A method, device, and medium for obtaining a digital twin of an airport flight area

Abstract

The present invention provides a method, device and medium for obtaining a digital twin of an airport flight area, relating to the technical field of digital twin. The method includes: obtaining physical facility data, equipment status data and operation data of the target airport flight area, creating a three-dimensional geometric model based on the collected object facility data and equipment status data, and performing dynamic simulation on the operation data in the flight area based on the three-dimensional geometric model to determine the target digital twin model, so as to realize automated and efficient management of the airport flight area.

Description

Technical Field

[0001] The present invention relates to the technical field of digital twins, and particularly to a method, device, and medium for obtaining a digital twin of an airport apron. Background Art

[0002] As a key hub of air transportation, the scale and number of airports are constantly increasing, and their functions are becoming increasingly complex and diverse. Digital twin technology, as a key technology, has been applied in the airport field. Wang Jiamian studied the check-in service system of the airport terminal. Using digital twin technology, by determining the key factors affecting the check-in efficiency, establishing a simulation model, and optimizing it using an improved genetic algorithm, the goal of minimizing the queuing waiting time was achieved. Conde J et al. proposed a reference concept and data model for airport digital twins based on FIWARE general enablers and Next Generation Service Interfaces - Linked Data standards to improve the efficiency of flight turnaround events.

[0003] Although digital twin technology is being increasingly widely applied in the civil aviation field, its application in the airport apron is still in its infancy and faces many challenges, such as great difficulty in technical implementation, low data utilization value, and unsatisfactory simulation software. Particular attention has been paid to the research on conflict detection and avoidance for aircraft in the airport apron.

[0004] As the core area of the airport, the apron undertakes important functions such as aircraft takeoff and landing, taxiing, parking, and ground traffic organization. Its complexity and importance require managers to master detailed and accurate geospatial information and the status of facilities and equipment. Traditional management methods often rely on two-dimensional drawings and manual inspections, which are difficult to meet the efficient and safe management needs of modern airports. Summary of the Invention

[0005] In view of the above technical problems, the technical solution adopted by the present invention is as follows: A method for obtaining a digital twin of an airport apron, the method comprising the following steps:

[0006] S100, obtaining physical facility data, equipment status data, and operation data of the target airport apron;

[0007] S200, creating a three-dimensional geometric model based on the collected physical facility data and equipment status data;

[0008] S300, performing dynamic simulation on the operation data in the apron based on the three-dimensional geometric model to generate a target digital twin model;

[0009] Wherein, the target digital twin model includes a target module, and the target module performs the following steps to implement conflict detection for the target aircraft:

[0010] S410. Obtain the current flight data of the target aircraft flying at the current time node, where the current flight data at least includes: current position coordinates (x, y, z), current speed v, and current movement direction θ; where x is the value on the horizontal axis of the plane rectangular coordinate converted from the longitude and latitude value of the position where the target aircraft is located at the current time node, y is the value on the vertical axis of the plane rectangular coordinate converted from the longitude and latitude value of the position where the target aircraft is located at the current time node, and z is the flight altitude of the target aircraft flying at the current time node.

[0011] S420. Generate an initial protective cover based on the current flight data of the target aircraft. Among them, the initial protective cover used to predict the conflict of the target aircraft is an initial polygon centered on the current position coordinates (x, y, z) and surrounding the target aircraft. The i-th initial vertex A i of the initial polygon satisfies the following requirements: A i = (x + r i cos(θ + α i ), y + r i sin(θ + α i ), where the value range of i is from 1 to m, and m is the number of initial vertices of the initial polygon; r i is the preset radius of the i-th initial vertex, and the i-th preset angle α i is the angle formed by connecting A i to the origin in the specified rectangular coordinate system. The specified rectangular coordinate system takes the center of gravity of the target aircraft as the origin, the current movement direction as the X-axis, and the direction obtained by rotating the current movement direction clockwise by 90 degrees as the Y-axis.

[0012] S430. Input the current flight data and the initial protective cover of the target aircraft into the target machine learning model, obtain the initial conflict result, and obtain the target conflict type based on the initial conflict result. The target conflict type is wake vortex conflict, head-on conflict, or crossing conflict.

[0013] S440. Based on the initial conflict result and the target conflict type, obtain the predicted collision position of the target aircraft and increase the number of vertices of the initial protective cover corresponding to the predicted collision position, so as to obtain an intermediate protective cover; among them, the shapes of the intermediate protective covers corresponding to different predicted collision positions are different.

[0014] S450. Input the current flight data and the intermediate protective cover of the target aircraft into the target machine learning model to obtain the target conflict result.

[0015] According to another aspect of the present invention, there is provided a non-transitory computer-readable storage medium, in which at least one instruction or at least one program segment is stored, and the at least one instruction or the at least one program segment is loaded and executed by a processor to implement the foregoing method.

[0016] According to another aspect of the present invention, an electronic device is provided, including a processor and the aforementioned non-transitory computer-readable storage medium.

