Low-altitude safety monitoring method based on multi-dimensional information space-time correlation and related device

Abstract

The application discloses a kind of low-altitude safety monitoring method and related device based on multidimensional information space-time association, it is related to low-altitude supervision technical field, excavate the data correlation of multiple dimensions, identify implicit risk path, pre-judge traffic safety risk that appears in the flight process of unmanned aerial vehicle.Method includes: receiving the multi-source trajectory data uploaded by unmanned aerial vehicle in low-altitude supervision area, when the illegal information appears in the multi-source trajectory data, create the data sequence corresponding to current illegal information, the element in the data sequence includes airspace type, time period, illegal type;Search the element with higher occurrence frequency in the multiple sub-sequences segmented, obtain the daily illegal mode that unmanned aerial vehicle occurrence frequency in low-altitude supervision area exceeds the preset frequency range;Output and display the daily illegal mode.

Description

Technical Field

[0001] This invention relates to the field of low-altitude monitoring technology, and in particular to a low-altitude safety monitoring method and related device based on multi-dimensional information spatiotemporal correlation. Background Technology

[0002] With the rapid development of the low-altitude economy, the number of low-altitude aircraft such as drones has increased exponentially, and they are widely used in logistics transportation, agricultural and forestry plant protection, power line inspection and other fields. However, safety accidents caused by their disorderly flight (such as unauthorized flights disrupting air traffic, equipment failures and crashes) occur frequently, posing a serious challenge to airspace management and public safety. Currently, existing technologies in the field of low-altitude safety supervision have the following significant problems: Existing regulatory systems are mostly scattered across different departments or platforms. Data from multiple sources, such as drone flight records (e.g., GPS trajectory, flight altitude), equipment status data (e.g., battery level, sensor parameters), and airspace application information (e.g., reporting time, approved scope), are heterogeneous in format and inconsistent in standard, forming "information silos."

[0003] Due to the problem of information silos, traditional technologies mostly adopt static statistical or isolated analysis models, which cannot identify hidden risk paths, resulting in weak risk presentation and decision support. Because they mostly adopt static statistical or isolated analysis models, early warning of violations relies on manual review, making it difficult to automatically identify high-risk precursors such as "pilots with insufficient qualifications frequently applying for sensitive airspace" through algorithms, resulting in passive supervision. Summary of the Invention

[0004] In view of this, this application provides a low-altitude safety monitoring method and related device based on multi-dimensional information spatiotemporal correlation, which can explore the correlation of data in multiple dimensions, identify hidden risk paths, and predict traffic safety risks that may occur during the flight of UAVs.

[0005] The first aspect of this application provides a low-altitude safety monitoring method and related apparatus based on multi-dimensional information spatiotemporal correlation, the method comprising: Receive multi-source trajectory data uploaded by drones in the low-altitude surveillance area. When violation information appears in the multi-source trajectory data, create a data sequence corresponding to the current violation information. The elements of the data sequence include airspace type, time period, and violation type. Search for elements with high frequency in the multiple subsequences obtained by segmentation to obtain the daily violation pattern of drones appearing more frequently than the preset frequency range in the low-altitude monitoring area; Output and display the daily violation pattern.

[0006] A second aspect of this application provides a low-altitude safety monitoring device based on multi-dimensional information spatiotemporal correlation, the device comprising: The receiving module is used to receive multi-source trajectory data uploaded by UAVs in the low-altitude monitoring area. When violation information appears in the multi-source trajectory data, a data sequence corresponding to the current violation information is created. The elements of the data sequence include airspace type, time period, and violation type. The search module is used to search for elements with high frequency in the multiple sub-sequences obtained by segmentation, and to obtain the daily violation patterns of drones appearing in the low-altitude monitoring area that exceed the preset frequency range. The display module is used to output and display the daily violation pattern.

[0007] A third aspect of this application provides an electronic device, comprising: a processor and a memory, the processor and the memory being connected via a communication bus; wherein the processor is configured to call and execute a program stored in the memory; the memory is configured to store the program, the program being configured to implement the low-altitude safety monitoring method based on multidimensional information spatiotemporal correlation as provided in the first aspect of this application.

