Method and system for dynamic updating of airport obstacle data

Abstract

This invention relates to the field of airspace management technology, specifically disclosing a method and system for dynamically updating airport airspace obstacle data. The method includes: acquiring regional benchmark data based on a first acquisition precision under standard operating conditions to construct a benchmark regional model; acquiring real-time regional data periodically based on a second acquisition precision under real-time operating conditions to construct a real-time regional model with time labels; identifying the real-time regional model based on an adaptive detection parameter set, locating abnormal candidate areas, and simultaneously locating abnormal obstacles; marking the location results, and visually displaying the real-time regional model. This invention acquires relevant airport information through an acquisition device with data transmission capabilities, performs 3D modeling of the airport, and, combined with a pre-acquired standard model, quickly determines abnormal areas and objects at different times, thus providing intuitive feedback to management personnel. The method boasts extremely high identification efficiency and readability.

Description

Technical Field

[0001] This invention relates to the field of airspace management technology, specifically a method and system for dynamically updating airport airspace obstacle data. Background Technology

[0002] Airport airspace, as the core area ensuring the safe takeoff, landing, and flight of aircraft, has key parameters such as the height and location of obstacles within it that directly affect flight procedure formulation and operational safety. Currently, most airports in China still use traditional management models, mainly relying on periodic manual ground inspections. In terms of technology application, existing systems are mostly limited to single data sources, such as periodic aerial imagery, and fail to effectively integrate multi-source information such as satellite remote sensing, UAV monitoring, ground sensor networks, and urban planning approval data. Furthermore, the data analysis process mainly relies on manual visual interpretation and simple geometric comparison, lacking AI-based automated change detection capabilities, precise 3D spatial modeling methods, and intelligent spatial conflict analysis with complex airspace restriction surfaces. This results in low identification efficiency and a high risk of misjudgment. Therefore, there is an urgent need for an intelligent identification architecture that can integrate multi-source dynamic data to improve identification efficiency and accuracy. Summary of the Invention

[0003] The purpose of this invention is to provide a method and system for dynamically updating airport airspace obstacle data, so as to solve the problems mentioned in the background art.

[0004] To achieve the above objectives, the present invention provides the following technical solution:

[0005] A method for dynamically updating airport airspace obstacle data, the method comprising:

[0006] The clearance area is determined based on the initial range input by the manager, the first acquisition accuracy is determined based on the clearance area, and the regional benchmark data is obtained based on the first acquisition accuracy under standard working conditions to construct a benchmark area model.

[0007] In real-time operation, real-time regional data is acquired periodically based on the second acquisition precision to construct a real-time regional model with time labels; the first acquisition precision uses peak parameters, and the second acquisition precision uses dynamic parameters.

[0008] Access the airport's item management database and determine an adaptive detection parameter set based on each item in the database; the adaptive detection parameter set includes the information comparison accuracy of the real-time area model and the anomaly judgment threshold of the real-time area model;

[0009] Based on the adaptive detection parameter set, the real-time region model is identified, abnormal candidate areas are located, abnormal obstacles are located simultaneously, and the second acquisition accuracy is recursively adjusted according to the location results.

[0010] The location results are marked and visualized in the real-time region model.

[0011] As a further aspect of the present invention: the steps of determining the clearance area based on the initial range input by the manager, determining the first acquisition accuracy based on the clearance area, acquiring regional benchmark data based on the first acquisition accuracy under standard working conditions, and constructing a benchmark area model include:

[0012] Receive the initial range input by the administrator; the initial range uses the outline information in a preset map;

[0013] Determine the initial region based on the initial range;

[0014] Establish a connection channel with the block planning information database to obtain block information, and divide the initial area into sub-regions based on the block information;

[0015] The adjacent regions of each edge sub-region are located based on the block information, and the merging status value is determined based on the block type of the adjacent regions; the merging status value includes zero and one, which are used to represent non-merging and merging, respectively;

[0016] Based on the merged state value, adjacent regions are merged into the initial range to obtain the clear area;

[0017] The first acquisition accuracy is determined based on the clearance area. Under standard working conditions, the regional benchmark data is obtained based on the first acquisition accuracy, and a benchmark regional model is constructed.

[0018] As a further aspect of the present invention: the step of determining the first acquisition accuracy based on the clearance area, acquiring regional reference data based on the first acquisition accuracy under standard working conditions, and constructing a reference area model includes:

[0019] For any sub-area within the airspace clearance area, query its corresponding monitoring equipment; the corresponding monitoring equipment indicates the monitoring equipment whose monitoring range includes the sub-area, and the types include fixed monitoring equipment, drones, and remote sensing platforms.

[0020] Query the maximum accuracy of different types of monitoring devices, select the maximum value from the maximum accuracy, query the monitoring device corresponding to the maximum value, use the maximum value as its acquisition accuracy;

[0021] Under standard operating conditions, it generates image acquisition commands for various types of monitoring devices to acquire image data covering the airspace area.

[0022] A baseline region model is constructed based on the image data.

[0023] As a further aspect of the present invention: the step of acquiring real-time regional data at regular intervals based on a second acquisition precision and constructing a real-time regional model containing time labels in real-time working state includes:

[0024] Generate data acquisition commands periodically while operating in real-time.