[0017] The present invention has at least the following beneficial effects: acquiring physical facility data, equipment status data, and operation data of the flight area of the target airport, creating a three-dimensional geometric model based on the collected object facility data and equipment status data, dynamically simulating the operation data in the flight area based on the three-dimensional geometric model to determine the target digital twin model, acquiring the current flight data of the target aircraft flying at the current time node, generating an initial protective cover based on the current flight data of the target aircraft, inputting the current flight data of the target aircraft and the initial protective cover into the target machine learning model to obtain an initial conflict result and obtaining the target conflict type based on the initial conflict result, obtaining the predicted collision position of the target aircraft based on the initial conflict result and the target conflict type and increasing the number of vertices of the initial protective cover corresponding to the predicted collision position to obtain an intermediate protective cover, and inputting the current flight data of the target aircraft and the intermediate protective cover into the target machine learning model to obtain the target conflict result, thereby improving the accuracy of aircraft conflict recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can also obtain other drawings based on these drawings without creative efforts.

[0019] Figure 1 It is a flowchart of a method for obtaining a digital twin of an airport flight area provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0020] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts fall within the scope of protection of the present invention.

[0021] An embodiment of the present invention provides a method for obtaining a digital twin of an airport flight area, as Figure 1 shown, the method includes the following steps:

[0022] S100, acquiring physical facility data, equipment status data, and operation data of the flight area of the target airport.

[0023] Among them, the physical facility data includes: geographical information data of the runway, taxiway, apron, terminal building, and parking lot in the flight area; the physical facility data also includes attribute data of facilities such as boarding bridges, baggage handling equipment, aircraft, and tractors.

[0024] Among them, the equipment status information includes the status information of pressure sensors, voltage sensors, current sensors, etc.

[0025] Among them, the operation data includes flight information data, traffic flow data, passenger flow data, etc. The flight information data includes flight schedule, flight dynamics (takeoff time, landing time, boarding gate change, etc.) data; the traffic flow data includes traffic flow, speed, and driving path data of aircraft and ground vehicles (tugboats, shuttle buses, baggage carts, etc.).

[0026] S200, create a three-dimensional geometric model based on the collected physical facility data and equipment status data.

[0027] Specifically, those skilled in the art know that any method of creating a three-dimensional geometric model based on data in the prior art belongs to the protection scope of the present invention. For example, using professional three-dimensional modeling software such as CAD software or 3ds Max to create a three-dimensional geometric model.

[0028] S300, based on the three-dimensional geometric model, perform dynamic simulation on the operation data in the flight area to determine the target digital twin model.

[0029] In summary, obtain the physical facility data, equipment status data, and operation data of the flight area of the target airport, create a three-dimensional geometric model based on the collected physical facility data and equipment status data, perform dynamic simulation on the operation data in the flight area based on the three-dimensional geometric model, and determine the target digital twin model, so as to realize the automated and efficient management of the aircraft flight area.

[0030] Among them, the target digital twin model includes a target module, and the target module performs the following steps to achieve conflict detection of the target aircraft:

[0031] S410, obtain the current flight data of the target aircraft flying at the current time node, and the current flight data at least includes: current position coordinates (x, y, z), current speed v, and current movement direction θ; where x is the value on the horizontal axis of the plane rectangular coordinate converted from the longitude and latitude value of the position where the target aircraft is located at the current time node, y is the value on the vertical axis of the plane rectangular coordinate converted from the longitude and latitude value of the position where the target aircraft is located at the current time node, and z is the flight altitude of the target aircraft flying at the current time node.

[0032] Specifically, the current flight data is obtained based on ADS-B broadcasts. Those skilled in the art will understand that any method in the prior art for converting longitude and latitude values into flat-angle rectangular coordinates falls within the protection scope of the present invention and will not be elaborated herein.

[0033] S420, generate an initial protective cover based on the current flight data of the target aircraft. Among them, the initial protective cover for predicting conflicts of the target aircraft is an initial polygon centered on the current position coordinates (x, y, z) and surrounding the target aircraft. The i-th initial vertex A i of the initial polygon satisfies the following requirements: A i = (x + r i cos(θ + α i ), y + r i sin(θ + α i ), z), where the value range of i is from 1 to m, and m is the number of initial vertices of the initial polygon; r i is the preset radius of the i-th initial vertex, and the i-th preset angle α i is the angle formed by connecting A i to the origin in the specified rectangular coordinate system. The specified rectangular coordinate system has the center of gravity of the target aircraft as the origin, the current movement direction as the X-axis, and the direction obtained by rotating the current movement direction clockwise by 90 degrees as the Y-axis.

[0034] Specifically, obtain the aircraft type and the protective cover parameter data corresponding to the aircraft type. The protective cover parameter data includes: the number of vertices, the preset radius corresponding to each vertex, and the preset angle corresponding to each vertex. By obtaining the aircraft type of the target aircraft, obtain the protective cover parameter data m, α i and r i corresponding to the target aircraft.

[0035] Specifically, the initial protective cover is a geometric area delimited around the aircraft, used to represent the safety range of the aircraft. When other aircraft or objects enter the initial protective cover, it is considered that there is a risk of conflict.

[0036] In an embodiment of the present invention, m = 6. A hexagon can better fit the outer contour of the aircraft, especially when considering the extension range of the leading edge and trailing edge of the aircraft wings. For example, the front and rear vertices of the hexagon can be determined according to the fuselage length of the aircraft and the safety buffer distance, while the left and right vertices can be set according to the wingspan and the safety buffer distance, and the oblique sides can consider the potential influence range of the leading edge and trailing edge of the aircraft wings.

