An unmanned aerial vehicle management method and system for unmanned aerial vehicle multi-source data fusion

Abstract

This invention discloses a UAV management method and system based on multi-source data fusion. It acquires raw monitoring data from various UAV detection devices, performs spatiotemporal standardization on this heterogeneous data to establish a unified spatiotemporal benchmark, then establishes UAV mapping relationships through target association matching, and fuses the standardized monitoring data after association mapping. Finally, it performs airspace regulatory analysis based on the generated comprehensive UAV status data, outputting an airspace regulatory decision report containing monitoring information, authentication status, and early warning prompts. This invention effectively solves the problems of perception blind spots and insufficient data reliability inherent in single detection devices through multi-source data collaborative perception and fusion, significantly improving the ability to accurately monitor both cooperative and non-cooperative UAVs around the clock, and providing reliable technical support for airspace safety management.

Description

Technical Field

[0001] This invention relates to the field of drone safety supervision technology, and in particular to a drone management method and system that integrates multi-source drone data. Background Technology

[0002] With the widespread application of drones in logistics, emergency rescue, and urban patrol, the need for low-altitude airspace management is becoming increasingly urgent. Currently, drone regulation mainly relies on single-type detection methods or drones proactively reporting data. While these methods can monitor cooperative targets to some extent, they still face many challenges in practical application. Single detection methods, due to the inherent characteristics of their technical principles, have limitations in detection range, identification accuracy, and anti-interference capabilities, making it difficult to meet the needs of all-weather, all-round drone monitoring in complex urban environments. On the other hand, relying on drones proactively reporting data requires establishing unified data interface standards and a robust collaboration mechanism, which is difficult to implement and cannot effectively monitor non-cooperative targets not connected to the system.

[0003] While existing technologies have attempted to enhance monitoring capabilities by deploying multiple detection devices, significant differences exist in the spatiotemporal references, data formats, and information dimensions of data collected by different types of devices in practical applications, making effective integration of multi-source data difficult. This data heterogeneity prevents the system from fully utilizing the advantages of various devices, hindering the formation of a unified and accurate overall situational awareness. Particularly in scenarios requiring real-time tracking and authentication of drones, existing technologies often fail to provide sufficiently reliable comprehensive status information, thus limiting further improvements in airspace regulatory effectiveness.

[0004] Therefore, how to effectively solve the problem of integrating multi-source heterogeneous data while making full use of existing detection equipment, and achieve accurate perception and comprehensive assessment of the operational status of UAVs, has become an important technical problem that needs to be solved in the field of low-altitude safety management. Summary of the Invention

[0005] In view of this, the purpose of this invention is to propose a drone management method and system based on multi-source data fusion of drones.

[0006] To achieve the aforementioned technical objectives, in a first aspect, this application provides a UAV management method based on multi-source data fusion, comprising: The system acquires raw monitoring data from various drone detection devices, including positioning data collected by 5G-A sensing devices, identity data acquired by RID devices, radio frequency data captured by passive spectrum detection devices, and azimuth data measured by AOA monitoring devices. The original monitoring data is subjected to spatiotemporal standardization processing to generate standardized monitoring data with a unified spatiotemporal benchmark; Target association matching is performed on standardized monitoring data to generate association mapping relationships for UAVs; Multi-source data fusion is performed on the standardized monitoring data after correlation mapping to generate comprehensive UAV status data; Based on comprehensive UAV status data, airspace regulatory analysis is performed to generate an airspace regulatory decision report, which includes monitoring information, certification status, and early warning prompts.

[0007] In some embodiments, spatiotemporal standardization processing includes establishing a global time reference system to align timestamps from multiple sources, using a sliding window mechanism to achieve time synchronization, and using a spatial indexing algorithm to achieve unified transformation of geographic coordinates. The raw monitoring data undergoes spatiotemporal standardization processing to generate standardized monitoring data with a unified spatiotemporal benchmark, including: Receive raw monitoring data, establish a global time reference system, and uniformly convert the timestamps of monitoring data from different detection devices to the standard time coordinate system; The original monitoring data after time conversion is processed by a sliding time window mechanism to generate time-aligned first intermediate data. A spatial grid index is constructed on the first intermediate data, and the geographic coordinates collected by different UAV detection devices are uniformly converted into a standard plane coordinate system to generate a spatially unified second intermediate data. The second intermediate data is subjected to data integrity verification. For missing data fields, a data completion strategy based on device characteristics is used to complete the data and output standardized monitoring data.

[0008] In some embodiments, data integrity verification is performed on the second intermediate data, and data completion processing is performed on missing data fields using a data completion strategy based on device characteristics, including: Identify missing fields in the second intermediate data; When the missing field is the vertical flight speed field, the vertical flight speed data for the corresponding time window is obtained from the historical data of the RID device. If the RID device data is unavailable, the missing field is estimated by differential calculation based on the historical rate of change of true altitude or altitude. When the missing field is the track angle field, the associated entity is used to obtain historical track angle data. If historical track angle data exists, the missing field is filled in using a linear extrapolation algorithm. If there is no historical track angle data, the missing field is left as null. When the missing field is true height or altitude, the available data of 5G-A sensing devices, RID devices and passive spectrum detection devices are weighted and fused based on the device confidence weight to fill in the missing field. When the missing field is the horizontal flight speed field, the available equipment data is weighted and fused based on equipment reliability, or the missing field is estimated based on the rate of change of position using a continuous position difference algorithm. When the missing field is the product model field, the product model information is inherited from the available data of the passive spectrum detection device. If the passive spectrum detection device data is unavailable, the corresponding field information is inherited from the available data of the AOA monitoring device to fill in the missing field.

[0009] In some embodiments, target association matching includes direct association based on identifier consistency and indirect association based on spatiotemporal similarity. The dynamic time warping algorithm is used to calculate trajectory similarity, and motion feature matching is combined to establish the correspondence between multi-source data and UAV targets. Target association matching is performed on standardized monitoring data to generate association mapping relationships for UAVs, including: Receive standardized monitoring data, perform first-level association matching based on identifier consistency, and establish direct association when the identifiers reported by multiple detection devices are consistent and the spatiotemporal conditions meet the preset threshold, generating preliminary association results; When the identifier is unavailable or there is a conflict, the spatiotemporal similarity association algorithm is activated to perform secondary association matching. By calculating the spatial proximity and temporal synchronization of the target location, association hypotheses are established and a candidate association set is generated. The dynamic time warping algorithm is used to calculate the trajectory morphology similarity of the candidate association set. Combined with the consistency analysis of motion feature parameters, the preliminary association results are verified and optimized to generate the optimized association mapping relationship. For ambiguous association mappings, a multi-hypothesis tracking mechanism is established, multiple association schemes are evaluated based on the principle of probabilistic optimality, and the association result with the highest confidence is selected as the final association mapping. Based on the final association mapping relationship, establish a correspondence table between UAV targets and multi-source monitoring data, and output the final association mapping relationship.

[0010] In some embodiments, a multi-hypothesis tracking mechanism is established for ambiguous association mappings, evaluating multiple association schemes based on the principle of probabilistic optimality, and selecting the association result with the highest confidence as the final association mapping, including: For ambiguous association mappings, multiple association hypothesis schemes are generated. Each association hypothesis scheme contains different target-data correspondences, including: The location matching probability of each association hypothesis scheme is calculated based on the location likelihood function, which is constructed according to the spatial proximity of the target location. The motion matching probability of each associated hypothesis scheme is calculated based on the consistency of motion features, including the consistency measures of velocity vector and heading angle. By combining the location matching probability and the motion matching probability, a comprehensive confidence score is calculated for each association hypothesis scheme; The association hypothesis with the highest overall confidence score is selected as the optimal association hypothesis, and the final association mapping relationship is generated.