[0008] The fourth aspect of this application provides a computer-readable storage medium storing computer-executable instructions for performing the low-altitude safety monitoring method based on multidimensional information spatiotemporal correlation as provided in the first aspect of this application.

[0009] Compared to existing technologies, this application provides a low-altitude safety monitoring method and related device based on multi-dimensional information spatiotemporal correlation. Its beneficial effects are as follows: In the embodiments of this application, multi-source trajectory data sequences collected according to continuous flight time are divided into subsequences. The airspace type, time period, and violation type of UAV illegal flight in the subsequences are arranged according to the segmented time period. Then, the association rules of airspace type, time period, and violation type in the subsequences are mined by the Apriori algorithm to obtain the daily violation patterns of UAVs. By establishing the association chain of "aircraft-airspace-pilot-event", the hidden risk paths in UAV flight are identified, breaking the information silos of heterogeneous multi-source data. Compared with the static statistical or isolated analysis mode of single-dimensional monitoring of whether UAVs exceed electronic fences and only recording the time of a single violation event, the method proposed in this application has a more comprehensive risk identification and can provide early warning for illegal flights that are difficult to detect. Attached Figure Description

[0010] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0011] Figure 1 This is a schematic diagram of the low-altitude safety monitoring system; Figure 2 This is a flowchart of the steps of the low-altitude safety monitoring method based on multi-dimensional information spatiotemporal correlation proposed in the embodiments of this application; Figure 3 This is a functional block diagram of the low-altitude safety monitoring device based on multi-dimensional information spatiotemporal correlation proposed in the embodiments of this application. Detailed Implementation

[0012] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0013] In this application, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0014] Example 1 As can be seen from the background technology described above, existing technologies have significant shortcomings in multi-source data fusion, spatiotemporal correlation mining, three-dimensional risk presentation, and predictive control, making it difficult to meet the needs of refined and intelligent low-altitude safety supervision. There is an urgent need to build an intelligent system that integrates multi-dimensional data and incorporates spatiotemporal analysis to achieve an upgrade from passive response to proactive prevention and control.

[0015] Therefore, this application provides a low-altitude safety monitoring method based on multi-dimensional information spatiotemporal correlation, which is applied to a low-altitude safety monitoring system. Figure 1 This is a schematic diagram of the low-altitude safety monitoring system, such as... Figure 1 As shown, the low-altitude safety monitoring system includes a data access layer that supports API interfaces, direct database connections, file imports, and other methods to achieve real-time and batch access of multi-source data; among which, Data processing layer: Includes sub-modules such as format conversion, cleaning and calibration, and fusion processing, and outputs standardized datasets.

[0016] The algorithm analysis layer deploys algorithm modules such as time series pattern mining, spatiotemporal cube modeling, and graph computation analysis.

[0017] The storage layer uses a relational database (MySQL) to store structured data, a time-series database (InfluxDB) to store device state sequences, and a graph database (Neo4j) to store relational network data.

[0018] Interface layer: Provides RESTful APIs for front-end calls, supporting functions such as data query, analysis result retrieval, and early warning information push.

[0019] Figure 2 This is a flowchart illustrating the steps of the low-altitude safety monitoring method based on multi-dimensional information spatiotemporal correlation proposed in this application. Figure 2 As shown, the method includes: S101 to S103; Before performing step S101, the multi-source trajectory data can be preprocessed by performing the following steps S01-S03.

[0020] Optional S01: Convert the multi-source trajectory data of the UAV to be monitored into a predetermined format; the multi-source trajectory data includes: flight records, equipment status data, and airspace application information.

[0021] The process of converting the multi-source trajectory data of the UAV to be monitored into a predetermined format includes: converting flight records, equipment status data, and airspace application information into a unified format using ETL tools, including converting the spatial coordinates of flight records, equipment status data, and airspace application information into the WGS84 coordinate system; converting the timestamps of flight records, equipment status data, and airspace application information into UTC time; establishing standardized fields according to equipment models to form a standardized field library, and filling the equipment status data into the corresponding standardized fields.

[0022] The flight records include flight trajectory, drone flight altitude, and drone flight speed; equipment status data includes battery level, motor speed, and sensor parameters; and airspace application information includes reporting time, approved scope, and approval status.