[0037] S430. Input the current flight data of the target aircraft and the initial protective cover into the target machine learning model to obtain the initial conflict result and, based on the initial conflict result, obtain the target conflict type, where the target conflict type is wake vortex conflict, head-on conflict, or crossing conflict. Specifically, the target machine learning model is a target support vector machine model.

[0038] Specifically, when the initial conflict result is in a conflict operation state, obtain the target conflict type based on the conflict result. Further, if the moving directions of the two aircraft in conflict are the same and the following aircraft enters the wake vortex area of the preceding aircraft, determine that the target conflict type is wake vortex conflict; if the moving directions of the two aircraft in conflict are opposite and they approach each other, determine that the target conflict type is head-on conflict; if the flight trajectories of the two aircraft in conflict intersect at a certain point, determine that the target conflict type is crossing conflict.

[0039] Specifically, S430 further includes: when the initial conflict result is in a normal operation state, wait for the next time node to execute S410.

[0040] S440. Based on the initial conflict result and the target conflict type, obtain the predicted collision position of the target aircraft and increase the number of vertices of the initial protective cover corresponding to the predicted collision position, thereby obtaining an intermediate protective cover; where the shapes of the intermediate protective covers corresponding to different predicted collision positions are different. Specifically, after determining the predicted collision position and the increased number of vertices, in an embodiment of the present invention, the increased vertices are added to the predicted collision position according to a preset addition rule, thereby obtaining the intermediate protective cover. The preset addition rule can be determined according to the actual situation. For example, the increased vertices are evenly distributed based on (x, y).

[0041] S450. Input the current flight data of the target aircraft and the intermediate protective cover into the target machine learning model to obtain the target conflict result.

[0042] In summary, obtain the current flight data of the target aircraft flying at the current time node. Based on the current flight data of the target aircraft, generate an initial protective cover. Input the current flight data of the target aircraft and the initial protective cover into the target machine learning model to obtain the initial conflict result and, based on the initial conflict result, obtain the target conflict type. Based on the initial conflict result and the target conflict type, obtain the predicted collision position of the target aircraft and increase the number of vertices of the initial protective cover corresponding to the predicted collision position, thereby obtaining an intermediate protective cover. Input the current flight data of the target aircraft and the intermediate protective cover into the target machine learning model to obtain the target conflict result. By increasing the number of vertices of the initial protective cover corresponding to the predicted collision position, the present invention further determines the target conflict result and improves the accuracy of aircraft conflict recognition without increasing too much data volume.

[0043] Further, S440 also includes:

[0044] S441. When the target conflict type is wake vortex conflict and the predicted collision position of the target aircraft is at the tail of the aircraft, the shape of the middle protective cover at the tail of the target aircraft is the first preset shape and the number of vertices n of the middle protective cover at the tail of the aircraft satisfies the following requirements: n = n0 + ceil(k × v / av), where the first preset shape is determined based on the shape of the aircraft tail, n0 is the initial number of vertices of the initial protective cover at the tail of the aircraft, k is the speed influence parameter, av is the preset average speed, and ceil() is the ceiling function.

[0045] S442. When the target conflict type is head-on conflict, the shape of the middle protective cover at the head of the target aircraft is the second preset shape and the number of vertices q of the middle protective cover at the head of the aircraft satisfies the following requirements: q = q0 + ceil(k × v / av), where the second preset shape is determined based on the shape of the aircraft head, and q0 is the initial number of vertices of the initial protective cover at the head of the aircraft.

[0046] S443. When the target conflict type is crossing conflict, obtain the crossing angle φ of the target aircraft based on the current flight data, and determine the predicted collision position of the target aircraft based on the crossing angle φ. The shape of the middle protective cover at the predicted collision position is the third preset shape and the number of vertices of the middle protective cover at the predicted collision position is greater than the corresponding initial number of vertices of the initial protective cover at the predicted collision position, where the third preset shape is determined based on the predicted collision position, and the crossing angle φ is the included angle formed by the current flight direction of the target aircraft and the current flight direction of the conflicting aircraft, and the conflicting aircraft is the aircraft that conflicts with the target aircraft in the initial conflict result.

[0047] Furthermore, S441 also includes: the number of vertices p of the middle protective cover at the head of the aircraft satisfies the following requirements: p = q0 - ceil(k × v / av).

[0048] In summary, when the target conflict type is wake vortex conflict and the predicted collision position of the target aircraft is at the tail of the aircraft, increase the number of vertices of the protective cover corresponding to the tail of the aircraft and decrease the number of vertices of the protective cover corresponding to the tail of the aircraft. When the target conflict type is head-on conflict, increase the number of vertices of the protective cover corresponding to the head of the aircraft. When the target conflict type is crossing conflict, obtain the crossing angle φ of the target aircraft based on the current flight data, and determine the predicted collision position of the target aircraft based on the crossing angle φ, and increase the number of vertices n of the protective cover corresponding to the predicted collision position.

[0049] Further, the time interval t between the next time node and the current time node satisfies the following condition: t = α / (β + γ × ED), where α is a preset basic time interval, β is a preset adjustment coefficient, γ is a preset event density influence coefficient, ED is the number of conflicts occurring to the target aircraft per unit time, and β < 1, γ < 1.