[0011] In some embodiments, multi-source data fusion includes fusing location information using an adaptive weighting algorithm, estimating motion state using a Kalman filter algorithm, and integrating identity features using a confidence voting mechanism to form a complete state description that includes spatial situation, motion parameters, and identity identifiers. Multi-source data fusion is performed on the standardized monitoring data after correlation mapping to generate comprehensive UAV status data, including: The system receives standardized monitoring data after association mapping and uses an adaptive weighting algorithm to fuse and calculate the location information from different UAV detection devices to generate fused spatial location data. The Kalman filter algorithm is used to estimate the motion state of the fused spatial location data, generating motion state data that includes velocity and acceleration. Based on a confidence-based voting mechanism, identity feature data from different UAV detection devices are integrated to generate unified identity identification data; The fused spatial location data, motion status data, and unified identity data are integrated to form a complete state description that includes spatial situation, motion parameters, and identity, and outputs comprehensive UAV state data.

[0012] In some embodiments, standardized monitoring data after association mapping is received, and an adaptive weighting algorithm is used to fuse and calculate the location information from different UAV detection devices to generate fused spatial location data, including: Obtain the location information of each UAV detection device in the standardized monitoring data after association mapping, and determine the initial confidence weight of each device based on the device type; The initial confidence weights are dynamically adjusted based on the real-time signal quality parameters of each device's location information to generate dynamic weight coefficients for each device. The location information of each device is weighted and averaged based on dynamic weighting coefficients to generate weighted and fused latitude and longitude coordinates. The weighted and fused latitude and longitude coordinates are subjected to inverse coordinate transformation to convert the plane coordinate system coordinates into geographic coordinate system coordinates, generating fused spatial location data. Outlier detection and filtering are performed on the fused spatial location data to output the final spatial location data.

[0013] In some embodiments, the fused spatial location data, motion state data, and unified identity data are integrated to form a complete state description including spatial situation, motion parameters, and identity, including: Receive fused spatial location data, motion status data, and unified identity data to establish a UAV status description framework; Map the latitude and longitude coordinates, true altitude, and altitude information in the spatial location data to the spatial situation field of the state description framework; Map the horizontal flight speed, vertical flight speed, and track angle information in the motion state data to the motion parameter fields of the state description framework; Map the serial number and product model information in the unified identity data to the identity field of the status description framework; Perform data consistency verification on the mapped state description framework to ensure that the logical relationships between the fields conform to the flight characteristics of the UAV. Based on the verification results, a complete state description including spatial situation, motion parameters and identification is generated, and the comprehensive state data of the UAV is output.

[0014] In some embodiments, airspace regulatory analysis includes inferring flight intent through a flight path prediction algorithm, assessing compliance in conjunction with an airspace rule base, and determining threat levels based on a risk assessment model; Based on comprehensive UAV status data, airspace regulatory analysis is performed to generate airspace regulatory decision reports, including: Receive comprehensive status data of UAVs, and use trajectory prediction algorithms to make short-term and medium-term predictions of the UAVs' flight trajectories, generating flight intent analysis results; The flight intent analysis results are matched and compared with the no-fly zones, restricted flight zones and special control areas in the airspace rule base to generate flight compliance assessment results; Based on the comprehensive status data and flight intention analysis results of the UAV, a threat level score is calculated through a risk assessment model. The threat level score comprehensively considers flight altitude anomaly, speed anomaly, heading deviation, and airspace sensitivity. Warning level information is generated based on flight compliance assessment results and threat level scores. Warning prompts are generated when illegal or high-risk flights are detected. By integrating flight intent analysis results, flight compliance assessment results, threat level scores, and warning level information, an airspace regulatory decision report is generated that includes monitoring information, certification status, and warning prompts.

[0015] In a second aspect, the present invention also provides a UAV management system for multi-source data fusion, applicable to the method described in the first aspect. The system includes a data acquisition module, a spatiotemporal standardization processing module, a target association matching module, a multi-source data fusion module, and an airspace supervision and analysis module. The data acquisition module is used to acquire raw monitoring data from various UAV detection devices, including positioning data collected by 5G-A sensing devices, identity data acquired by RID devices, radio frequency data captured by passive spectrum detection devices, and azimuth data measured by AOA monitoring devices. The spatiotemporal standardization processing module is used to perform spatiotemporal standardization processing on the raw monitoring data to generate standardized monitoring data with a unified spatiotemporal reference. The target association matching module is used to perform target association matching on the standardized monitoring data to generate association mapping relationships for UAVs. The multi-source data fusion module is used to perform multi-source data fusion on the standardized monitoring data after association mapping to generate comprehensive UAV status data. The airspace supervision and analysis module is used to perform airspace supervision and analysis based on the comprehensive UAV status data to generate an airspace supervision and decision report, which includes monitoring information, authentication status, and early warning prompts.

[0016] Compared with existing technologies, the present invention, employing the above technical solution, has the following beneficial effects: By acquiring positioning data collected by 5G-A sensing devices, identity data acquired by RID devices, radio frequency data captured by passive spectrum detection devices, and azimuth data measured by AOA monitoring devices, spatiotemporal standardization processing is performed on these heterogeneous data to establish a unified spatiotemporal benchmark; a drone mapping relationship is established through target association matching, and multi-source data fusion is performed on the standardized monitoring data after association; finally, airspace supervision analysis is performed based on the generated comprehensive drone status data, outputting an airspace supervision decision report containing monitoring information, authentication status, and early warning prompts. Through multi-source data collaborative perception and fusion, the present invention effectively solves the problems of perception blind spots and insufficient data reliability existing in single detection devices, significantly improving the all-time accurate supervision capability of cooperative and non-cooperative drones, and providing reliable technical support for airspace safety management. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention 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 some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram of steps S101 to S105 of the drone management method described in the specific implementation embodiment; Figure 2This is a schematic diagram of the multi-source data fusion process described in the specific implementation method; Figure 3 This is a flowchart illustrating the spatiotemporal standardization process described in the specific implementation method; Figure 4 This is a schematic diagram of the trajectory similarity calculation process described in the specific implementation method; Figure 5 This is a schematic diagram of the structure of the unmanned aerial vehicle management system described in the specific implementation method.

[0019] The reference numerals for the above figures are as follows: 1. Unmanned Aerial Vehicle (UAV) Management System; 11. Data acquisition module; 12. Spatiotemporal standardization processing module; 13. Target association and matching module; 14. Multi-source data fusion module; 15. Airspace monitoring and analysis module. Detailed Implementation

[0020] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be particularly noted that the following embodiments are for illustrative purposes only and do not limit the scope of the invention. Similarly, the following embodiments are only some, not all, embodiments of the present invention, and all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Please see Figure 1 In a first aspect, this embodiment provides a drone management method based on multi-source data fusion, comprising: S101. Acquire raw monitoring data from various drone detection devices, including positioning data collected by 5G-A sensing devices, identity data acquired by RID devices, radio frequency data captured by passive spectrum detection devices, and azimuth data measured by AOA monitoring devices. S102. Perform spatiotemporal standardization processing on the raw monitoring data to generate standardized monitoring data with a unified spatiotemporal benchmark; S103. Perform target association matching on standardized monitoring data to generate association mapping relationships for UAVs; S104. Perform multi-source data fusion on the standardized monitoring data after correlation mapping to generate comprehensive UAV status data; S105. Based on the comprehensive status data of UAVs, conduct airspace supervision and analysis to generate an airspace supervision and decision report, which includes monitoring information, certification status, and early warning prompts.