[0025] For any sub-region, acquire a region image based on the second acquisition precision, and construct a real-time region model based on the region image;

[0026] The real-time region model is registered based on the aforementioned reference region model;

[0027] Record the time of instruction generation, insert the registered real-time region model, and obtain a real-time region model with time labels.

[0028] As a further aspect of the present invention: the step of determining an adaptive detection parameter set based on each item in the access airport's item management database includes:

[0029] Access the airport's event management database and, starting from the current time, query events containing time periods within a preset time range; these events include route usage information and construction information.

[0030] Query the transit area corresponding to each item, and assign values ​​to the transit area according to the type of item and its time period;

[0031] For any region traversed, the cumulative values ​​assigned to the queried items are used to determine the adaptive detection parameter set.

[0032] As a further aspect of the present invention: the step of identifying the real-time region model based on the adaptive detection parameter set, locating abnormal candidate regions, simultaneously locating abnormal obstacles, and recursively adjusting the second acquisition accuracy based on the location results includes:

[0033] The information of each sub-region is read from the adaptive detection parameter set to compare the accuracy and anomaly detection threshold.

[0034] Query the sub-model corresponding to the sub-region in the real-time region model and the baseline region model respectively to obtain the real-time sub-model and the baseline sub-model;

[0035] The segmentation network is determined based on the information comparison accuracy; the segmentation grid is a three-dimensional grid, the grid unit is a rectangular prism, and the unit size is inversely proportional to the information comparison accuracy.

[0036] Based on the same origin, vectors pointing from the origin to each grid node in the segmented grid are created to obtain the real-time vector set and the reference vector set, which correspond to the real-time sub-model and the reference sub-model, respectively.

[0037] Compare the real-time vector set with the baseline vector set, calculate the set difference as the anomaly score, and when the anomaly score reaches the anomaly determination threshold, mark the real-time sub-model as an anomaly candidate region.

[0038] The abnormal candidate area is identified, abnormal obstacles are located, and the second acquisition accuracy is recursively adjusted based on the location results.

[0039] As a further aspect of the present invention: the steps of identifying abnormal candidate regions, locating abnormal obstacles, and recursively adjusting the second acquisition accuracy based on the location results include:

[0040] The system identifies abnormal candidate regions based on a pre-defined convolutional recognition model, and determines the name and location of the subject.

[0041] Query the standard location range corresponding to each entity name, and mark the location as an abnormal obstacle when it exceeds the standard location range;

[0042] The query counts the number of entities whose names are staff members, and the query counts the number of abnormal obstacles;

[0043] The number of staff members is assigned a weight, the number of abnormal obstacles is adjusted, and the second acquisition accuracy is recursively adjusted based on the adjusted number.

[0044] The weighting uses a negative indicator of the intermediate number of staff members, and the second collection accuracy is proportional to the adjusted number.

[0045] The present invention also provides a dynamic update system for airport airspace obstacle data, the system comprising:

[0046] The benchmark model creation module is used to determine the clearance area based on the initial range input by the manager, determine the first acquisition accuracy based on the clearance area, acquire regional benchmark data based on the first acquisition accuracy under standard working conditions, and construct a benchmark area model.

[0047] The real-time model creation module is used to periodically acquire real-time regional data based on a second acquisition precision in real-time operation, and construct a real-time regional model with time labels; the first acquisition precision uses peak parameters, and the second acquisition precision uses dynamic parameters;

[0048] The detection parameter determination module is used to access the airport's item management database and determine an adaptive detection parameter set based on each item in the item management database. The adaptive detection parameter set includes the information comparison accuracy of the real-time area model and the anomaly judgment threshold of the real-time area model.

[0049] The anomaly localization module is used to identify the real-time region model based on the adaptive detection parameter set, locate anomaly candidate areas, simultaneously locate abnormal obstacles, and recursively adjust the second acquisition accuracy according to the localization results.

[0050] The visualization module is used to mark the positioning results and visualize the real-time area model.

[0051] As a further aspect of the present invention: the benchmark model creation module includes:

[0052] An initial range receiving unit is used to receive the initial range input by the administrator; the initial range adopts the outline information in a preset map;

[0053] An initial region determination unit is used to determine an initial region based on the initial range;

[0054] The region segmentation unit is used to establish a connection channel with the block planning information database, obtain block information, and segment the initial region according to the block information to obtain sub-regions;

[0055] The adjacency differentiation and identification unit is used to locate the adjacent regions of each edge sub-region according to the block information, and to determine the merging status value according to the block type of the adjacent regions; the merging status value includes zero and one, which are used to represent non-merging and merging respectively;

[0056] The region merging unit is used to merge adjacent regions into the initial range based on the merging status value to obtain the clear area;

[0057] The first creation execution unit is used to determine the first acquisition accuracy based on the clearance area, acquire regional benchmark data based on the first acquisition accuracy under standard working conditions, and construct a benchmark area model.

[0058] As a further aspect of the present invention: the real-time model creation module includes:

[0059] The instruction generation unit is used to generate data acquisition instructions periodically in real-time operation.