[0050] Optionally, α = 0.1, β = 0.05, γ = 0.01.

[0051] Specifically, it further includes obtaining a target machine learning model through the following steps:

[0052] S001. Obtain the sample flight data of a number of sample aircraft and the true conflict results corresponding to the sample flight data. The sample flight data at least includes: sample position coordinates (x0, y0, z0), sample speed v0, and sample movement direction θ0. Among them, x0 is the value on the horizontal axis of the plane rectangular coordinate converted from the longitude and latitude value of the position where the sample aircraft is located at the sample time node, y0 is the value on the vertical axis of the plane rectangular coordinate converted from the longitude and latitude value of the position where the sample aircraft is located at the sample time node, and z is the flight altitude of the sample aircraft at the sample time node.

[0053] S002. Generate a sample protective cover based on the sample flight data of the sample aircraft. The sample protective cover is a sample polygon surrounding the target aircraft based on the sample position coordinates (x0, y0, z0). The j-th sample vertex B j of the sample polygon satisfies the following requirements: B j = (x0 + r 0j cos(θ0 + α 0j ), y0 + r 0j sin(θ0 + α 0j ), z0); r 0j is the preset radius of the i-th sample vertex, and the i-th preset angle α 0j is the angle formed by the connection between B j and the origin in the plane rectangular coordinate. The value range of j is from 1 to n, and n is the number of sample vertices.

[0054] Specifically, based on the aircraft type and the protective cover parameter data corresponding to the aircraft type, obtain the protective cover parameter data n, α 0j and r 0j corresponding to the sample aircraft.

[0055] S003. Construct a machine learning model, and use the sample flight data, sample protective cover, and true conflict results to train the constructed machine learning model to obtain sample training results.

[0056] S004, if the sample training result meets the preset training requirements, use the trained machine learning model as the target machine learning model.

[0057] Specifically, the preset training requirements can be set according to actual needs.

[0058] In summary, obtain the sample flight data of several sample aircraft and the corresponding real conflict results of the sample flight data. Based on the sample flight data of the sample aircraft, generate sample protective covers, construct a machine learning model, and use the sample flight data, sample protective covers, and real conflict results to train the constructed machine learning model to obtain sample training results. If the sample training results meet the preset training requirements, use the trained machine learning model as the target machine learning model.

[0059] In a specific embodiment of the present invention, the specific implementation strategy after a conflict includes:

[0060] In a wake vortex conflict, when one of the two aircraft enters this section, that aircraft serves as the master aircraft. When the second aircraft enters this section, the distance between the second aircraft and the first aircraft is detected in real time. When the distance is less than the safe distance, the second aircraft decelerates and stops until the distance is greater than the safe distance, and then the second aircraft starts to move forward until one of the two aircraft leaves the wake vortex monitoring area, and the monitoring ends. When multiple aircraft are following in the wake vortex at the same time, the subsequent aircraft are controlled by the nearest master aircraft; when a wake vortex conflict is detected, the following aircraft needs to accurately calculate the safe deceleration amplitude based on factors such as the distance from the preceding aircraft, the type of the preceding aircraft, and the current meteorological conditions. For example, if the preceding aircraft is a large airliner, its generated wake vortex is stronger, and the following aircraft may need a greater deceleration amplitude. Generally, the deceleration amplitude of the following aircraft can be between 5 - 15 knots, and the specific value is obtained through a wake vortex intensity calculation model, which comprehensively considers the influence of parameters such as the wingspan, engine thrust, and flight altitude of the preceding aircraft on the wake vortex intensity. During the deceleration process, the following aircraft should maintain a stable deceleration rate to avoid other risks caused by sudden braking or excessive deceleration. At the same time, continuously monitor the distance from the preceding aircraft. Once the distance returns to the safe range, the following aircraft can gradually resume the normal taxiing speed, and the rate of resuming the speed also needs to be controlled according to airport regulations and actual situations, generally not exceeding 5 knots per second.

[0061] In a head-on conflict, when one of the two aircraft enters this section, that aircraft serves as the master aircraft and monitors in real time whether the second aircraft is too close to this monitoring section. When the distance is less than a certain threshold, the second aircraft decelerates and stops until the master aircraft leaves the monitoring area, and then the monitoring ends and the second aircraft is released. When multiple aircraft have a head-on conflict at the same time, the master aircraft controls all subsequent aircraft and queues them to pass through this section.

[0062] In cross conflicts, when one of the two aircraft enters the section, that aircraft serves as the master aircraft, which monitors in real time whether the second aircraft is too close to the monitored section. When the distance is less than a certain threshold, the second aircraft decelerates and stops until the master aircraft leaves the monitored area, at which point the monitoring ends and the second aircraft is released. When multiple aircraft simultaneously encounter cross conflicts, the master aircraft controls all subsequent aircraft, and they queue up to pass through the section. Specifically, due to the short length of some taxiways, there may be a situation where intersections merge, in which case the two intersections are treated as one intersection, giving it exclusivity. In head-on conflicts or cross conflicts, the relevant aircraft or vehicles must immediately stop or decelerate to the minimum safe speed. For example, when an aircraft conflicts near a runway-taxiway intersection, it should reduce its speed to 0 knots or maintain an extremely low idle speed (such as below 5 knots) within the shortest time to ensure no collision occurs within the conflict area. The decision to stop or decelerate is based on the relative positions, velocity vectors of the conflicting parties, and the safety thresholds preset by the airport. Through an accurate calculation model, the conflict risk is evaluated in real time, and once the risk exceeds the safety threshold, a stop or decelerate command is immediately triggered.