[0022] In step S101, the raw monitoring data refers to a heterogeneous data set collected from various UAV detection devices. The 5G-A sensing device acquires positioning data through millimeter-wave signal analysis, the RID device obtains identity data by parsing UAV broadcast signals, the passive spectrum detection device captures radio frequency data by passively intercepting communication signals, and the AOA monitoring device obtains azimuth data by measuring the signal angle of arrival using a multi-antenna array. These devices have complementary characteristics in terms of data dimension, accuracy, and update frequency, providing multi-dimensional information support for subsequent data fusion. This step, through a parallel data acquisition mechanism, achieves comprehensive perception coverage of both cooperative and non-cooperative UAVs.

[0023] In step S102, spatiotemporal standardization refers to the crucial preprocessing step of unifying heterogeneous data to a standard spatiotemporal reference. This is achieved by establishing a global time reference system to align timestamps from multiple sources, using a sliding window mechanism for time synchronization, and employing a spatial indexing algorithm for unified geographic coordinate transformation. Preferably, time synchronization uses UTC millisecond-level timestamps, and spatial coordinate transformation uses the WGS84 standard coordinate system. This process effectively solves the inconsistency problem of multi-source data in terms of time and spatial reference, establishing a unified comparison benchmark for subsequent data association.

[0024] In step S103, target association matching refers to the process of establishing a correspondence between multi-source monitoring data and individual UAVs. Through identifier consistency analysis and spatiotemporal similarity calculation, a mapping relationship between UAVs and monitoring data is constructed. Preferably, when identifiers are unavailable or conflicting, an association algorithm based on location proximity and time synchronization is used. This step, through a multi-level association strategy, effectively solves the problem of inconsistent target identification caused by device heterogeneity, ensuring that multi-source data from the same UAV can be correctly collected.

[0025] In step S104, multi-source data fusion refers to the process of integrating and calculating the standardized monitoring data after correlation. An adaptive weighted algorithm is used to fuse location information, Kalman filtering is used to estimate motion state, and a confidence-based voting mechanism is used to integrate identity features. Preferably, location fusion considers device confidence and real-time signal quality, and motion state estimation uses complementary data from multiple sensors. This fusion process fully utilizes the advantages of each device to generate a complete state description including spatial situation, motion parameters, and identity identifiers.

[0026] In step S105, airspace regulatory analysis refers to the process of risk assessment and decision support based on comprehensive UAV status data. Flight intentions are inferred through trajectory prediction algorithms, compliance is assessed using an airspace rule base, and threat levels are determined based on a risk assessment model. Preferably, the generation of early warning alerts comprehensively considers multiple dimensions such as flight altitude anomalies, speed anomalies, and airspace sensitivity. This analysis process provides airspace managers with decision-making support including monitoring information, certification status, and early warning alerts.

[0027] This embodiment establishes a complete multi-source data fusion processing chain for unmanned aerial vehicles (UAVs), achieving fully automated processing from raw data acquisition to regulatory decision-making. Standardized data interfaces and a unified processing framework are used between each step to ensure the system's good scalability and compatibility. Through collaborative perception and intelligent analysis of multi-source data, the accuracy, real-time performance, and comprehensiveness of UAV supervision are effectively improved, providing reliable technical support for low-altitude safety management.

[0028] In some embodiments, spatiotemporal standardization processing includes establishing a global time reference system to align timestamps from multiple sources, using a sliding window mechanism to achieve time synchronization, and using a spatial indexing algorithm to achieve unified transformation of geographic coordinates. The raw monitoring data undergoes spatiotemporal standardization processing to generate standardized monitoring data with a unified spatiotemporal benchmark, including: Receive raw monitoring data, establish a global time reference system, and uniformly convert the timestamps of monitoring data from different detection devices to the standard time coordinate system; The original monitoring data after time conversion is processed by a sliding time window mechanism to generate time-aligned first intermediate data. A spatial grid index is constructed on the first intermediate data, and the geographic coordinates collected by different UAV detection devices are uniformly converted into a standard plane coordinate system to generate a spatially unified second intermediate data. The second intermediate data is subjected to data integrity verification. For missing data fields, a data completion strategy based on device characteristics is used to complete the data and output standardized monitoring data.

[0029] In this embodiment, the global time reference system is constructed using UTC standard time, uniformly converting the different timestamps of each detection device to a time coordinate system with millisecond-level precision. Preferably, timestamp alignment uses a network time protocol to synchronize the clocks of each device. This process solves the problem of data timing disorder caused by clock differences among multiple source devices, providing an accurate time reference for subsequent data fusion.

[0030] The sliding time window mechanism groups data into segments of fixed duration. Preferably, the window size can be dynamically adjusted according to the device data update frequency, and monitoring data falling within the same window are considered as observations at the same time. The window sliding step size can be set to half the window size to ensure data continuity. This mechanism effectively handles the temporal asynchrony of multi-source data and generates time-aligned intermediate data.

[0031] The spatial grid index uses a regular grid partitioning method to establish the spatial query structure. Unified geographic coordinate transformation is achieved by converting latitude and longitude data from different coordinate systems to the UTM plane coordinate system. Preferably, the grid size is set according to the monitoring accuracy requirements, and spatial queries are accelerated by establishing a mapping relationship between latitude and longitude and grid coding. This process solves the problem of inconsistent spatial reference systems among multi-source devices, achieving unified management of spatial data.

[0032] Data integrity verification involves traversing data fields to detect missing values. A data completion strategy based on device characteristics employs differentiated completion methods according to the data features of each detection device. Preferably, for the vertical velocity field unique to RID devices, estimation is performed using historical data interpolation or differential calculation based on the rate of change of height when data is missing. For the location field shared by multiple devices, a weighted average method is used to fuse available data. This process ensures the integrity of the dataset, providing a reliable data foundation for subsequent analysis.

[0033] This embodiment systematically solves the inconsistency problem of multi-source heterogeneous data in the temporal and spatial dimensions by establishing a complete spatiotemporal standardization processing flow. Each processing step employs standardized algorithms and configurable parameters, ensuring both the reliability of the processing results and providing flexibility to adapt to different application scenarios. This standardization process lays a solid data foundation for subsequent target association and data fusion, significantly improving the data processing capabilities and reliability of the UAV monitoring system.

[0034] In some embodiments, data integrity verification is performed on the second intermediate data, and data completion processing is performed on missing data fields using a data completion strategy based on device characteristics, including: Identify missing fields in the second intermediate data; When the missing field is the vertical flight speed field, the vertical flight speed data for the corresponding time window is obtained from the historical data of the RID device. If the RID device data is unavailable, the missing field is estimated by differential calculation based on the historical rate of change of true altitude or altitude. When the missing field is the track angle field, the associated entity is used to obtain historical track angle data. If historical track angle data exists, the missing field is filled in using a linear extrapolation algorithm. If there is no historical track angle data, the missing field is left as null. When the missing field is true height or altitude, the available data of 5G-A sensing devices, RID devices and passive spectrum detection devices are weighted and fused based on the device confidence weight to fill in the missing field. When the missing field is the horizontal flight speed field, the available equipment data is weighted and fused based on equipment reliability, or the missing field is estimated based on the rate of change of position using a continuous position difference algorithm. When the missing field is the product model field, the product model information is inherited from the available data of the passive spectrum detection device. If the passive spectrum detection device data is unavailable, the corresponding field information is inherited from the available data of the AOA monitoring device to fill in the missing field.