[0023] Executing S01 to convert multi-source trajectory data into a predetermined format is a preprocessing procedure for the data. Preprocessing steps S11 to S14 can also be executed.

[0024] S12: Perform outlier cleaning on multi-source trajectory data, and identify outlier and missing data; S13: Use the 3σ criterion to identify and remove abnormal data that exceeds the reasonable range in multi-source trajectory data.

[0025] For example, data with a flight altitude greater than 1000 meters in the flight log can be removed, and data with a battery level less than -5% in the device status data can be removed.

[0026] S14: For missing values ​​in continuous multi-source trajectory data, forward imputation is used; for discrete multi-source trajectory data, mean imputation is used.

[0027] S02: Receive the flight trajectory of the drone to be monitored, and mark the flight trajectory with a unique identifier of the drone to be monitored.

[0028] S03: For flight trajectories with the same unique identifier, a Kalman filter algorithm is used for fusion calibration.

[0029] The fusion calibration of the GPS flight trajectory and BeiDou positioning flight trajectory of the same aircraft using the Kalman filter algorithm can be performed using formula (1): (1) It is the optimal value of the estimated flight trajectory at time k. It is the state transition matrix. It is Kalman gain. These are observed values. It is the optimal value of the estimated flight trajectory at time k-1. It is the observation matrix at time k, used to map the state vector to the observation space and establish a linear relationship between the state and the observation. It is the control matrix at time k, representing the degree of influence of the control input on the system state. It is the control input at time k, that is, the control quantity acting on the UAV system.

[0030] S101: Receive multi-source trajectory data uploaded by UAVs in the low-altitude monitoring area. When violation information appears in the multi-source trajectory data, create a data sequence corresponding to the current violation information. The elements in the data sequence include airspace type, time period, and violation type.

[0031] Executing S101 includes the following sub-steps: K11: Divide the multi-source trajectory data sequence collected according to continuous flight time into subsequences.

[0032] For example, using the SAX algorithm and the sliding window technique, with a window size of 5 minutes and a step size of 2 minutes, the continuous flight time series is divided into multiple time periods.

[0033] Each time period is sequentially taken as the current time period, and the sub-data corresponding to the current time period is Z-normalized using formula (2): (2); in, It is the sequence mean of the subsequence. It is the standard deviation of the sub-data for the current time period, reflecting the dispersion of the sub-data. It is the first The data values ​​are Z-standardized. Z-standardization transforms the original data into a standard normal distribution with a mean of 0 and a standard deviation of 1. Represents the first in the original sub-data Data points.

[0034] After Z-normalization, the subsequence is compressed into w feature points using the PAA algorithm.

[0035] The breakpoints are determined based on the Gaussian probability density function, and the feature points are mapped to a sequence containing 5 symbols to achieve data dimensionality reduction.

[0036] In one example, the sub-data can also be smoothed and denoised by using the Moen algorithm (window size k=3) to smooth the symbol sequence and remove high-frequency noise.

[0037] For example, the multi-source trajectory data sequence collected according to continuous flight time is divided into subsequences as follows: {[t1-t2]- Airport airspace-peak hours-over-high altitude flight, [t2-t3]- Military restricted area-night-illegal entry, [t3-t4]- Airport airspace-peak hours-unreported flight}.

[0038] S102: Search for elements with high frequency in the multiple subsequences obtained from the segmentation to obtain the daily violation pattern in the low-altitude monitoring area where the frequency of drones exceeds the preset frequency range.

[0039] For example, Table 1 shows the multiple subsequences obtained from the segmentation: Search for attributes that appear simultaneously in multiple subsequences exceeding a preset frequency range, and combine them to form a daily violation pattern: drones illegally entering the airport's airspace during peak hours.

[0040] Another example that might be found in the search results for routine violation patterns includes: multiple drones flying in unreported airspace, and aircraft flying non-compliantly in unapproved airspace.

[0041] One way to search for frequently occurring elements in the multiple subsequences obtained from segmentation is as follows: The Apriori algorithm was used to mine association rules, with a minimum support of 5% and a minimum confidence of 70%, to generate an association rule set of "spatial type → time period → violation type" as the daily violation pattern.

[0042] S103: Output and display the daily violation pattern.