[0063] Specifically, when the conflict cannot be resolved by speed adjustment, the system quickly initiates path replanning. The replanned path is selected based on a comprehensive analysis of the real-time traffic conditions in the entire airport flight area, including the positions, movement directions, speeds of other aircraft and vehicles, as well as the occupancy of runways and taxiways. Advanced path planning algorithms such as the Dijkstra algorithm or A algorithm are used to search for alternative paths in the airport's topological map. These algorithms consider factors such as the length of the path, congestion level, and intersections with other traffic flows to find the optimal conflict resolution path. For example, during peak hours, paths with less traffic and higher taxiway grades are preferentially selected to improve taxiing efficiency. Path replanning must strictly follow the airport's facility layout and operating rules. For example, it is necessary to avoid guiding aircraft to taxiways that are under maintenance or closed, and ensure that the navigation facilities and signs on the path are intact and available. For special areas such as runway end safety areas and clearance protection areas, path planning should ensure that aircraft and vehicles stay away from these areas to prevent threats to the safety of airport operations.

[0064] Furthermore, emergency mission flights such as medical rescue flights and fire fighting flights are always given the highest priority. These flights have unconditional right of way during conflict resolution, and other aircraft and vehicles must immediately give way to them to ensure the rapid execution of emergency missions. The system quickly identifies emergency flights through real-time information interaction with the airport emergency command center and air traffic control department, and sets up special identification and priority handling mechanisms for them.

[0065] Due to the large number of passengers on large airliners and their relatively large occupancy of airport resources, they are usually given a higher priority in conflict resolution. Departing flights, which are related to flight punctuality and subsequent air traffic flow management, are also given relatively high priority. For large airliners, according to their aircraft types and the distribution of boarding bridge resources at the airport, they are preferentially arranged to use positions close to the terminal building with complete boarding bridge facilities to reduce the boarding time of passengers and the use of airport shuttle buses. Departing flights are then reasonably arranged for taxiing routes and conflict resolution sequences according to their estimated departure times and the sequencing requirements of air traffic control.

[0066] Furthermore, for aircraft or vehicles waiting for a long time on the taxiway or apron, their priorities are appropriately increased during conflict resolution. Through real-time monitoring and statistics of the waiting time, when the waiting time exceeds a certain threshold (such as 15 minutes), the system automatically adjusts their priorities to reduce their ground residence time and improve the overall utilization rate of airport resources. This priority adjustment helps to balance the service fairness of different objects within the airport and avoid a series of subsequent problems caused by some flights or vehicles waiting for a long time, such as passenger dissatisfaction and crew fatigue.

[0067] In the case of resource shortages, such as limited resources like boarding bridges and refueling equipment, aircraft or vehicles that can quickly release resources are preferentially arranged for conflict resolution. For example, flights that are about to complete the boarding process are preferentially arranged to taxi to the runway to vacate the boarding bridge resources for subsequent flights; vehicles that have completed refueling are preferentially guided to leave the refueling area to increase the turnover rate of refueling facilities.

[0068] In another specific embodiment of the present invention, the target digital twin model further includes: a speed rule curve model of the aircraft. The speed rule curve model is mainly divided into curve fitting in the takeoff stage, curve fitting in the landing stage, and linear fitting in the taxiing out and into the port stages. The curve model of the aircraft's takeoff stage can be fitted based on the position, altitude, and speed information of ADS -B through algorithms such as polynomial regression and spline interpolation. The curve fitting in the landing stage fits the braking deceleration stage through an exponential decay function or uses a piecewise function to fit the entire landing process. The speed curve of the aircraft's taxiing movement is fitted linearly.

[0069] Specifically, polynomial regression is a commonly used curve fitting method. For the takeoff acceleration stage, the change in the aircraft's speed over time can be approximately represented by a polynomial function. For example, by methods such as the least squares method, the coefficients are solved based on the speed and time information in the collected ADS -B data, thereby obtaining the fitting curve. The polynomial regression method is simple and intuitive and can provide a good fitting effect for relatively smooth acceleration processes. It can select an appropriate polynomial degree according to the characteristics of the data. For example, if the acceleration process is more complex, a higher-degree polynomial may be needed to better capture the speed change.

[0070] Spline interpolation is a method for constructing a smooth curve between data points. It fits the curve by using low-degree polynomials (usually cubic splines) between adjacent data points and ensures that the curve has continuous first and second derivatives at the data points, making the curve smoother. Spline interpolation can better adapt to situations with more complex speed changes. Especially during takeoff, an aircraft may have different acceleration phases, such as initial acceleration and engine thrust adjustment phases. It can more accurately describe the changing trend of speed and avoid overfitting or underfitting problems that may occur in polynomial regression.