[0035] In this embodiment, missing field identification employs a field status marking mechanism, assigning a valid status identifier to each data field. When a field value is detected to be empty or outside a reasonable range, it is marked as missing, triggering the corresponding completion process. This process ensures the reliability and quality of the dataset, providing complete data support for subsequent analysis.

[0036] To address the issue of missing vertical flight speed fields, the completion process prioritizes retrieving available data for the corresponding time window from the historical data of the RID device. If RID data is unavailable, an estimation method is used based on the historical rate of change of true altitude or altitude, employing a differential calculation approach. An approximate vertical speed value is obtained by dividing the altitude difference between adjacent time points by the time interval. This completion method fully leverages the advantages of RID devices in vertical speed measurement and provides a reliable backup solution through physical relationship derivation.

[0037] For missing track angle fields, historical track angle data sequences for the UAV are obtained through associated entities. If historical data exists, a linear extrapolation algorithm is used to predict and complete the data based on the track angle change trends at recent time points; if no historical data is available, null values ​​are retained to avoid introducing erroneous information. This approach ensures both data continuity and accuracy.

[0038] When the true altitude or altitude field is missing, the available device data is weighted and fused based on device confidence weights. Preferably, the weight allocation considers the accuracy characteristics of each device in altitude measurement, and the fused altitude value is calculated through a weighted average. This fusion method effectively utilizes the complementarity of multi-source data, improving the reliability of altitude data.

[0039] Two options are available for completing the horizontal flight speed field: weighted fusion based on equipment reliability, or an estimation algorithm based on continuous position differences. The former assigns weights to the data fusion according to the measurement accuracy of each device, while the latter calculates an approximate speed value by dividing the distance change between adjacent positions by the time interval. These two methods can be flexibly selected based on data availability to ensure the integrity of the speed data.

[0040] When the product model field is missing, the model information is first inherited from the data of the passive spectrum detection device. If that is not available, the corresponding information is obtained from the data of the AOA monitoring device. This inheritance mechanism ensures the consistency of identification information and avoids the problem of model information conflicts for the same UAV in different data sources.

[0041] This embodiment systematically solves the problem of missing values ​​in the multi-source data acquisition process by establishing differentiated completion strategies for different field characteristics. Each completion method is designed based on the device's measurement characteristics and physical relationships, ensuring both the rationality of the completed data and fully utilizing the complementary advantages of multi-source data. This data integrity guarantee mechanism significantly improves the quality and availability of UAV status data, providing a reliable data foundation for subsequent regulatory decisions.

[0042] In some embodiments, target association matching includes direct association based on identifier consistency and indirect association based on spatiotemporal similarity. The dynamic time warping algorithm is used to calculate trajectory similarity, and motion feature matching is combined to establish the correspondence between multi-source data and UAV targets. Target association matching is performed on standardized monitoring data to generate association mapping relationships for UAVs, including: Receive standardized monitoring data, perform first-level association matching based on identifier consistency, and establish direct association when the identifiers reported by multiple detection devices are consistent and the spatiotemporal conditions meet the preset threshold, generating preliminary association results; When the identifier is unavailable or there is a conflict, the spatiotemporal similarity association algorithm is activated to perform secondary association matching. By calculating the spatial proximity and temporal synchronization of the target location, association hypotheses are established and a candidate association set is generated. The dynamic time warping algorithm is used to calculate the trajectory morphology similarity of the candidate association set. Combined with the consistency analysis of motion feature parameters, the preliminary association results are verified and optimized to generate the optimized association mapping relationship. For ambiguous association mappings, a multi-hypothesis tracking mechanism is established, multiple association schemes are evaluated based on the principle of probabilistic optimality, and the association result with the highest confidence is selected as the final association mapping. Based on the final association mapping relationship, establish a correspondence table between UAV targets and multi-source monitoring data, and output the final association mapping relationship.

[0043] In this embodiment, direct association based on identifier consistency achieves rapid matching by comparing identification information such as serial numbers reported by different devices. When the identifiers reported by multiple devices are consistent and the spatiotemporal conditions meet a preset threshold, a direct association is established, making full use of the inherent identification information of the devices and achieving efficient and accurate target association.

[0044] When identifiers are unavailable or conflicting, a spatiotemporal similarity association algorithm is activated for secondary association matching. This algorithm establishes association assumptions by calculating the spatial proximity and temporal synchronization of the target locations. Spatial proximity is calculated based on Euclidean distance, and temporal synchronization is determined by a time difference threshold. Preferably, the spatial distance threshold can be dynamically adjusted according to the UAV speed. This association method effectively solves the matching problem when identifiers are missing or conflicting.

[0045] The Dynamic Time Warping (RTW) algorithm is used to calculate trajectory shape similarity by finding the optimal alignment path between two trajectories to measure their shape similarity. Consistency analysis of motion feature parameters, including consistency measures of velocity vectors and heading angles, is combined to further validate and optimize the initial association results. This algorithm can effectively handle trajectory data with different sampling frequencies, improving association accuracy.

[0046] A multi-hypothesis tracking mechanism is established to address ambiguous association mappings. This involves generating multiple possible association schemes and evaluating them based on the principle of probabilistic optimality. The confidence score of each association scheme is calculated by combining the location matching probability and the motion feature consistency probability, and the scheme with the highest overall confidence score is selected as the final result. This mechanism effectively resolves the association ambiguity problem in dense target environments.

[0047] The correspondence table between UAV targets and multi-source monitoring data records the mapping relationship between each UAV entity and its corresponding multi-source data entry. This structured data organization method provides a clear foundation for subsequent data fusion.

[0048] This embodiment systematically solves the problem of establishing the correspondence between multi-source data and UAV targets by establishing a multi-level association matching mechanism. From direct identifier matching to spatiotemporal similarity analysis, and then to trajectory morphology verification and multiple hypothesis evaluation, a complete association processing chain is formed. This hierarchical and progressive association strategy ensures both matching efficiency and association accuracy, providing a reliable association foundation for the fusion of multi-source UAV data.

[0049] In some embodiments, a multi-hypothesis tracking mechanism is established for ambiguous association mappings, evaluating multiple association schemes based on the principle of probabilistic optimality, and selecting the association result with the highest confidence as the final association mapping, including: For ambiguous association mappings, multiple association hypothesis schemes are generated. Each association hypothesis scheme contains different target-data correspondences, including: The location matching probability of each association hypothesis scheme is calculated based on the location likelihood function, which is constructed according to the spatial proximity of the target location. The motion matching probability of each associated hypothesis scheme is calculated based on the consistency of motion features, including the consistency measures of velocity vector and heading angle. By combining the location matching probability and the motion matching probability, a comprehensive confidence score is calculated for each association hypothesis scheme; The association hypothesis with the highest overall confidence score is selected as the optimal association hypothesis, and the final association mapping relationship is generated.

[0050] In this embodiment, the multi-hypothesis tracking mechanism handles ambiguous mapping relationships by generating multiple possible association schemes. Each association hypothesis scheme defines a different combination of target-data correspondences. By evaluating multiple possibilities in parallel, it avoids the limitations of a single association decision, making it particularly suitable for scenarios with dense multiple targets or conflicting data.

[0051] The location likelihood function is constructed based on a Gaussian distribution model, measuring the matching probability by calculating the spatial distance between the target location and the observation location. The closer the distance, the higher the probability value; specifically, the probability distribution shape is adjusted using a preset spatial error standard deviation parameter. This calculation fully considers the differences in position measurement accuracy between different devices.