[0060] The second creation execution unit is used to acquire a region image based on the second acquisition precision for any sub-region, and to construct a real-time region model based on the region image.

[0061] The model registration unit is used to register the real-time region model based on the reference region model;

[0062] The time stamp insertion unit is used to record the instruction generation time and insert it into the registered real-time region model to obtain a real-time region model containing time stamps.

[0063] Compared with the prior art, the beneficial effects of the present invention are: the present invention acquires relevant information of the airport through a data acquisition device with data transmission function, performs three-dimensional modeling of the airport with extremely high fidelity, and, combined with a pre-acquired standard model, quickly identifies abnormal areas and objects at different times, and then provides intuitive feedback to management personnel, with extremely high recognition efficiency and readability, making it easy to manage. Attached Figure Description

[0064] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention.

[0065] Figure 1 A flowchart illustrating the method for dynamically updating airport airspace obstacle data.

[0066] Figure 2 A block diagram illustrating the structure of an airport airspace obstacle data dynamic update system. Detailed Implementation

[0067] To make the technical problems to be solved, the technical solutions, and the beneficial effects of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present invention and are not intended to limit the present invention.

[0068] Figure 1 This is a flowchart illustrating a method for dynamically updating airport airspace obstacle data. In this embodiment of the invention, a method for dynamically updating airport airspace obstacle data includes:

[0069] Step S100: Determine the clearance area based on the initial range input by the manager, determine the first acquisition accuracy based on the clearance area, acquire the area benchmark data based on the first acquisition accuracy under standard working conditions, and construct the benchmark area model;

[0070] The manager refers to the person in charge of the airport airspace clearance process, and is also the user of the method provided by the technical solution of this invention. The manager inputs a range, called the initial range, to determine the airspace clearance area. The initial range can be determined by a coordinate set. The determined airspace clearance area is analyzed to determine the first acquisition accuracy. Under standard working conditions, the regional benchmark data is obtained based on the first acquisition accuracy to construct a benchmark regional model. This process generally occurs during unused periods and is used to obtain the standard state of the entire airport.

[0071] Step S200: In real-time operation, acquire real-time regional data periodically based on the second acquisition precision, and construct a real-time regional model with time labels; the first acquisition precision uses peak parameters, and the second acquisition precision uses dynamic parameters;

[0072] Real-time operating status refers to the usage period during which aircraft take-off and landing or other on-site activities occur. During this period, real-time data of the area is acquired periodically based on the second acquisition precision, and a model is constructed, called the real-time area model. Since the construction process is timed, time stamps are used. In fact, both the real-time area model and the baseline area model are three-dimensional models corresponding to the airport and use the same scale. Furthermore, the first acquisition precision is used to determine the standard state, and only one acquisition is needed during the non-use period until the next non-use period. Its acquisition precision needs to be higher, so peak parameters are used. The second acquisition precision is a timed acquisition process, which is a multiple acquisition process. The possibility of large changes within the airport is not high. Therefore, dynamic parameters are used to determine the dynamic acquisition process based on the actual situation.

[0073] Step S300: Access the airport's item management database and determine an adaptive detection parameter set based on each item in the item management database; the adaptive detection parameter set includes the information comparison accuracy of the real-time regional model and the anomaly judgment threshold of the real-time regional model;

[0074] The airport's event management database refers to the events that need to be handled during airport operations. This data is existing and stored in a database, which can be directly accessed in this invention. Analyzing each event in the event management database determines its impact on different areas, thereby determining the subsequent detection process for each area, represented by a dataset of detection parameters. The adaptive detection parameter set includes the information comparison accuracy of the real-time area model and the anomaly detection threshold of the real-time area model. Its purpose is to construct a rasterized model comparison process for partitions, allowing more resources to be applied to certain important areas. It has a similar function to the second acquisition accuracy using dynamic parameters, both reducing the amount of ineffective resources invested and indirectly improving resource utilization.

[0075] Step S400: Identify the real-time region model based on the adaptive detection parameter set, locate the abnormal candidate region, locate the abnormal obstacle simultaneously, and recursively adjust the second acquisition accuracy according to the positioning result;

[0076] After determining the adaptive detection parameter set, the real-time region model is identified based on the adaptive detection parameter set. The identification result consists of two steps: first, the location result of the abnormal candidate area, and second, the location result of the abnormal obstacle. The location result is the identification result of the airport obstacle. The second acquisition accuracy is recursively adjusted according to this identification result, so that the second acquisition accuracy becomes a dynamic parameter. The second acquisition accuracy is applied in the next real-time acquisition.

[0077] Step S500: Mark the positioning results and visualize the real-time region model;

[0078] The real-time region model is used to reflect the overall status of the airport. It is a three-dimensional model that can be directly interfaced with the display process. When displaying the real-time region model, the positioning results are marked in the real-time region model. The marking method can be to highlight the located abnormal candidate areas and abnormal obstacles.