[0071] For the curve fitting of the landing phase, after the aircraft touches down and enters the braking deceleration phase, the speed usually shows an exponential decay trend. An exponential decay function can be used for fitting. By fitting the speed and time information during the braking deceleration phase in the ADS-B data, the values of and are determined. The exponential decay function can well describe the process of rapid speed decline, which conforms to the physical laws of braking deceleration. It can accurately reflect the performance of the braking system and the deceleration effect based on actual data. At the same time, by comparing the fitting parameters under different aircraft or different landing conditions, the working conditions of the braking system and the impact of runway conditions on deceleration can be evaluated.

[0072] The landing process includes different phases such as approach, touchdown, and braking deceleration, and the speed change law is different in each phase. A piecewise function can be used to fit the speed curve of the entire landing process. For example, a constant function can be used to represent a relatively stable speed during the approach phase, a linear function can be used to represent the transition of speed from the approach speed to the touchdown speed during the touchdown phase, and an exponential decay function is used during the braking deceleration phase. The piecewise function can more accurately describe the complex speed changes during the landing process, taking into account the physical characteristics and actual situations of different phases.

[0073] For the method of fitting the speed curve of taxiway movement, since the speed of the aircraft on the taxiway is relatively low and stable, a linear function is usually used to fit the change of speed with time. Here, is the initial speed, is the slope, representing the rate of change of speed. In most cases, the value of is small because the taxiing speed changes little. The linear fitting method is simple and has a small amount of calculation, and can provide a sufficiently accurate fitting result for the relatively stable speed situation on the taxiway. It can quickly obtain the changing trend of speed with time, and some basic characteristics during the taxiing process, such as average speed and speed fluctuation, can be analyzed based on the fitting result. By analyzing the slope and intercept of the fitting curve, the speed change situation of the aircraft during taxiing out of the port can also be understood, as well as whether it meets the taxiing speed requirements stipulated by the airport.

[0074] For the fitted motion curve, verification and evaluation are required. Metrics such as root mean square error (RMSE) and mean absolute error (MAE) can be used to measure the closeness of the fitted curve to the actual data, ensuring that the fitted curve can accurately reflect the taxiing motion of the aircraft. Integrate the fitted motion curve into the simulation model. This requires converting the mathematical expression of the motion curve into a form that the simulation model can recognize and process. According to the motion curve, set the motion parameters of the aircraft at different stages, such as speed, acceleration, steering angle, etc. For example, during the takeoff stage, set the acceleration process of the aircraft according to the motion curve; during the taxiing stage, set the appropriate taxiing speed and steering logic.

[0075] In another specific embodiment of the present invention, the target digital twin model further includes: a simulation time management algorithm and a simulation optimization algorithm.

[0076] The simulation time management algorithm advances the simulation time triggered by various events in the airport flight area, ensures accuracy by dynamically adjusting the time step, uses distributed clock synchronization technology to ensure time synchronization among modules, and enables precise synchronization between real-time data and simulation time. The time advancement algorithm takes various events occurring in the airport flight area (such as aircraft takeoff and landing, taxiing, vehicle driving, etc.) as the trigger points for time advancement. When an event occurs, the system updates the system state according to the event type and relevant rules, and dynamically adjusts the time step based on the busyness of the airport operation and the density of events. A smaller time step (such as 0.1 second) is adopted during peak hours with high traffic flow and frequent events to accurately capture details, and the time step is appropriately increased (such as 1 second) during periods with low traffic flow and few events to improve simulation efficiency while ensuring no important information is lost. The system monitors the running state of the simulation environment in real time and automatically adjusts the time step. For example, when airplanes are queuing up for takeoff on the runway, it switches to a small time step, thus accurately simulating the airport operation scenario and making the simulation process consistent with the actual operation in terms of time logic. The time synchronization algorithm ensures time synchronization among different modules (such as aircraft motion models, vehicle motion models, conflict detection modules, etc.) in the simulation system. Distributed clock synchronization technology (such as Network Time Protocol NTP or Precision Time Protocol PTP) is used to synchronize the clocks of each module, and timestamp information is transmitted among modules to ensure the processing order and time consistency of events in different modules. For example, after the aircraft motion model updates the aircraft position, it transmits the timestamped position information to the conflict detection module to ensure that the detection result accurately corresponds to the actual position of the aircraft at the same time point. When processing real-time data updates, an accurate timestamp (consistent with the time accuracy of the simulation system, such as microsecond level) is added to each piece of real-time data collected. Through timestamp matching and calibration, the real-time data is accurately integrated into the simulation system. The real-time data timestamp and the simulation time are regularly compared and calibrated, and a synchronization lock mechanism is adopted to avoid data conflicts, ensuring the accuracy and reliability of the collaborative work of each module in the system. The time scheduling algorithm is responsible for reasonably scheduling various events in the simulation process. It determines the execution order and timing of events based on factors such as the priority of events, time sequence, and the availability of system resources. For example, events related to emergency mission flights (such as medical rescue flights, fire-fighting flights, etc.) are given the highest priority to ensure their prior execution; for events of other regular flights, they are scheduled according to predefined rules (such as first come, first served, according to the importance of the flight, etc.). At the same time, the event scheduling algorithm also needs to consider the allocation of system resources to avoid event execution being blocked or incorrect due to resource conflicts. Through reasonable event scheduling, it is ensured that the simulation process can accurately reflect various situations in the actual operation of the airport, improving the authenticity and effectiveness of the simulation.