[0043] This application's embodiments divide multi-source trajectory data sequences collected according to continuous flight time into subsequences. The airspace type, time period, and violation type of drone violations in the subsequences are arranged according to the segmented time periods. Then, the Apriori algorithm is used to mine the association rules of airspace type, time period, and violation type in the subsequences to obtain the daily violation patterns of drones. By establishing the association chain of "aircraft-airspace-pilot-event", hidden risk paths in drone flights can be identified, breaking down the information silos of heterogeneous multi-source data. Compared with single-dimensional monitoring of whether drones exceed electronic fences and static statistics or isolated analysis modes that only record the time of a single violation event, the method proposed in this application provides more comprehensive risk identification and enables early warning of difficult-to-detect violations.

[0044] In addition to mining the correlation chain of "aircraft-airspace-pilot-event" to obtain the hidden risk flight patterns of drones, another embodiment of this application also establishes a model for full life cycle health assessment of equipment status data to provide early warning of potential failures such as crash risk caused by motor aging.

[0045] The low-altitude safety monitoring method based on multidimensional information spatiotemporal correlation proposed in this application includes optional step S104: constructing an equipment health assessment model based on the LOF algorithm, using the KD tree algorithm to quickly find the k=10 nearest neighbor UAV equipment status data for each data point, and determining an abnormal state when the LOF value is >1.5.

[0046] In the above search calculation process, the reachability distance is calculated using formula (3): (3), among which, It is the Euclidean distance between p and o. Let be the k-th nearest neighbor distance of o. It refers to the reachable distance of the drone.

[0047] Local density and LOF value calculation, formula for calculating local reachable density (4): LOF= (4); among which, Expressed using formula (5); = (5); in, It is the data point currently being analyzed in the status data. yes Other data points within the k-neighborhood, This represents the size of the k-neighborhood of p, usually referring to the number of data points contained in the k-neighborhood of p, reflecting the scale of the neighborhood. The k-neighborhood set of p is usually composed of the k data points that are closest to p. It is the core of defining the local density range, and all density calculations are based on the points in this neighborhood.

[0048] The larger the local density and LOF value, the greater the difference between the state data of the target UAV and the state data of the k=10 nearest neighbor devices. When the LOF value is greater than 1.5, the target device is an abnormal device.

[0049] By establishing a monitoring system for the status of drone equipment through the above steps, and by comparing the local density of the target drone equipment with the average density of neighboring drone equipment, abnormal drone equipment can be detected, thereby achieving full life cycle supervision of drone equipment and providing early warning of potential faults, such as the risk of crashes caused by motor aging.

[0050] To dynamically display the changes in violation density in a certain airspace at different times, thereby intuitively presenting the spatiotemporal evolution of risk, the low-altitude safety monitoring method based on multi-dimensional information spatiotemporal correlation proposed in another embodiment of this application further includes step S105: In this embodiment, step S105 is performed: a three-dimensional spatial mapping is carried out to establish a corresponding three-dimensional space for the airspace where the UAV flies. M11: Latitude and longitude (x,y) are converted into planar coordinates (using Mercator projection), and time (t) is converted into a continuous variable in hours, forming three-dimensional coordinates (x,y,t).

[0051] Data standardization is performed using M11.

[0052] M12: For discretely distributed data points, spatiotemporal kriging is used for interpolation filling.

[0053] For example, interpolation filling is performed using formula (6): (6); in, These are weighting coefficients, determined by the spatiotemporal distance between the data points and the target point, and the reliability of the data. Given the attribute values ​​of the data points, where i represents the sequence of data points. Indicates spatial location ,time The multi-source data fusion estimate at the location.

[0054] S106: Divide the three-dimensional space into multiple cubic units, with each cubic unit serving as the basic analysis unit.

[0055] S107: For each cubic unit, aggregate the following violation indicators, which include: number of violation events, number of flights, average flight altitude, and number of equipment anomalies; S108: The violation indicators are calculated using the cumulative or average method to form the attribute value of each cubic unit.

[0056] S109: The attribute values ​​in the cubic cells are stored in netCDF format, and the drone violation data under a specific dimension is extracted in a three-dimensional manner.