[0079] Regarding step S100, the steps of determining the clearance area based on the initial range input by the manager, determining the first acquisition accuracy based on the clearance area, acquiring regional reference data based on the first acquisition accuracy under standard working conditions, and constructing a reference area model include:

[0080] Receive the initial range input by the administrator; the initial range uses the outline information in a preset map;

[0081] Determine the initial region based on the initial range;

[0082] Establish a connection channel with the block planning information database to obtain block information, and divide the initial area into sub-regions based on the block information;

[0083] The adjacent regions of each edge sub-region are located based on the block information, and the merging status value is determined based on the block type of the adjacent regions; the merging status value includes zero and one, which are used to represent non-merging and merging, respectively;

[0084] Based on the merged state value, adjacent regions are merged into the initial range to obtain the clear area;

[0085] The first acquisition accuracy is determined based on the clearance area. Under standard working conditions, the regional benchmark data is obtained based on the first acquisition accuracy, and a benchmark regional model is constructed.

[0086] The above describes a specific process for generating a baseline area model. The administrator inputs an initial range, which uses outline information from a preset map. This essentially limits the input method; however, inputting a set of coordinates is also a feasible technical solution. The initial range itself defines a boundary, and the area within that boundary is the initial area. Further, a connection is established with a block planning information database to obtain block information. This block information, existing planning information, can be represented by map features to determine the type of each plot of land near the airport. Based on this block information, the initial area is divided to obtain... Multiple sub-regions; based on the segmentation process, adjacent regions (not included in the initial region) of the edge sub-regions in the initial region are obtained, the block type of the adjacent regions is identified, and a merge status value is determined. The merge status value includes zero and one, which are used to represent not merging and merging, respectively. Based on the merge status value, it is determined whether to merge the adjacent regions into the initial region. After merging, it itself becomes a new edge sub-region, and the merge judgment process can be executed again. When the merge status value of all adjacent regions indicates not merging or the distance between the adjacent regions and the center point of the initial region is far enough, the merging process is stopped. The finally merged region is called the empty region.

[0087] Finally, a first acquisition accuracy is determined based on the clearance area, and regional benchmark data is acquired based on the first acquisition accuracy under standard operating conditions to construct a benchmark region model. Specifically, the steps of determining the first acquisition accuracy based on the clearance area, acquiring regional benchmark data based on the first acquisition accuracy under standard operating conditions, and constructing a benchmark region model include:

[0088] For any sub-area within the airspace clearance area, query its corresponding monitoring equipment; the corresponding monitoring equipment indicates the monitoring equipment whose monitoring range includes the sub-area, and the types include fixed monitoring equipment, drones, and remote sensing platforms.

[0089] Query the maximum accuracy of different types of monitoring devices, select the maximum value from the maximum accuracy, query the monitoring device corresponding to the maximum value, use the maximum value as its acquisition accuracy;

[0090] Under standard operating conditions, it generates image acquisition commands for various types of monitoring devices to acquire image data covering the airspace area.

[0091] A baseline region model is constructed based on the image data.

[0092] For any sub-area within the cleared airspace, query its corresponding monitoring equipment. The corresponding monitoring equipment refers to the monitoring equipment whose monitoring range includes the sub-area. Types include fixed monitoring equipment, drones, and remote sensing platforms. Simply put, any monitoring equipment capable of acquiring images of that sub-area is a corresponding monitoring equipment. It's conceivable that, due to the extremely high degree of freedom of drones and the extremely wide monitoring range of remote sensing platforms, the monitoring equipment corresponding to each sub-area could include both drones and remote sensing platforms. However, it doesn't necessarily include both, such as in concealed areas or no-fly zones. For fixed monitoring equipment, it is generally a camera, and its installation location is limited. Cameras are installed in most areas inside the airport, but because this invention also involves some adjacent areas, not every... Each sub-region corresponds to a fixed monitoring device. The maximum accuracy of different types of monitoring devices is queried, and the maximum value is selected from the maximum accuracy values. The monitoring device corresponding to the maximum value is then selected as the acquisition device, and the maximum value is used as its acquisition accuracy. In this invention's technical solution, due to permission restrictions or other reasons, the accuracy of a sub-region acquired by fixed monitoring devices, drones, and remote sensing platforms is different. The acquisition parameters of drones and remote sensing platforms are actually fixed, such as the drone's altitude and resolution. This makes the acquisition accuracy of drones and remote sensing platforms fixed. However, for cameras, a sub-region may be captured by a camera at a great distance. In this case, the acquisition accuracy will be lower. The acquisition accuracy can be understood as the actual area corresponding to a pixel in the captured image. The larger the actual area, the lower the acquisition accuracy.

[0093] After determining the acquisition equipment and acquisition accuracy, under standard operating conditions (the meaning of standard operating conditions has been explained previously), an image acquisition command is generated pointing to the corresponding monitoring equipment to obtain image data covering the airspace area. A baseline area model is then constructed based on the image data. It should be noted that the model construction process based on image data is not complicated, because the airport itself has a BIM model. By identifying the image data, determining the coverage information, and filling it into the BIM model, the corresponding three-dimensional model, called the baseline area model, can be obtained.

[0094] Regarding step S200, the step of periodically acquiring real-time regional data based on the second acquisition precision in real-time working state and constructing a real-time regional model with time labels includes:

[0095] Generate data acquisition commands periodically while operating in real-time.