[0077] The simulation optimization algorithms cover intelligent optimization algorithms, deep learning model optimization, and multi-agent reinforcement learning motion model optimization. Through the collaborative action of various technical means, they improve the efficiency of airport resource allocation, enhance the accuracy of traffic flow prediction, and optimize traffic management strategies. Intelligent optimization algorithms include genetic algorithms, particle swarm optimization, ant colony algorithms, simulated annealing, etc., and are commonly used to solve complex optimization problems. In airport digital twins, particle swarm optimization is suitable for handling real-time scheduling in dynamic environments, such as flight scheduling and resource allocation, and can converge quickly; genetic algorithms perform well in dealing with complex multi-objective optimization problems (such as passenger flow management and facility layout optimization) and can explore a wider solution space. Taking gate assignment as an example, the genetic algorithm encodes gates and flights as gene elements to construct chromosomes, and selects excellent individuals through a fitness function (considering factors such as passenger walking distance, flight delay cost, and gate facility utilization rate), and generates new solutions through crossover and mutation operations. After multiple iterations, the gate assignment scheme is optimized. For airport ground vehicle scheduling, the particle swarm optimization algorithm regards vehicles as particles, and according to their positions, speeds, and fitness functions (such as minimizing the total travel time, reducing energy consumption, and improving the task completion rate), finds the optimal driving path and task execution order through the cooperation and competition among particles, improves the airport operation efficiency, and can also be widely applied to problems such as ground traffic flow management and resource allocation optimization, supporting airport emergency management and promoting the intelligent development of airports. The deep learning optimization algorithm optimizes tasks such as airport traffic flow prediction by constructing a deep learning model that combines the advantages of convolutional neural networks (CNNs) and recurrent neural networks (RNNs). The CNN extracts the spatial layout features of the airport (such as runway, taxiway layout, and facility distribution), and the RNN processes time series data (such as flight takeoffs and landings and passenger flow changes), and adjusts the network layer parameters (convolution kernel size, number, RNN hidden unit number, etc.) to improve the model fitting ability. When training the model, an adaptive learning rate strategy is adopted. A larger learning rate is used in the initial stage to accelerate parameter updates, and then reduced in the later stage to avoid oscillations, ensuring convergence to the global optimum or a near-optimum solution. At the same time, regularization techniques (such as L2 regularization) are used to add penalty terms to the loss function to constrain the parameter size, prevent overfitting, and improve the generalization ability of the model in complex operating environments, accurately predicting airport traffic flow in different situations and providing strong support for operation management. The reinforcement learning motion model optimization algorithm uses the deep Q-network (DQN) and its improved strategies to optimize the motion models of multi-agents (aircraft and ground vehicles). The DQN uses a deep neural network to process the complex state space (combinations of aircraft and vehicle positions, speeds, headings, etc.) and action space (various operation actions) of airport surface traffic. By introducing an experience replay mechanism, the experience data of the interaction between the agent and the environment is stored in the buffer, and samples are randomly selected for learning, breaking the sample correlation to improve the learning efficiency; the target network update strategy is used to stabilize the Q-value estimation result and accelerate model convergence.The agent selects actions based on a carefully designed reward function (which balances individual goals with the overall operating requirements of the airport, such as reducing taxiing time for aircraft, avoiding collisions, and maximizing runway utilization, and ensuring seamless traffic connection between vehicles and aircraft), and selects actions according to the real-time environmental state of the airport. It explores the optimal motion strategy in continuous interaction with the environment, adapts to complex situations such as flight dynamic changes and emergencies, and ensures the orderly and efficient operation of airport traffic. For example, it can still make reasonable decisions in bad weather or peak hours, and can also adapt to the operation characteristics of new equipment or new types of aircraft to improve surface traffic management.

[0078] Embodiments of the present invention also provide a non-transitory computer-readable storage medium, which can be set in an electronic device to store at least one instruction or at least one program related to a method in the method embodiments. The at least one instruction or the at least one program is loaded and executed by the processor to implement the method provided in the above embodiments.

[0079] Embodiments of the present invention also provide an electronic device, including a processor and the aforementioned non-transitory computer-readable storage medium.

[0080] Although some specific embodiments of the present invention have been described in detail by way of examples, those skilled in the art should understand that the above examples are only for illustration purposes and not for limiting the scope of the present invention. Those skilled in the art should also understand that various modifications can be made to the embodiments without departing from the scope and spirit of the present invention.