[0052] Motion matching probability is evaluated by comparing the consistency of velocity vectors and heading angles. Velocity consistency is measured using the cosine of the angle between the vectors, while heading angle consistency is calculated using the absolute value of the angle difference. The two are weighted and combined to form the final motion matching probability, with the weights dynamically adjusted according to the importance of the motion features.

[0053] The overall confidence score is calculated as a weighted product of the location matching probability and the motion matching probability. Preferably, the weighting takes into account the reliability differences of location and motion features in different scenarios. A time decay factor is also incorporated into the score calculation, giving more weight to recent data.

[0054] The optimal association scheme is selected using the maximum confidence principle, choosing the scheme with the highest overall score from all candidate schemes as the final result. When multiple schemes have similar scores, historical association stability can be used as an auxiliary decision-making factor to ensure the continuity of the association results.

[0055] This embodiment systematically solves the problem of association ambiguity in complex scenarios by establishing a complete probabilistic evaluation framework. Each evaluation index is designed based on physical characteristics and statistical principles, ensuring both the scientific rigor of the evaluation and providing flexibility to adapt to different scenarios. This multi-hypothesis tracking mechanism significantly improves the accuracy and robustness of target association, providing a reliable foundation for subsequent data processing.

[0056] In some embodiments, multi-source data fusion includes fusing location information using an adaptive weighting algorithm, estimating motion state using a Kalman filter algorithm, and integrating identity features using a confidence voting mechanism to form a complete state description that includes spatial situation, motion parameters, and identity identifiers. Multi-source data fusion is performed on the standardized monitoring data after correlation mapping to generate comprehensive UAV status data, including: The system receives standardized monitoring data after association mapping and uses an adaptive weighting algorithm to fuse and calculate the location information from different UAV detection devices to generate fused spatial location data. The Kalman filter algorithm is used to estimate the motion state of the fused spatial location data, generating motion state data that includes velocity and acceleration. Based on a confidence-based voting mechanism, identity feature data from different UAV detection devices are integrated to generate unified identity identification data; The fused spatial location data, motion status data, and unified identity data are integrated to form a complete state description that includes spatial situation, motion parameters, and identity, and outputs comprehensive UAV state data.

[0057] In this embodiment, the adaptive weighted algorithm achieves optimized fusion by dynamically adjusting the weights of the location information of each device. The weight allocation is based on the inherent precision of the device type and real-time signal quality parameters, and the fused position is calculated through a weighted average. This algorithm fully utilizes the advantages of high-precision devices while reducing the impact of low-quality data.

[0058] The Kalman filter algorithm uses a state-space model to optimally estimate the motion state. The algorithm consists of two phases: prediction and update. The prediction phase extrapolates the state based on the motion model, while the update phase refines the estimate using observation data. Through recursive computation, it effectively filters out observation noise, generating smooth and accurate velocity and acceleration data.

[0059] The confidence-based voting mechanism reaches a consensus by comparing the identity feature data provided by various devices. Each device's voting weight is set based on its identity recognition reliability; the result is adopted when a majority of devices report a consistent identity. This mechanism effectively resolves identity information conflicts and ensures the accuracy of identity identification.

[0060] The data integration process organizes three types of information—spatial location, motion status, and identification—into a structured description. Spatial status includes latitude, longitude, and altitude information; motion parameters record speed and direction data; and identification clearly identifies the target's characteristics. Through unified data formats and standardized field definitions, a complete description of the UAV's status is formed.

[0061] This embodiment achieves effective integration of multi-source information by establishing a multi-level data fusion framework. Each fusion method is designed for the characteristics of different types of data, ensuring fusion accuracy while fully utilizing the complementarity of multi-source data. Through a systematic fusion strategy, the completeness and reliability of UAV status description are significantly improved, providing an accurate data foundation for airspace supervision.

[0062] In some embodiments, standardized monitoring data after association mapping is received, and an adaptive weighting algorithm is used to fuse and calculate the location information from different UAV detection devices to generate fused spatial location data, including: Obtain the location information of each UAV detection device in the standardized monitoring data after association mapping, and determine the initial confidence weight of each device based on the device type; The initial confidence weights are dynamically adjusted based on the real-time signal quality parameters of each device's location information to generate dynamic weight coefficients for each device. The location information of each device is weighted and averaged based on dynamic weighting coefficients to generate weighted and fused latitude and longitude coordinates. The weighted and fused latitude and longitude coordinates are subjected to inverse coordinate transformation to convert the plane coordinate system coordinates into geographic coordinate system coordinates, generating fused spatial location data. Outlier detection and filtering are performed on the fused spatial location data to output the final spatial location data.

[0063] In this embodiment, the initial confidence weights are pre-set based on the inherent accuracy characteristics of each device type. Preferably, RID devices are assigned higher weights because they directly acquire UAV positioning information, while passive spectrum detection devices are assigned medium weights based on their signal quality. This weight allocation strategy fully utilizes the technical advantages of each device, laying the foundation for subsequent fusion calculations.

[0064] Dynamic weight adjustment is achieved by analyzing real-time signal quality parameters. These parameters include indicators such as signal-to-noise ratio, data update frequency, and signal strength. Weight adjustment coefficients are calculated using a multi-parameter comprehensive evaluation model. This mechanism can adapt to environmental changes and maintain the stability of the fusion effect when equipment performance fluctuates.

[0065] The weighted average calculation multiplies the coordinates of each device's location by its dynamic weight coefficient, sums the results, and then divides by the total weight to obtain the merged coordinates. Double-precision floating-point arithmetic is used during the calculation to ensure accuracy, while abnormal weights are normalized to prevent numerical overflow. This calculation method guarantees accuracy while exhibiting good numerical stability.

[0066] The inverse coordinate transformation process converts the fused results from the planar coordinate system back to the geographic coordinate system. A reverse mapping algorithm is used to recover latitude and longitude coordinates, ensuring the output data is compatible with standard geographic information systems. The transformation process takes into account the curvature correction of the Earth, improving coordinate accuracy in large-scale monitoring scenarios.

[0067] Outlier detection uses statistical analysis to identify data points that deviate from the normal range, while filtering smooths data fluctuations using a moving average algorithm. These post-processing steps effectively improve the reliability and usability of location data, providing high-quality spatial information for subsequent applications.

[0068] This embodiment achieves intelligent integration of multi-source location data by establishing a complete adaptive weighted fusion process. From initial weight setting to dynamic adjustment, and then to precise calculation and post-processing, a systematic location fusion scheme is formed. This scheme maintains the efficiency of the algorithm while ensuring the accuracy of the fusion results, significantly improving the reliability of UAV spatial positioning.

[0069] In some embodiments, the fused spatial location data, motion state data, and unified identity data are integrated to form a complete state description including spatial situation, motion parameters, and identity, including: Receive fused spatial location data, motion status data, and unified identity data to establish a UAV status description framework; Map the latitude and longitude coordinates, true altitude, and altitude information in the spatial location data to the spatial situation field of the state description framework; Map the horizontal flight speed, vertical flight speed, and track angle information in the motion state data to the motion parameter fields of the state description framework; Map the serial number and product model information in the unified identity data to the identity field of the status description framework; Perform data consistency verification on the mapped state description framework to ensure that the logical relationships between the fields conform to the flight characteristics of the UAV. Based on the verification results, a complete state description including spatial situation, motion parameters and identification is generated, and the comprehensive state data of the UAV is output.

[0070] In this embodiment, the UAV state description framework is constructed using a structured data model, defining three core information fields: the spatial situation field records location-related data, the motion parameter field describes dynamic characteristics, and the identification field stores target recognition information. This classification and organization facilitates the systematic management and application of data.