[0096] For any sub-region, acquire a region image based on the second acquisition precision, and construct a real-time region model based on the region image;

[0097] The real-time region model is registered based on the aforementioned reference region model;

[0098] Record the time of instruction generation, insert the registered real-time region model, and obtain a real-time region model with time labels.

[0099] In one example of the technical solution of this invention, the construction process of the real-time region model is described. Data acquisition instructions are generated periodically in real-time operating mode. The meaning of real-time operating mode has been explained previously. For any sub-region, a region image is acquired based on a second acquisition precision. A real-time region model is constructed based on the region image. The process of constructing the model based on the image has also been explained previously and will not be repeated here. It should be noted that the real-time region model may experience some fluctuations during construction. Registration is performed on the real-time region model using the reference region model. The registration process ensures that they share the same origin and coordinate direction to prevent model deformation. Furthermore, the data acquisition instructions are generated periodically; that is, the current real-time region model is obtained every certain period, resulting in real-time region models at different times. The time of instruction generation is recorded, and a time stamp is inserted into the actual region model to indicate the state at which the actual region model represents.

[0100] Regarding step S300, the step of accessing the airport's item management database and determining the adaptive detection parameter set based on each item in the item management database includes:

[0101] Access the airport's event management database and, starting from the current time, query events containing time periods within a preset time range; these events include route usage information and construction information.

[0102] Query the transit area corresponding to each item, and assign values ​​to the transit area according to the type of item and its time period;

[0103] For any region traversed, the cumulative values ​​assigned to the queried items are used to determine the adaptive detection parameter set.

[0104] The above describes the process of determining the detection parameter set. It involves accessing the airport's event management database, starting from the current time, and querying events containing time periods within a preset time range. In this invention's technical solution, these events include route usage information and construction information. The process involves querying the transit areas corresponding to each event, assigning values ​​to these transit areas based on the event type and its time period. The assignment results represent the occupancy status of a specific sub-region. For any transit area, the assignment results of the queried events are accumulated. Based on these assignment results, the detection parameter set is determined, thus obtaining the specific collection parameters for each sub-region, referred to as the adaptive detection parameter set.

[0105] Regarding the assignment process, a base value is pre-determined for different types, and then a variable is determined by the time period. The longer the time period, the larger the variable. The product of the base value and the variable is calculated to obtain the assignment result reflecting the occupancy status. For any traversed area, the assignment results of the queried items are accumulated. Based on the assignment results, an adaptive detection parameter set is determined. The adaptive detection parameters include the information comparison accuracy of the real-time area model and the anomaly judgment threshold of the real-time area model. The larger the accumulated value, the higher the information comparison accuracy and the more accurate the comparison result. The smaller the anomaly judgment threshold, the easier it is to be judged as an anomaly candidate area, triggering the recognition process. The information comparison accuracy and the anomaly judgment threshold are both parameters in step S400, and are explained in detail below:

[0106] Regarding step S400, the step of identifying the real-time region model based on the adaptive detection parameter set, locating anomaly candidate regions, simultaneously locating abnormal obstacles, and recursively adjusting the second acquisition accuracy based on the location results includes:

[0107] The information of each sub-region is read from the adaptive detection parameter set to compare the accuracy and anomaly detection threshold.

[0108] Query the sub-model corresponding to the sub-region in the real-time region model and the baseline region model respectively to obtain the real-time sub-model and the baseline sub-model;

[0109] The segmentation network is determined based on the information comparison accuracy; the segmentation grid is a three-dimensional grid, the grid unit is a rectangular prism, and the unit size is inversely proportional to the information comparison accuracy.

[0110] Based on the same origin, vectors pointing from the origin to each grid node in the segmented grid are created to obtain the real-time vector set and the reference vector set, which correspond to the real-time sub-model and the reference sub-model, respectively.

[0111] Compare the real-time vector set with the baseline vector set, calculate the set difference as the anomaly score, and when the anomaly score reaches the anomaly determination threshold, mark the real-time sub-model as an anomaly candidate region.

[0112] The abnormal candidate area is identified, abnormal obstacles are located, and the second acquisition accuracy is recursively adjusted based on the location results.

[0113] In one example of the technical solution of this invention, the process of determining the second acquisition accuracy is described. The second acquisition accuracy itself is dynamic accuracy. In step S200, the second acquisition accuracy is directly applied for model creation. The second acquisition accuracy is actually determined in step S400. After the second acquisition accuracy is determined, it is used in the next model creation. Specifically, step S400 is actually divided into two steps: first, anomaly localization, and then determining the second acquisition accuracy based on the anomaly localization result. For the anomaly localization process, the sub-models corresponding to the sub-regions are queried in the real-time region model and the reference region model respectively to obtain the real-time sub-model and the reference sub-model. Then, the information comparison accuracy and anomaly judgment threshold of each sub-region are read from the adaptive detection parameter set. The segmentation network is determined according to the information comparison accuracy. Vectors pointing from the origin to each grid node in the segmentation grid are created based on the same origin to obtain the real-time vector set and the reference vector set, which correspond to the real-time sub-model and the reference sub-model, respectively. At this time, the real-time vector set and the reference vector set are compared, and the set difference is calculated as the anomaly degree. When the anomaly degree reaches the anomaly judgment threshold, the real-time sub-model is marked as an anomaly candidate area. Further identification is carried out in the anomaly candidate area to locate abnormal obstacles. The second acquisition accuracy is recursively adjusted according to the positioning results.

