the trajectory The first on 1 From these points, we can further obtain the symmetrical one-way great circle distance from trajectory T1 to trajectory T2. As shown in equation (6): （6）； Calculate the set of entry trajectories The symmetric unidirectional great circle distance between each pair of trajectories generates the distance matrix between trajectories. The element in the i-th row and j-th column As shown in equation (7): （7） ； Where N is the set of entry trajectories. The number of trajectories included; (2.2) Flight flow identification based on spectral clustering: (2.2.1) Constructing the similarity matrix between trajectories Calculate the trajectory similarity in the i-th row and j-th column. As shown in equation (8): (8)； Where σ is a parameter that controls the width of the neighborhood; (2.2.2) Estimate the set of approach trajectories by using the Laplace eigenvalues ​​of the similarity matrix W between trajectories. The number of clusters is estimated by calling the spectralcluster function in MATLAB 2022b or later, taking a maximum cluster estimate k1 as input (k1 is an integer greater than 1), and the spectralcluster function outputs the eigenvalue matrix E to estimate the number of clusters. In the eigenvalue matrix E, the number of the first few eigenvalues ​​equal to zero reflects the number of clusters in the trajectory data; assuming that the first k² eigenvalues ​​in E are equal to 0, the number of clusters in the set of entering trajectories is k², where... ; (2.2.3) Call the MATLAB function `spectralcluster` for the spectral clustering algorithm, taking the trajectory similarity matrix W and the number of clusters k2 as input, to obtain the representation of the classification of each trajectory. The vector idx1; Vector idx1 gives the set of approach trajectories The trajectories are classified into Z categories, which represents the set of entry trajectories. The trajectory in the middle is divided into By dividing the flight flow into Z subsets, flight flow identification is achieved.

4. The method for predicting arrival flight plans for terminal area traffic simulation according to claim 1, characterized in that, Step 3) includes: (3.1) Determine the proportion of high-rise buildings used. (3.1.1) Set the height layer set There are a total of max There are several altitude layers, where q is the index of the altitude layer; (3.1.2) For the z-th flight flow Choose any trajectory Calculate the trajectory The altitude h of the first waypoint start With each height layer in height layer set A The distance is used to determine the trajectory, and the height level corresponding to the minimum distance is taken as the trajectory. The altitude layer used when flying into the terminal area; (3.1.3) Traverse the z-th flight flow Get all trajectories in the dataset, obtain the height layer used for all trajectories, and calculate the height layer used. Number of trajectories , This indicates that in flight flow The number of trajectories using the q-th height layer; (3.1.4) Statistical analysis of the flow of the z-th flight The number of trajectories using the q-th altitude layer accounts for a certain percentage of the z-th flight flow. The proportion of all trajectories is used to obtain the flow of the z-th flight. Using the proportional distribution of the qth height layer , ; where N z Indicates the flow of the z-th flight The total number of all trajectories; (3.2) Determine the usage ratio of different machine types (3.2.1) Statistical set of entry trajectories All aircraft types operated by China are listed in the aircraft type collection. ; (3.2.2) Count the number of trajectories using the m-th model. ; (3.2.3) Calculate the usage ratio of the m-th model. , ; Where N represents the set of entry trajectories. The total number of aircraft in the country.

5. The method for predicting arrival flight plans for terminal area traffic simulation according to claim 1, characterized in that, Step 5) includes: (5.1) Let the statistical time period be 1 hour, denoted as S. , At the start time, The termination time; (5.2) Obtain the flow of the z-th flight Trajectory in time series of flight paths , where K is the trajectory Number of track points included; (5.3) Judgment Does it contain at least one track time? If it contains, then the trajectory is considered. Within the time range S; if not included, the trajectory is considered to be... Not within the time range S; (5.4) Traverse the z-th flight flow The total number of trajectories within the time range S will be used as the data for the z-th flight in the S-th hour. Flight traffic ; (5.5) Obtain the flow of the z-th flight respectively Flight traffic in China during 24 time periods.

6. The method for predicting arrival flight plans for terminal area traffic simulation according to claim 1, characterized in that, Step 6) includes: (6.1) Hourly traffic flow forecast for air routes (6.1.1) Let the average annual growth rate of flight volume be... The predicted year is y years in the future, where y is an integer; (6.1.2) Planned flight route R z Flight traffic during the Sth time period of a 24-hour day in the next y years Compared with the current z-th flight flow Flight traffic during the corresponding time period The relationship between them is shown in equation (13): （13）； (6.1.3) Repeat step (6.1.2) to predict the flight flow for the z-th flight in the next y years. Flight traffic at different times of the day 24 hours ; (6.2) Constructing flight plans (6.2.1) Take the flow of the z-th flight Planned flight route R z Flight traffic in the Sth time period ,create There are several simulated flight plans, where the p-th simulated flight plan is denoted as FlightPlan(p). The elements of a simulated flight plan (FlightPlan(p)) include the identification number (FlightPlan(p).Id), the planned flight route (FlightPlan(p).Route), the aircraft type (FlightPlan(p).Type), the altitude level (FlightPlan(p).Attitude), and the time of occurrence (FlightPlan(p).Time). (6.2.2) Let p be the identifier of the simulated flight plan FlightPlan(p), and let the z-th flight be... Planned flight route R z The planned flight path for the simulated flight plan FlightPlan(p); (6.2.3) Using random sampling, generate the altitude layer for each simulated flight plan FlightPlan(p); (6.2.4) Using random sampling, generate the aircraft type for each simulated flight plan FlightPlan(p); (6.2.5) Using a random sampling method, generate the occurrence time for each simulated flight plan FlightPlan(p); (6.2.6) Sort all the flights generated in the Sth time period according to their appearance time and combine them into the arrival flight operation simulation plan for the Sth time period.

7. The method for predicting arrival flight plans for terminal area traffic simulation according to claim 6, characterized in that, Step (6.2.3) includes: (a) Generation A random number between 0 and 1, following a uniform distribution. ; (b) If Then, the flight plan FlightPlan(p) uses the l-th altitude layer, FlightPlan(p).Altitude. Indicates the flow of the z-th flight Use the proportional distribution of the qth height layer; (c) Iterate through all random numbers Generate the corresponding usage altitude layer for each simulated flight plan FlightPlan(p).

8. The method for predicting arrival flight flow plans for terminal area traffic simulation according to claim 6, characterized in that, Step (6.2.4) includes: (a) Generation A random number between 0 and 1, following a uniform distribution. ; (b) If Then, the aircraft type FlightPlan(p).Type of the simulated flight plan FlightPlan(p) is the g-th aircraft type; where, This indicates the proportion of users using the m-th model. (c) Iterate through all random numbers Generate the corresponding aircraft type for each simulated flight plan FlightPlan (p).

9. The method for predicting arrival flight plans for terminal area traffic simulation according to claim 6, characterized in that, Section (6.2.5) includes: (a) Generation A random number between 0 and 1, following a uniform distribution. ; (b) The time interval between two consecutive flights of aircraft follows a negative exponential distribution. Calculate as shown in equation (14): （14）； in, The set time interval represents the minimum safe time interval between two aircraft flying consecutively in a flight flow; (c) Set the occurrence time of the first simulated flight plan FlightPlan(1) to 0 seconds, and calculate the occurrence time of all flights in sequence according to formula (15): FlightPlan(p+1).Time=FlightPlan(p).Time+ (15); The arrival time of an incoming flight indicates the moment it enters the terminal area boundary.

Images