[0071] The data mapping process fills the framework with the fused data according to the field correspondences. Latitude and longitude coordinates are in decimal degree format, and true altitude and elevation are in meters to ensure data standardization. When mapping motion parameters, unit consistency and accuracy preservation are considered, and identity information directly adopts the consensus results.

[0072] Data consistency verification uses logical rules to validate the reasonableness of each field. The checks include the matching of position and direction of movement, the coordination of altitude and speed, and the consistency of identity characteristics. When anomalies are detected, a data correction process is initiated to ensure that the status description matches the actual flight characteristics of the drone.

[0073] A complete status description is generated based on a validated data framework and organized according to a preset format. The description includes required core fields and optional extended fields, forming a standardized record of comprehensive UAV status data.

[0074] This embodiment achieves the organic unification of multi-source information by establishing a systematic data integration process. From framework construction to data mapping and consistency verification, a complete state description generation chain is formed. This solution ensures the structured and standardized nature of UAV state data, providing a reliable data foundation for subsequent airspace surveillance and analysis.

[0075] In some embodiments, airspace regulatory analysis includes inferring flight intent through a flight path prediction algorithm, assessing compliance in conjunction with an airspace rule base, and determining threat levels based on a risk assessment model; Based on comprehensive UAV status data, airspace regulatory analysis is performed to generate airspace regulatory decision reports, including: Receive comprehensive status data of UAVs, and use trajectory prediction algorithms to make short-term and medium-term predictions of the UAVs' flight trajectories, generating flight intent analysis results; The flight intent analysis results are matched and compared with the no-fly zones, restricted flight zones and special control areas in the airspace rule base to generate flight compliance assessment results; Based on the comprehensive status data and flight intention analysis results of the UAV, a threat level score is calculated through a risk assessment model. The threat level score comprehensively considers flight altitude anomaly, speed anomaly, heading deviation, and airspace sensitivity. Warning level information is generated based on flight compliance assessment results and threat level scores. Warning prompts are generated when illegal or high-risk flights are detected. By integrating flight intent analysis results, flight compliance assessment results, threat level scores, and warning level information, an airspace regulatory decision report is generated that includes monitoring information, certification status, and warning prompts.

[0076] In this embodiment, the trajectory prediction algorithm infers future flight paths by analyzing historical trajectory data. Short-term predictions are based on extrapolation of the current motion state, while medium-term predictions combine common flight pattern recognition. The algorithm considers the maneuverability characteristics of the UAV and generates possible flight intent analysis results through curve fitting and pattern matching.

[0077] The airspace rule base stores the geographical boundary information of no-fly zones, restricted-fly zones, and specially controlled areas. The matching and comparison process analyzes spatial relationships to determine the intersection between the predicted trajectory and the controlled area, generating detailed compliance assessment results. This process monitors potential spatial conflict risks in real time.

[0078] The risk assessment model uses a multi-factor weighted method to calculate the threat level score. Flight altitude anomaly is determined by comparing with standard flight altitude thresholds; speed anomaly is assessed based on the reasonable speed range corresponding to the airspace type; heading deviation analyzes the difference between the flight direction and the expected path; and airspace sensitivity is set according to the regional importance classification. The weights of each factor are configured according to actual regulatory needs.

[0079] The generation of warning level information comprehensively considers compliance assessments and threat level scores. When entry into a no-fly zone is detected or the threat score exceeds a preset threshold, the corresponding level of warning is automatically triggered. Warning levels can be graded according to urgency to facilitate differentiated response measures.

[0080] The regulatory decision-making report integrates all analytical results into a structured output. Monitoring information includes real-time status and predictive data, authentication status records identity verification results, and early warning prompts clearly state the risk type and action recommendations. The report uses a standardized format for easy subsequent processing and application.

[0081] This embodiment establishes a complete airspace regulatory analysis chain, achieving intelligent regulation from status monitoring to risk assessment. Each analysis stage is based on comprehensive judgment using multi-dimensional data, ensuring both comprehensive regulation and precise decision support. This systematic regulatory approach effectively improves the efficiency and safety of airspace management.

[0082] Please see Figure 5In a second aspect, this embodiment also provides a UAV management system 1 for multi-source data fusion of UAVs, applicable to the method described in the first aspect. The system includes a data acquisition module 11, a spatiotemporal standardization processing module 12, a target association matching module 13, a multi-source data fusion module 14, and an airspace supervision and analysis module 15. The data acquisition module 11 is used to acquire raw monitoring data from various UAV detection devices. The raw monitoring data includes positioning data collected by 5G-A sensing devices, identity data acquired by RID devices, radio frequency data captured by passive spectrum detection devices, and azimuth data measured by AOA monitoring devices. The spatiotemporal standardization processing module 12 is used to perform spatiotemporal standardization processing on the raw monitoring data to generate standardized monitoring data with a unified spatiotemporal reference. The target association matching module 13 is used to perform target association matching on the standardized monitoring data to generate association mapping relationships of UAVs. The multi-source data fusion module 14 is used to perform multi-source data fusion on the standardized monitoring data after association mapping to generate comprehensive UAV status data. The airspace supervision and analysis module 15 is used to perform airspace supervision and analysis based on the comprehensive UAV status data to generate an airspace supervision and decision report. The airspace supervision and decision report includes monitoring information, authentication status, and early warning prompts.

[0083] In this embodiment, the data acquisition module 11 acquires raw monitoring data through interfaces of multiple types of detection devices, and the data acquired by each device has complementary characteristics in terms of dimension and accuracy. The spatiotemporal standardization processing module 12 establishes a unified spatiotemporal benchmark and eliminates spatiotemporal differences between multi-source data through time alignment and coordinate transformation. The target association and matching module 13 establishes the correspondence between data and targets and uses a multi-level association strategy to solve the problem of inconsistent identifiers. The multi-source data fusion module 14 performs integrated calculations on the associated data and generates a complete state description through adaptive weighting and state estimation. The airspace supervision and analysis module 15 performs risk assessment based on the fusion results and generates a supervision decision report containing early warning information.

[0084] This system utilizes a modular architecture to handle the entire process of UAV multi-source data processing, from acquisition to regulatory decision-making. After the data acquisition module 11 acquires heterogeneous data from multiple sources, the spatiotemporal standardization module 12 unifies the data benchmark, the target association and matching module 13 establishes correct correspondences, the multi-source data fusion module 14 integrates and optimizes status information, and finally, the airspace regulatory analysis module 15 outputs regulatory decisions. This systematic processing architecture effectively overcomes the limitations of single-device monitoring, enabling comprehensive supervision of both cooperative and non-cooperative UAVs, and providing reliable technical support for airspace safety management.

[0085] Furthermore, the following examples can be derived from the above technical solutions: By analyzing and summarizing the detection principles and the reliability of detection data of various types of drone detection equipment, this study supports the deployment planning and design of detection equipment, specifically including 5G-A integrated sensing equipment, RID equipment, passive spectrum detection equipment, and AOA monitoring equipment.

[0086] The 5G-A integrated sensing device actively illuminates drones using millimeter-wave signals (24.25–27.5 GHz) from 5G-A base stations. By analyzing the propagation delay, Doppler shift, and beam angle changes of the echo signal, it calculates the target's distance, speed, and azimuth. This integrated sensing technology combines communication and sensing functions into a single hardware device, achieving "dual-purpose network." The device relies on dense base station deployment, achieving centimeter-level positioning accuracy (a significant improvement over 5G's meter-level accuracy), with a horizontal speed error of <0.5 m / s, making it particularly suitable for low-altitude urban monitoring. Based on the high-precision clock of the 5G network, the timestamp error is <1 ms, making it suitable for multi-source data fusion. It achieves a 99% recognition rate for "low, slow, and small" drones and can distinguish drones from birds through micro-motion feature analysis. However, its vertical resolution is relatively low (with a large true altitude error) and it depends on base station coverage density; additional equipment is needed in rural blind areas.