[0057] Extracting drone violation data under a specific dimension using a three-dimensional indexing method is achieved through the following process: The (x,y,t) index represents a three-dimensional index method, enabling efficient spatiotemporal queries and supporting data slicing and extraction based on spatial range and time interval.

[0058] By fixing one dimension parameter in the (x, y, t) dimension, the dataset for the other two dimensions is extracted. The extracted data consists of unit attribute values ​​such as "number of violations, number of flights, average flight altitude, and number of equipment malfunctions". The specific method is as follows: Slice extraction logic: If a fixed time dimension t is used (e.g., "10:00-11:00 on June 1, 2024"), slices of x (longitude) and y (latitude) dimensions are extracted, which are the spatial distribution data of the regulatory area within this time period (e.g., "the number of violations in each 50m×50m grid within this hour"). If the spatial dimension x or y is fixed (e.g., "116.3°-116.4° East"), slices of the y and t dimensions are extracted, which are the time series data within this spatial range (e.g., "the average flight altitude change per hour within this longitude band"). If the height dimension is fixed (some extended models include height parameters), then spatiotemporal slices of x, y, and t are extracted.

[0059] The data content extracted using the above method: A slice extracts the set of attribute values ​​of all cubic cells under a specific combination of dimensions. For example, extracting a slice of "a certain airspace from 18:00 to 20:00 every day of the week" yields statistical values ​​such as the number of violations and equipment malfunctions of each grid in that airspace during that time period each day. This data can be directly used to analyze "the daily risk variation pattern of this airspace in the evening".

[0060] Compared to existing visualization methods, which are mostly two-dimensional (such as marking violation points on a map), these methods cannot intuitively present the spatiotemporal evolution of risks. For example, they cannot dynamically display the changes in violation density in a certain airspace at different times, nor can they present a three-dimensional risk distribution of "altitude-time-area," making it difficult for regulatory personnel to quickly locate key control areas and resulting in delayed emergency response.

[0061] This application embodiment establishes a three-dimensional mapping space for the airspace using the above method. The three-dimensional mapping space is divided into multiple cubic units. Based on the UAV violation data of the cubic units, the three-dimensional mapping space can be displayed. Then, the data stored in the cubic units is sliced ​​and calculated to dynamically display the change in violation density of a certain airspace at different times, presenting a three-dimensional risk distribution of "altitude-time-area" in a three-dimensional way, ensuring that regulatory personnel can quickly locate key control areas.

[0062] In this embodiment of the application, step S101 is to receive multi-source trajectory data uploaded by the drone. When violation information appears in the multi-source trajectory data, after creating the data sequence corresponding to the current violation information, steps S110-S111 can be executed to establish an abnormal information network, which includes aircraft nodes, airspace nodes, pilot nodes, and event nodes. Then, the abnormal information network is searched to identify community groups where abnormal behaviors are clustered.

[0063] S110: An anomaly information network is established using aircraft nodes, airspace nodes, pilot nodes, and event nodes as network nodes, and the lines connecting any two of these nodes as edges.

[0064] Aircraft node: Attributes include device ID, model, manufacturer, registration time, and current status (normal or abnormal).

[0065] Airspace node: Attributes include airspace ID, geographic boundary, control level, permitted flight altitude range, pilot node, and event node.

[0066] The attributes of a drone pilot node include identity ID, certification level, registration time, and number of historical violations.

[0067] Event nodes have attributes including event ID, type (violation or failure), occurrence time, and severity.

[0068] The edges of the abnormal information network include: Aircraft - Airspace, directed edge, attributes include flight time period, flight path length, and whether it is within the approved area.

[0069] Pilot - Aircraft: Undirected edge, attributes are binding time, number of operations, and last operation time; Pilot - Event: Directed edge, attributes are degree of association (0-1) and event processing result; Aircraft - Event: Directed edge, with attributes being the device state parameters at the time the event occurs and the cause of the event (preliminary determination).

[0070] S120: Use breadth-first search to search the abnormal information network and identify drone clusters with anomalous behavior.

[0071] The above search method identifies groups with "tight internal connections and sparse external connections" in abnormal information networks. Based on the objective fact that abnormal drones often appear in clusters, it discovers drone swarms with abnormal behavior.