Claims

1. A method for obtaining a digital twin of an airport flight area, characterized in that: The method comprises the following steps: S100, obtaining physical facility data, equipment status data and operation data of the flight area of ​​the target airport; S200, creating a three-dimensional geometric model based on the collected physical facility data and equipment status data; S300, based on the 3D geometric model, dynamically simulates the operation data in the flight zone and generates a target digital twin model; Among them, the target digital twin model includes a target module, which performs the following steps to realize conflict detection of the target aircraft: S410, obtaining current flight data of the target aircraft flying at the current time node, wherein the current flight data at least includes: current position coordinates (x, y, z), current speed v, and current motion direction θ; wherein x is the value of the longitude and latitude values ​​of the position of the target aircraft at the current time node converted into the horizontal axis of the plane rectangular coordinate, y is the value of the longitude and latitude values ​​of the position of the target aircraft at the current time node converted into the vertical axis of the plane rectangular coordinate, and z is the flight altitude of the target aircraft flying at the current time node; S420, generating an initial protective cover based on the current flight data of the target aircraft, wherein the initial protective cover for predicting the target aircraft conflict is an initial polygon centered at the current position coordinates (x, y, z) and surrounding the target aircraft, wherein the i-th initial vertex A of the initial polygon is i Meet the following requirements: A i =(x+r i cos(θ+α i ), y+r i sin(θ+α i ), z), i ranges from 1 to m, where m is the number of initial vertices of the initial polygon; r i is the preset radius of the i-th initial vertex, and the i-th preset angle α i Yes A i The angle formed by connecting the origin in the specified rectangular coordinate system, wherein the specified rectangular coordinate system has the center of gravity of the target aircraft as the origin, the current direction of movement as the X-axis, and the current direction of movement rotated 90 degrees clockwise as the Y-axis; S430, inputting the current flight data and the initial protection cover of the target aircraft into the target machine learning model, obtaining an initial conflict result and obtaining a target conflict type based on the initial conflict result, wherein the target conflict type is a wake conflict, a head-on conflict or a cross conflict; S440, based on the initial collision result and the target collision type, obtaining the predicted collision position of the target aircraft and increasing the number of vertices of the initial protective cover corresponding to the predicted collision position, thereby obtaining an intermediate protective cover; wherein the shapes of the intermediate protective covers corresponding to different predicted collision positions are different; S450, input the current flight data and intermediate protection cover of the target aircraft into the target machine learning model to obtain the target conflict result.

2. The method for obtaining the digital twin of an airport flight area according to claim 1, characterized in that: The physical facility data includes: geographic information data of the airfield runway, taxiway, apron, aviation building, and parking lot.

3. The method for obtaining the digital twin of an airport flight area according to claim 1, characterized in that: S430 also includes: when the initial conflict result is in a normal operating state, waiting for the next time node to execute S410.

4. The method for obtaining the digital twin of an airport flight area according to claim 1, characterized in that: The S440 also includes: S441, when the target conflict type is a wake conflict, and the predicted collision position of the target aircraft is the tail of the aircraft, the shape of the intermediate protective cover at the tail of the target aircraft is a first preset shape and the number of vertices n of the intermediate protective cover at the tail of the aircraft meets the following requirements: n=n0+ceil(k×v / av), wherein the first preset shape is determined based on the shape of the tail of the aircraft, n0 is the initial number of vertices of the initial protective cover at the tail of the aircraft, k is a speed influence parameter, av is a preset average speed, and ceil() is a rounding up function; S442, when the target conflict type is a head-on conflict, the shape of the intermediate protective cover at the head of the target aircraft is a second preset shape and the number of vertices q of the intermediate protective cover at the head of the aircraft meets the following requirement: q=q0+ceil(k×v / av), wherein the second preset shape is determined based on the shape of the head of the aircraft, and q0 is the initial number of vertices of the initial protective cover at the head of the aircraft; S443, when the target conflict type is an intersection conflict, the intersection angle φ of the target aircraft is obtained based on the current flight data, and the predicted collision position of the target aircraft is determined based on the intersection angle φ, the shape of the intermediate protective cover at the predicted collision position is a third preset shape and the number of vertices of the intermediate protective cover at the predicted collision position is greater than the initial number of vertices corresponding to the initial protective cover at the predicted collision position, wherein the third preset shape is determined based on the predicted collision position, wherein the intersection angle φ is the angle formed by the current flight direction of the target aircraft and the current flight direction of the conflicting aircraft, and the conflicting aircraft is the aircraft that conflicts with the target aircraft in the initial conflict result.

5. The method for obtaining the digital twin of an airport flight area according to claim 4, characterized in that: S441 also includes: the number p of vertices of the intermediate protective cover at the aircraft head meets the following requirement: p=q0-ceil(k×v / av).

6. The method for obtaining the digital twin of an airport flight area according to claim 3, characterized in that: The time interval t between the next time node and the current time node satisfies the following condition: t=α / (β+γ×ED), where α is the preset basic time interval, β is the preset adjustment coefficient, γ is the preset event density influence coefficient, ED is the number of conflicts occurring to the target aircraft per unit time, and β<1, γ<1.

7. The method for obtaining the digital twin of an airport flight area according to claim 6, characterized in that: α=0.1, β=0.05, γ=0.

01.

8. The method for acquiring the digital twin of an airport flight area according to claim 1, characterized in that: The target machine learning model is a target support vector machine model.

9. A non-transitory computer-readable storage medium, wherein at least one instruction or at least one program is stored in the storage medium, characterized in that: The at least one instruction or the at least one program is loaded and executed by the processor to implement the method for acquiring the digital twin of the airport flight area as described in any one of claims 1-8.

10. An electronic device, characterized in that: The invention comprises a processor and the non-transitory computer-readable storage medium as claimed in claim 9.

Images