[0114] It should be noted that the segmented mesh is a three-dimensional mesh, and the mesh unit is a rectangular prism. The unit size is inversely proportional to the information comparison accuracy. If the rectangular prism is a cube with three sides of equal length, then the unit size is the side length. The higher the information comparison accuracy, the smaller the side length, the more mesh nodes are obtained, the more vectors in the vector set, and the more refined the set difference calculation. The calculation process of the set difference is not complicated because the segmented mesh is the same and the mesh nodes are one-to-one, which makes the vectors also one-to-one. The difference vector is calculated, and then the magnitude of the difference vector is calculated as the vector distance. The sum of the vector distances is calculated to obtain the set difference. The set difference is mapped to a preset interval to obtain the anomaly degree (of course, it can also be directly used as the anomaly degree).

[0115] Furthermore, the steps of identifying abnormal candidate regions, locating abnormal obstacles, and recursively adjusting the second acquisition accuracy based on the location results include:

[0116] The system identifies abnormal candidate regions based on a pre-defined convolutional recognition model, and determines the name and location of the subject.

[0117] Query the standard location range corresponding to each entity name, and mark the location as an abnormal obstacle when it exceeds the standard location range;

[0118] The query counts the number of entities whose names are staff members, and the query counts the number of abnormal obstacles;

[0119] The number of staff members is assigned a weight, the number of abnormal obstacles is adjusted, and the second acquisition accuracy is recursively adjusted based on the adjusted number.

[0120] The weighting uses a negative indicator of the intermediate number of staff members, and the second collection accuracy is proportional to the adjusted number.

[0121] In one example of the technical solution of this invention, the process of identifying abnormal candidate regions and determining the second acquisition accuracy is specifically described. Based on a preset convolutional recognition model, abnormal candidate regions are identified to determine the subject name and location. This process involves recognizing objects within a known abnormal region to determine each object and its location, referred to as the subject name and location. The standard location range corresponding to each subject name is queried. When the location exceeds the standard location range, it indicates that the subject is in an inappropriate place, and it is marked as an abnormal obstacle. It is worth noting that locating subjects includes locating workers, who are also subjects. When marking abnormal obstacles, workers not at their work positions are also queried. The number of subjects whose subject name is "worker" is queried, and the number of abnormal obstacles is queried. The number of "worker" subjects is weighted, and the number of abnormal obstacles is adjusted accordingly. The second acquisition accuracy is recursively adjusted based on the adjusted number.

[0122] Specifically, the weighting uses a negative indicator of the intermediate number of staff members. The second acquisition accuracy is directly proportional to the adjusted number. The actual meaning of this process is that if staff members are present near an abnormal obstacle, they are considered to be the personnel handling the abnormal obstacle. In this case, the number of abnormal obstacles can be slightly reduced. However, if there are too many or too few staff members, they may no longer be the personnel handling the abnormal obstacle, or the handling process may have brought some redundancy. The number of abnormal obstacles needs to be amplified. The number of abnormal obstacles is used to determine the second acquisition accuracy. The second acquisition accuracy is directly proportional to the adjusted number. In the process of determining the second acquisition accuracy, the amplified number is actually a numerical value that only serves a calculation purpose. In fact, the above weighting can also be understood as a coefficient that determines the second acquisition accuracy based on the number of abnormal obstacles. Too many or too few staff members, or too many abnormal obstacles, will all increase the second acquisition accuracy.

[0123] The relationship between weight and the number of staff is such that the weight increases when there are too many or too few staff, as this is considered a more dangerous situation requiring higher data collection accuracy. Conversely, the weight decreases when the number of staff is moderate, indicating greater safety but lower data collection accuracy. Therefore, the relationship between weight and the number of staff needs to utilize an intermediate indicator, meaning the closer the number is to a certain value (the middle), the better. This means that the closer the number is to the middle value, the greater the weight, which is clearly contrary to the desired outcome. Therefore, a negative indicator needs to be added. A feasible relationship is as follows:

[0124] ; where, in the formula, Indicates weight, Indicates the number of staff. For the preset optimal quantity, Used for regulation The range of values ​​for , This is a preset constant used to adjust the range of weight values.

[0125] Figure 2 This is a block diagram illustrating the structural composition of an airport airspace obstacle data dynamic update system. In this embodiment of the invention, an airport airspace obstacle data dynamic update system 10 includes:

[0126] The benchmark model creation module 11 is used to determine the clearance area based on the initial range input by the manager, determine the first acquisition accuracy based on the clearance area, acquire regional benchmark data based on the first acquisition accuracy under standard working conditions, and construct a benchmark area model.