[0087] RID devices are remote identification devices that receive Remote ID signals (similar to avionics license plates) actively broadcast by drones and directly analyze their built-in location, speed, and identity data. The broadcast protocol follows the ASTM F3411 standard, with a signal frequency band of 2.4 GHz / 5.8 GHz. The device's serial number and product model are directly derived from the drone's firmware, making them unique and tamper-proof, ensuring the highest level of reliability. Based on the drone's built-in barometer and IMU (Inertial Measurement Unit), vertical accuracy reaches ±0.5 meters, superior to external detection devices. Through continuous position differential calculation, the dynamic update frequency is ≥1 Hz, providing strong real-time data performance. However, it relies on compliant drone broadcast signals; unauthorized drones may disable the RID function, requiring the use of passive devices. Passive spectrum detection devices can passively intercept communication signals between drones and remote controllers (such as Wi-Fi, image transmission, and remote control links). They identify targets through radio frequency fingerprint analysis (signal modulation method, frequency hopping mode) and the micro-Doppler effect, without actively emitting electromagnetic waves, thus offering strong stealth capabilities. Product model identification is based on radio frequency fingerprint database matching (such as DJI drone-specific communication protocols), with an accuracy rate >90%. Motion direction is deduced from the signal arrival phase difference (TDOA) and micro-Doppler frequency shift, with an angular error <3°. It can still operate in electromagnetically complex environments (such as urban areas) with a lower false alarm rate than optoelectronic devices. However, this device cannot directly acquire altitude data and requires integration with other equipment; it also fails when the signal is blocked.

[0088] AOA (Angle of Arrival) monitoring equipment is used for angle of arrival monitoring. By deploying a multi-antenna array, it measures the phase difference of the target signal arriving at each antenna and calculates the azimuth angle through triangulation. It is often linked with TDOA (Diverterless Target AoA) technology to improve positioning accuracy. This equipment has an azimuth error of <2° under multi-base station collaboration, which is better than a single sensor. It identifies drones by binding signal characteristics, assisting passive devices in target identification. Its response time is ≤3 seconds, making it suitable for rapid early warning. However, this equipment cannot independently provide distance / altitude data and requires fusion with radar or 5G-A; its performance degrades in rain and fog.

[0089] Based on the above analysis of the characteristics of various types of equipment, equipment can be selectively planned and deployed according to the actual terrain and application scenario requirements: For sensitive core areas, multiple types of equipment can be deployed in combination. For example, an international airport adopted a "5G-A sensing + photoelectric detection" fusion solution, which reduced the false alarm rate to 0.1 times / hour and successfully distinguished drones from ground vehicles; For areas with special regulatory needs, a certain type of equipment can be selectively deployed according to the terrain environment, such as passive spectrum equipment with strong anti-interference capabilities that can be deployed in areas with complex electromagnetic environments and countermeasure scenarios.

[0090] However, only some devices can accurately identify the drone's serial number. The raw serial number represents a unique identifier for the detected flight target. Different types of detection devices have different definitions of this unique identifier, and different drone detection technologies acquire "serial numbers" with different concepts and sources. Therefore, for the same drone, the identifier definitions identified by these devices are usually not the same: RID devices directly obtain the "official drone serial number required by regulations"; passive spectrum detection obtains a "fingerprint or feature code" based on radio signal characteristics, which is usually not equal to the official serial number; 5G-A detection obtains the drone's "IMEI / PEI" (and possibly SIM card information) as a 5G terminal device, which is different from the official drone serial number; AOA monitoring devices do not define a unique identifier for the flight target and generate it automatically by the monitoring device. Therefore, data fusion cannot be performed based on the drone's serial number.

[0091] To correlate data from various detection devices and identify whether the trajectory positioning reported by these devices represents the same drone during the same flight, this invention employs spatiotemporal correlation (location and time) for data binding. ① Spatiotemporal correlation: Targets detected by different devices at the same time and location can be highly suspected to be the same drone. For example, when the RID receiver obtains the serial number and location, the system can search for flying targets detected by 5G-A near the same location, or by feature signals detected by passive detection, and confirm that they are the same target based on preset correlation relationships; ②Trajectory comparison: Combined with trajectory comparison analysis algorithms, it is further determined whether the flight activities are for a unified target; ③ Trajectory Reverse Lookup: After confirming that they are the same target, their respective flight trajectories are completed by reverse lookup using the unique identifier of their respective monitoring equipment.

[0092] Specifically, this invention sets up a multi-source data fusion scheme for UAV detection equipment. Based on the given characteristics of the equipment data (including data integrity and missing data), a systematic fusion framework is constructed. By integrating heterogeneous data from devices such as 5G-A, RID devices, passive spectrum detection, and AOA, more comprehensive, accurate, and real-time UAV status information (such as position, speed, heading, etc.) is generated, while handling issues such as data missing data, time asynchrony, and correlation.

[0093] This multi-source data fusion solution leverages the strengths of each device (e.g., RID devices provide official serial numbers and complete speed information, while passive spectrum detection provides product model information) to compensate for the shortcomings of individual devices (e.g., AOA devices lack information such as altitude and speed). It needs to handle asynchronous data timing and differences in update frequency (assuming different device update frequencies, but the data time field can be used for synchronization), and address noise and missing values. Serial numbers are used as the primary association key, but when serial numbers are unreliable (e.g., missing or inconsistent), spatiotemporal association (location and time) is used for data binding, combined with trajectory comparison algorithms for target confirmation and data association. The fusion solution consists of four core modules: data preprocessing, data association, data fusion, and result output. It is easy to expand and maintain, prioritizing lightweight algorithms (e.g., Kalman filters, weighted averages) to ensure real-time performance. For complex scenarios, machine learning models (e.g., sensor confidence learning based on historical data) can be introduced.

[0094] Please see Figure 2 The multi-source data fusion solution mainly includes four modules: data preprocessing module, data association module, data fusion module, and result output module.

[0095] The data preprocessing module cleans, aligns, and normalizes the raw data to prepare it for fusion. Key operations include: time alignment and resolution of spatial heterogeneity, handling of missing values, and noise filtering.

[0096] Please see Figure 3 Time alignment and resolving spatial heterogeneity include the following steps: The "Data Time" field for all devices is converted to a UTC millisecond-level timestamp; Define a global time window (e.g., 200ms), group the data by timestamp, and divide it into time slices (e.g., 200ms window). Data falling into the same window are considered to be observed at the same time. For example, maintain a time slot buffer for each drone entity, and data within the window are considered to occur simultaneously. Handling latency: Use a sliding window mechanism to allow early or late data (e.g., maximum tolerable latency of 400ms), discard timed-out data or use it for prediction; Latitude and longitude coordinates are uniformly converted to UTM plane coordinates (unit: meters); Establish a spatial grid index (50m×50m grid) to accelerate nearest neighbor search.