[0072] The specific search process for executing S120 includes: Path analysis: Using the breadth-first search (BFS) algorithm, with "pilot-aircraft-airspace-event" as the path template, search for associated paths with a length ≤ 4, calculate the path weight (product of the attribute values ​​of each edge), and filter out high-risk paths with a weight > 0.6.

[0073] Community detection: Based on the Louvain algorithm, community partitioning is performed on the graph network by maximizing the modularity Q in formula (7): Q = δ (7); in It is an adjacency matrix. For node degree, For community tags, , is a node in the anomaly information network; m is half the sum of the weights of all edges in the anomaly information network.

[0074] Through the cluster analysis of the abnormal behavior described above, it can be found that when the same type of UAV repeatedly loses signal in similar airspaces, there is an inherent connection between the equipment model, airspace environment and failure event.

[0075] Example 2 Based on the low-altitude safety monitoring method based on multi-dimensional information spatiotemporal correlation provided in Embodiment 1 of this application, correspondingly, Embodiment 2 of this application also provides a low-altitude safety monitoring device based on multi-dimensional information spatiotemporal correlation. Figure 3 This is a functional block diagram of the low-altitude safety monitoring device based on multi-dimensional information spatiotemporal correlation proposed in the embodiments of this application, as follows: Figure 3 As shown, the device includes: The receiving module 301 is used to receive multi-source trajectory data uploaded by UAVs in the low-altitude monitoring area. When violation information appears in the multi-source trajectory data, a data sequence corresponding to the current violation information is created. The elements of the data sequence include airspace type, time period, and violation type. The search module 302 is used to search for elements with high frequency in the multiple sub-sequences obtained by segmentation, and to obtain the daily violation pattern in which the frequency of drones in the low-altitude monitoring area exceeds the preset frequency range. Display module 303 is used to output and display the daily violation pattern.

[0076] In one possible implementation, the device further includes: a lookup module, used to construct a device health assessment model based on the LOF algorithm, and to quickly look up the status data of the k=10 nearest neighbor drone devices for each data point using the KD tree algorithm, and to determine an abnormal state when the LOF value is greater than 1.5.

[0077] In one possible implementation, the apparatus further includes: The space creation module is used to create a corresponding three-dimensional space for the airspace where the UAV flies. The element division module is used to divide the three-dimensional space into multiple cubic elements, each of which serves as a basic analysis unit. The aggregation module is used to aggregate the following violation indicators for each cubic unit. The violation indicators include: number of violation events, number of flights, average flight altitude, and number of equipment anomalies. The calculation module is used to calculate the violation indicators using either cumulative or average calculation methods, forming the attribute value for each cubic cell; The extraction module is used to store attribute values ​​in cubic cells in netCDF format and extract drone violation data in a specific dimension using a three-dimensional approach.

[0078] In one possible implementation, the apparatus further includes: The network establishment module is used to establish an anomaly information network by using aircraft nodes, airspace nodes, pilot nodes, and event nodes as network nodes, and the lines connecting every two of these nodes as edges. The identification module is used to search the abnormal information network using a breadth-first search method to identify drone swarms with clusters of abnormal behaviors.

[0079] In one possible implementation, the device further includes a preprocessing module for converting the multi-source trajectory data of the UAV to be monitored into a predetermined format; the multi-source trajectory data includes: flight records, equipment status data, and airspace application information; upon receiving the flight trajectory of the UAV to be monitored, the module marks the flight trajectory with a unique identifier of the UAV to be monitored; and for flight trajectories with the same unique identifier, a Kalman filter algorithm is used for fusion calibration.

[0080] In one possible implementation, the spatial establishment module is specifically used to convert latitude and longitude into planar coordinates and time into continuous variables in hours to form three-dimensional coordinates; for discretely distributed data points, spatiotemporal kriging is used to interpolate and fill the three-dimensional coordinates.

[0081] The specific principles and execution processes of each unit in the low-altitude safety monitoring device disclosed in Embodiment 2 of this application are the same as those of the low-altitude safety monitoring method based on multi-dimensional information spatiotemporal correlation disclosed in Embodiment 1 of this application. Please refer to the corresponding part of the low-altitude safety monitoring method based on multi-dimensional information spatiotemporal correlation disclosed in Embodiment 1 of this application. They will not be repeated here.