[0127] The real-time model creation module 12 is used to periodically acquire real-time regional data based on the second acquisition precision in real-time working state, and construct a real-time regional model with time labels; the first acquisition precision adopts peak parameters, and the second acquisition precision adopts dynamic parameters;

[0128] The detection parameter determination module 13 is used to access the airport's item management database and determine an adaptive detection parameter set based on each item in the item management database; the adaptive detection parameter set includes the information comparison accuracy of the real-time area model and the anomaly judgment threshold of the real-time area model;

[0129] Anomaly localization module 14 is used to identify the real-time region model based on the adaptive detection parameter set, locate anomaly candidate areas, simultaneously locate abnormal obstacles, and recursively adjust the second acquisition accuracy according to the localization results.

[0130] The visualization module 15 is used to mark the positioning results and visualize the real-time area model.

[0131] Furthermore, the baseline model creation module 11 includes:

[0132] An initial range receiving unit is used to receive the initial range input by the administrator; the initial range adopts the outline information in a preset map;

[0133] An initial region determination unit is used to determine an initial region based on the initial range;

[0134] The region segmentation unit is used to establish a connection channel with the block planning information database, obtain block information, and segment the initial region according to the block information to obtain sub-regions;

[0135] The adjacency differentiation and identification unit is used to locate the adjacent regions of each edge sub-region according to the block information, and to determine the merging status value according to the block type of the adjacent regions; the merging status value includes zero and one, which are used to represent non-merging and merging respectively;

[0136] The region merging unit is used to merge adjacent regions into the initial range based on the merging status value to obtain the clear area;

[0137] The first creation execution unit is used to determine the first acquisition accuracy based on the clearance area, acquire regional benchmark data based on the first acquisition accuracy under standard working conditions, and construct a benchmark area model.

[0138] Specifically, the real-time model creation module 12 includes:

[0139] The instruction generation unit is used to generate data acquisition instructions periodically in real-time operation.

[0140] The second creation execution unit is used to acquire a region image based on the second acquisition precision for any sub-region, and to construct a real-time region model based on the region image.

[0141] The model registration unit is used to register the real-time region model based on the reference region model;

[0142] The time stamp insertion unit is used to record the instruction generation time and insert it into the registered real-time region model to obtain a real-time region model containing time stamps.

[0143] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for dynamically updating airport airspace obstacle data, characterized in that, The method includes: The clearance area is determined based on the initial range input by the manager, the first acquisition accuracy is determined based on the clearance area, and the regional benchmark data is obtained based on the first acquisition accuracy under standard working conditions to construct a benchmark area model. In real-time operation, real-time regional data is acquired periodically based on the second acquisition precision to construct a real-time regional model with time labels; the first acquisition precision uses peak parameters, and the second acquisition precision uses dynamic parameters. Access the airport's item management database and determine an adaptive detection parameter set based on each item in the database; the adaptive detection parameter set includes the information comparison accuracy of the real-time area model and the anomaly judgment threshold of the real-time area model; Based on the adaptive detection parameter set, the real-time region model is identified, abnormal candidate areas are located, abnormal obstacles are located simultaneously, and the second acquisition accuracy is recursively adjusted according to the location results. The location results are marked and visualized in the real-time region model.

2. The method for dynamically updating airport airspace obstacle data according to claim 1, characterized in that, The steps of determining the clearance area based on the initial range input by the manager, determining the first acquisition accuracy based on the clearance area, acquiring regional benchmark data based on the first acquisition accuracy under standard working conditions, and constructing a benchmark area model include: Receive the initial range input by the administrator; the initial range uses the outline information in a preset map; Determine the initial region based on the initial range; Establish a connection channel with the block planning information database to obtain block information, and divide the initial area into sub-regions based on the block information; The adjacent regions of each edge sub-region are located based on the block information, and the merging status value is determined based on the block type of the adjacent regions; the merging status value includes zero and one, which are used to represent non-merging and merging, respectively; Based on the merged state value, adjacent regions are merged into the initial range to obtain the clear area; The first acquisition accuracy is determined based on the clearance area. Under standard working conditions, the regional benchmark data is obtained based on the first acquisition accuracy, and a benchmark regional model is constructed.

3. The method for dynamically updating airport airspace obstacle data according to claim 2, characterized in that, The steps of determining the first acquisition accuracy based on the clearance area, acquiring regional reference data based on the first acquisition accuracy under standard working conditions, and constructing a reference region model include: For any sub-area within the airspace clearance area, query its corresponding monitoring equipment; the corresponding monitoring equipment indicates the monitoring equipment whose monitoring range includes the sub-area, and the types include fixed monitoring equipment, drones, and remote sensing platforms. Query the maximum accuracy of different types of monitoring devices, select the maximum value from the maximum accuracy, query the monitoring device corresponding to the maximum value, use the maximum value as its acquisition accuracy; Under standard operating conditions, it generates image acquisition commands for various types of monitoring devices to acquire image data covering the airspace area. A baseline region model is constructed based on the image data.

4. The method for dynamically updating airport airspace obstacle data according to claim 1, characterized in that, The steps of acquiring real-time regional data at regular intervals based on the second acquisition precision and constructing a real-time regional model with time labels in real-time working state include: Generate data acquisition commands periodically while operating in real-time. For any sub-region, acquire the region image based on the second acquisition precision, and construct a real-time region model based on the region image; The real-time region model is registered based on the aforementioned reference region model; Record the time of instruction generation, insert the registered real-time region model, and obtain a real-time region model with time labels.

5. The method for dynamically updating airport airspace obstacle data according to claim 1, characterized in that, The steps for determining the adaptive detection parameter set based on each item in the access airport's item management database include: Access the airport's event management database, starting from the current time, and query events containing time periods within a preset time range; the events include route usage information and construction information; Query the transit area corresponding to each item, and assign values ​​to the transit area according to the type of item and its time period; For any region traversed, the cumulative values ​​assigned to the queried items are used to determine the adaptive detection parameter set.

6. The method for dynamically updating airport airspace obstacle data according to claim 1, characterized in that, The steps of identifying the real-time region model based on the adaptive detection parameter set, locating abnormal candidate regions, simultaneously locating abnormal obstacles, and recursively adjusting the second acquisition accuracy based on the location results include: The information of each sub-region is read from the adaptive detection parameter set to compare the accuracy and anomaly judgment threshold. Query the sub-model corresponding to the sub-region in the real-time region model and the baseline region model respectively to obtain the real-time sub-model and the baseline sub-model; The segmentation network is determined based on the information comparison accuracy; the segmentation grid is a three-dimensional grid, the grid unit is a rectangular prism, and the unit size is inversely proportional to the information comparison accuracy. Based on the same origin, vectors pointing from the origin to each grid node in the segmented grid are created to obtain the real-time vector set and the reference vector set, which correspond to the real-time sub-model and the reference sub-model, respectively. Compare the real-time vector set with the baseline vector set, calculate the set difference as the anomaly score, and when the anomaly score reaches the anomaly determination threshold, mark the real-time sub-model as an anomaly candidate region. The abnormal candidate area is identified, abnormal obstacles are located, and the second acquisition accuracy is recursively adjusted based on the location results.

7. The method for dynamically updating airport airspace obstacle data according to claim 6, characterized in that, The steps of identifying abnormal candidate regions, locating abnormal obstacles, and recursively adjusting the second acquisition accuracy based on the location results include: The system identifies abnormal candidate regions based on a pre-defined convolutional recognition model, and determines the name and location of the subject. Query the standard location range corresponding to each entity name, and mark the location as an abnormal obstacle when it exceeds the standard location range; The query counts the number of entities whose names are staff members, and the query counts the number of abnormal obstacles; The number of staff members is assigned a weight, the number of abnormal obstacles is adjusted, and the second acquisition accuracy is recursively adjusted based on the adjusted number. The weighting uses a negative indicator of the intermediate number of staff members, and the second collection accuracy is proportional to the adjusted number.

8. A dynamic data update system for airport airspace obstacles, characterized in that, The system includes: The benchmark model creation module is used to determine the clearance area based on the initial range input by the manager, determine the first acquisition accuracy based on the clearance area, acquire regional benchmark data based on the first acquisition accuracy under standard working conditions, and construct a benchmark area model. The real-time model creation module is used to periodically acquire real-time regional data based on a second acquisition precision in real-time operation, and construct a real-time regional model with time labels; the first acquisition precision uses peak parameters, and the second acquisition precision uses dynamic parameters; The detection parameter determination module is used to access the airport's item management database and determine an adaptive detection parameter set based on each item in the item management database. The adaptive detection parameter set includes the information comparison accuracy of the real-time area model and the anomaly judgment threshold of the real-time area model. The anomaly localization module is used to identify the real-time region model based on the adaptive detection parameter set, locate anomaly candidate areas, simultaneously locate abnormal obstacles, and recursively adjust the second acquisition accuracy according to the localization results. The visualization module is used to mark the positioning results and visualize the real-time area model.

9. The airport airspace obstacle data dynamic update system according to claim 8, characterized in that, The baseline model creation module includes: An initial range receiving unit is used to receive the initial range input by the administrator; the initial range adopts the outline information in a preset map; An initial region determination unit is used to determine an initial region based on the initial range; The region segmentation unit is used to establish a connection channel with the block planning information database, obtain block information, and segment the initial region according to the block information to obtain sub-regions; The adjacency differentiation and identification unit is used to locate the adjacent regions of each edge sub-region according to the block information, and to determine the merging status value according to the block type of the adjacent regions; the merging status value includes zero and one, which are used to represent non-merging and merging respectively; The region merging unit is used to merge adjacent regions into the initial range based on the merging status value to obtain the clear area; The first creation execution unit is used to determine the first acquisition accuracy based on the clearance area, acquire regional benchmark data based on the first acquisition accuracy under standard working conditions, and construct a benchmark area model.

10. The airport airspace obstacle data dynamic update system according to claim 8, characterized in that, The real-time model creation module includes: The instruction generation unit is used to generate data acquisition instructions periodically in real-time operation. The second creation execution unit is used to acquire a region image based on the second acquisition precision for any sub-region, and to construct a real-time region model based on the region image. The model registration unit is used to register the real-time region model based on the reference region model; The time stamp insertion unit is used to record the instruction generation time and insert it into the registered real-time region model to obtain a real-time region model containing time stamps.

Images