[0082] Example 3 Embodiment 3 of this application provides an electronic device, including: a processor and a memory, the processor and the memory being connected via a communication bus; wherein, the processor is used to call and execute a program stored in the memory; the memory is used to store the program, the program being used to implement the low-altitude safety monitoring method based on multi-dimensional information spatiotemporal correlation as provided in Embodiment 1 of this application.

[0083] Example 4 Embodiment 4 of this application provides a computer-readable storage medium storing computer-executable instructions for executing the low-altitude safety monitoring method based on multi-dimensional information spatiotemporal correlation provided in Embodiment 1 of this application.

[0084] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computing software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0085] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the invention.

[0086] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A low-altitude safety monitoring method based on multi-dimensional information spatiotemporal correlation, characterized in that, The method includes: Receive multi-source trajectory data uploaded by drones in the low-altitude surveillance area. When violation information appears in the multi-source trajectory data, create a data sequence corresponding to the current violation information. The elements of the data sequence include airspace type, time period, and violation type. The data sequence is divided into multiple subsequences based on time periods and using a sliding window. Search for the attributes of elements that appear more frequently than a preset value and appear simultaneously in the multiple sub-sequences obtained by segmentation to obtain the daily violation patterns of drones in the low-altitude monitoring area. Output and display the daily violation pattern; The method further includes: An anomaly information network is established using aircraft nodes, airspace nodes, pilot nodes, and event nodes as network nodes, and the lines connecting any two of these nodes as edges. The aircraft node includes: device ID, model, manufacturer, registration time, and current status; The airspace nodes include: airspace ID, geographical boundary, control level, and permitted flight altitude range; Pilot nodes include: identity ID, certification level, registration time, and number of historical violations; Event nodes include: event ID, violation or fault type, time of occurrence, and severity; The edges of the abnormal information network include: Aircraft - Airspace, directed edge, attributes include flight time period, flight trajectory length, and whether it is within the approved area; Pilot - Aircraft: Undirected edge, with attributes including binding time, number of operations, and last operation time; Pilot - Event: Directed edge, attributes are degree of association and event handling result; Aircraft-Event: Directed edge, with attributes being the device state parameters at the time the event occurred and the cause of the event; The breadth-first search method is used to search the abnormal information network and identify drone swarms with clusters of abnormal behaviors. Specifically, the method of using breadth-first search to search the abnormal information network and identify drone swarms exhibiting abnormal behavior involves: based on the objective characteristic that abnormal drones appear in groups, a breadth-first search algorithm is used to search for groups in the abnormal information network whose internal connection density is greater than their external connection density, thereby identifying drone swarms exhibiting abnormal behavior; during the search process, using "pilot-aircraft-airspace-event" as the path template, a breadth-first search algorithm is used to search for associated paths with a length not exceeding 4, and the path weight of each associated path is calculated. The path weight is the product of the attribute values ​​of each edge in the path, and associated paths with a path weight greater than 0.6 are selected as high-risk paths; then, based on the Louvain algorithm, the graph network is divided into communities by maximizing the modularity Q. The formula for calculating the modularity Q is: in It is an adjacency matrix. For node degree, For community tags, , is a node in the abnormal information network; m is half the sum of the weights of all edges in the abnormal information network; through the high-risk path and the community division, the intrinsic relationship between the equipment model, airspace environment and failure event is obtained when the same type of UAV loses signal multiple times in similar airspace.

2. The method according to claim 1, characterized in that, The method further includes: A device health assessment model for the nearest neighbor UAV is constructed based on the LOF algorithm. The KD tree algorithm is used to find the status data of the target UAV. When the LOF value is greater than 1.5, the UAV corresponding to the currently searched data point is determined to be in an abnormal state.

3. The method according to claim 1, characterized in that, The method further includes: Establish a corresponding three-dimensional space for the airspace where the drone flies; The three-dimensional space is divided into multiple cubic units, and each cubic unit is used as the basic analysis unit. For each cubic unit, the following violation indicators are aggregated: number of violation events, number of flights, average flight altitude, and number of equipment anomalies. The violation indicators are calculated using either cumulative or average methods to form the attribute value for each cubic unit; The attribute values ​​in the cubic cells are stored in netCDF format, and the drone violation data under a specific dimension is extracted using a three-dimensional approach.

4. The method according to claim 1, characterized in that, The method further includes: Convert the multi-source trajectory data of the UAV to be monitored into a predetermined format; the multi-source trajectory data includes: flight records, equipment status data, and airspace application information. The flight trajectory of the drone to be monitored is received, and the flight trajectory is marked with a unique identifier of the drone to be monitored; For flight trajectories with the same unique identifier, a Kalman filter algorithm is used for fusion calibration.

5. The method according to claim 1, characterized in that, Establish a corresponding three-dimensional space for the airspace where the drone flies, including: Latitude and longitude are converted into planar coordinates, and time is converted into a continuous variable in hours to form a three-dimensional coordinate system. For discretely distributed data points, spatiotemporal kriging is used to interpolate and fill the three-dimensional coordinates.

6. A low-altitude safety monitoring device, characterized in that, The apparatus for carrying out the method as described in any one of claims 1 to 5 comprises: The receiving module receives multi-source trajectory data uploaded by drones in the low-altitude monitoring area. When violation information appears in the multi-source trajectory data, it creates a data sequence corresponding to the current violation information. The elements of the data sequence include airspace type, time period, and violation type. A segmentation module is used to segment the data sequence into multiple subsequences based on a time period and using a sliding window. The search module is used to search for the attributes of elements that appear more frequently than a preset value and appear simultaneously in the multiple sub-sequences obtained by segmentation, so as to obtain the daily violation patterns of drones in the low-altitude monitoring area. The display module is used to output and display the daily violation pattern; The network establishment module is used to establish an anomaly information network using aircraft nodes, airspace nodes, pilot nodes, and event nodes as network nodes, and the lines connecting every two nodes as edges. The aircraft nodes include: device ID, model, manufacturer, registration time, and current status; the airspace nodes include: airspace ID, geographical boundary, control level, and permitted flight altitude range; the pilot nodes include: identity ID, certification level, registration time, and historical violation count; and the event nodes include: event ID, violation or fault type, occurrence time, and severity. The edges of the anomaly information network include: aircraft-airspace (directed edge, attributes: flight time period, flight trajectory length, and whether within the approved range); pilot-aircraft (undirected edge, attributes: binding time, number of operations, and most recent operation time); pilot-event (directed edge, attributes: correlation degree and event processing result); and aircraft-event (directed edge, attributes: device status parameters at the time of the event, and event cause). The identification module is used to search the abnormal information network using a breadth-first search method to identify drone swarms with clusters of abnormal behaviors; Specifically, the identification module is used to: based on the objective characteristic that abnormal drones appear in clusters, use a breadth-first search algorithm to search for clusters with internal connection density greater than external connection density in the abnormal information network, in order to discover drone swarms exhibiting abnormal behavior; during the search process, using "pilot-aircraft-airspace-event" as the path template, use a breadth-first search algorithm to search for associated paths with a length no greater than 4, calculate the path weight of each associated path, where the path weight is the product of the attribute values ​​of each edge in the path, and filter associated paths with a path weight greater than 0.6 as high-risk paths; then, based on the Louvain algorithm, perform community partitioning on the graph network by maximizing the modularity Q, where the formula for calculating the modularity Q is: in It is an adjacency matrix. For node degree, For community tags, It is a node in the abnormal information network; It is half the sum of the weights of all edges in the abnormal information network; based on the high-risk path and the community division, the intrinsic relationship between the equipment model, airspace environment and failure event is obtained when the same type of UAV loses signal multiple times in similar airspace.

7. The apparatus according to claim 6, characterized in that, The device also includes a search module, which is used to construct a device health assessment model based on the LOF algorithm. The KD tree algorithm is used to search for the status data of the nearest neighboring drones for each data point. When the LOF value is greater than 1.5, the drone corresponding to the currently searched data point is determined to be in an abnormal state.

8. An electronic device, characterized in that, include: A processor and a memory are connected via a communication bus; wherein the processor is used to call and execute a program stored in the memory; The memory is used to store a program for implementing the method as described in any one of claims 1-5.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions for performing the method as described in any one of claims 1-5.

Images