[0097] Missing value handling includes missing field handling and data normalization. For missing fields, different strategies are adopted based on device type and data availability: Vertical flight speed: Only provided by RID equipment. If other equipment is missing, RID data is used directly if it is available within the time window; otherwise, it is estimated from the rate of change of altitude (using historical data of true altitude or altitude, calculated by difference). Track angle: Missing in 5G-A, but provided by other devices. If missing, use linear extrapolation if the associated entity has historical track angles; otherwise, leave it empty (do not merge). True altitude / altitude: If AOA equipment is missing, it will be fused from other equipment (5G-A, RID, passive spectrum detection); if all are missing, the default value (such as ground altitude 0) will be used or ignored. Horizontal flight speed: AOA equipment is missing. When missing, it is estimated from other equipment or from the rate of change of position (continuous position difference). Product Model: Available only for passive spectrum detection and AOA devices. If missing (e.g., 5G-A, RID), leave it blank or inherit it from the associated history.

[0098] Data normalization: such as converting latitude and longitude to a unified coordinate system (e.g., WGS84), unifying the unit of altitude to meters, and unifying the unit of speed to m / s.

[0099] Noise filtering includes the following steps: Simple filtering: Use moving average or median filtering to remove outliers from continuous data (such as location, speed); Confidence weight: An initial weight is assigned to each device (based on device type, such as RID devices are more reliable, weight 0.9; passive spectrum 0.8; AOA monitoring 0.6; 5G-A monitoring 0.5), and the weight can be adjusted in real time (based on historical error).

[0100] The data association module binds data from different devices to the same drone entity, resolving the question of "which data belongs to the same drone". The serial number is the primary key, but inconsistencies (such as serial number conflicts or missing numbers) need to be handled. The specific association strategy is as follows: First-level association (serial number priority): If multiple devices report the same serial number and the time difference is within a threshold (e.g., ≤200ms), they are directly associated. When the location difference of the same serial number is large (e.g., >100 meters), it is considered an error and a secondary association is triggered.

[0101] Second-level association (spatiotemporal matching): When the serial number is missing or inconsistent, location and time are used for association.

[0102] Spatial association: Calculate Euclidean distance (based on latitude and longitude). If the distance is within a threshold (e.g., ≤50 meters, the threshold can be dynamically adjusted according to speed), it is considered as the same entity. Time correlation: The time difference is within the window (e.g., ≤200ms); Motion assistance: If velocity information is available, use motion consistency (such as track angle similarity) to assist in the association; Multiple Hypothesis Tracking (MHT): For high-density scenes, a lightweight MHT algorithm is used to generate multiple related hypotheses, and the optimal one is selected based on probability (such as position and velocity likelihood).

[0103] Third-level association (short-term trajectory similarity): Please see Figure 4 When single-point matching is ambiguous (e.g., multiple drones are nearby), a trajectory comparison processing method is introduced. The trajectory similarity is calculated as follows: Position sequence: Calculate the trajectory shape distance using DTW (Dynamic Time Warping); Motion characteristics: Cosine similarity of velocity / heading; Overall rating: ; Furthermore, if data cannot be associated with an existing entity, a new drone track is created (initialized based on serial number or location). In entity management, each drone entity maintains a state vector (including position, velocity, etc.) and a lifecycle (deleted if no data is available after timeout).

[0104] The data fusion module performs fusion calculations on the correlated data to predict the complete flight state of the UAV. Its core is a multi-sensor fusion algorithm, which designs fusion rules for different state variables. The state vector is defined as: [Serial Number, Latitude and Longitude, True Altitude, Altitude, Horizontal Flight Speed, Vertical Flight Speed, Track Angle, Product Model, Data Time] The specific fusion strategy is as follows: Latitude and longitude (provided by all devices): Method: Weighted average fusion, with weights based on device confidence (initial weights: RID=0.9, others=0.8) and signal quality (e.g., signal-to-noise ratio); Example: Merged longitude = (w1 * longitude 1 + w2 * longitude 2 + ...) / Σw, latitude is calculated similarly; True altitude and altitude (AOA equipment missing): Method: Complementary fusion, true altitude and altitude can be converted to each other (requires digital elevation model DEM), but independent fusion is performed in simple scenarios. True altitude fusion uses a weighted average of data from 5G-A, RID, and passive spectrum probes. If AOA is missing, it is not included. Altitude fusion is similar to true altitude. If all data is missing, DEM is used for estimation or it is ignored. Horizontal flight speed (AOA equipment missing): Method: Kalman filter fusion, with RID data as the baseline (most complete), and other devices used as measurement updates; if RID is missing, a weighted average of 5G-A or passive spectrum probes is used; if AOA is missing, it is not included. Vertical flight speed (provided by RID devices only): Method: RID data is prioritized; if RID is missing, the change rate of true altitude / altitude is estimated (e.g., Historical data is used for differential analysis; other devices are missing and not integrated. Track angle (missing for 5G-A, but provided by other devices): Methods: Weighted average or vector synthesis (averaging after converting angles to unit vectors); 5G-A is missing and not included; if the angle difference is large (e.g., >30°), priority: passive spectrum data > AOA > RID; Product models (available only for passive spectrum detection and AOA devices): Method: Voting or weighted priority. If multiple sources are consistent, the one that is adopted directly is adopted. In case of conflict, the priority is passive spectrum > AOA device (subsequent access of other devices can be expanded).

[0105] After the serial numbers are associated, they are uniformly adopted (prioritizing RID, and sorting by priority if there is no RID, RID > passive > AOA > 5g-a), and the data time is the latest timestamp after fusion.

[0106] The core algorithm is as follows: Kalman Filter (KF): Used for dynamic state (position, velocity) estimation. The KF state vector includes position and velocity. Measurement inputs come from various devices. It handles noise and missing data, has good real-time performance, can process time series data, and can be extended to multiple model KF (such as considering UAV maneuver models). Weighted average: used for static or low-dynamic states (such as height), simple and efficient; Uncertainty management: Output a confidence score (0-1) for each fusion state, based on sensor weights and data consistency.

[0107] The results output module generates the fused UAV flight status data, and its output includes: Required fields: Serial number, latitude and longitude, true altitude, horizontal flight speed, data time (contributed by all devices); Optional fields: Vertical flight speed (dependent on RID), track angle (dependent on multiple devices), product model (dependent on specific devices); Additional information: Data source combination (e.g., "RID+5G-A").

[0108] Note that the serial number and product model of the entire fused trajectory remain unchanged. Once the source is determined, these two values ​​are fixed for each trajectory point; any missing values ​​need to be filled in.

[0109] The output format is as follows: Real-time streaming: JSON messages, updated once per second (example): { "sn": "DRN123", / / Drone serial number "latitude": 39.9042, / / latitude "longitude": 116.4074, / / longitude "height": 120.5, / / Really tall "altitude": 150.2, / / altitude "GS": 15.3, / / Horizontal flight speed "VS": 0.5, / / Vertical flight speed, from RID or estimated "heading_angle": 45.0, / / "model": "DJI Mavic", / / Product model number, derived from passive spectrum detection or AOA. "timestamp": "2025-09-01T12:00:00Z", / / current time "sources": ["RID", "5G-A", "Spectrum"] / / Data source combination } The aforementioned technical solution addresses the challenges of integrating with full-scale UAV data and the inability to monitor "non-cooperative targets." It requires only integration with a few detection devices to achieve full-airspace UAV monitoring. It also overcomes the high deployment costs, vulnerabilities, limitations, blind spots, and insufficient reliability and accuracy associated with deploying single detection devices. Through multi-source heterogeneous data collaborative sensing, cross-validation, and complementary enhancement, this solution achieves the goal of "comprehensive, accurate, clear, and trustworthy" UAV monitoring, constructing a robust and reliable low-altitude safety monitoring system.

[0110] By adopting the above technical solutions, this invention differs from existing technologies and possesses the following beneficial effects: By integrating multi-source sensing technologies such as radar, radio, optoelectronics, and 5G-A, a holographic sensing network for low-altitude airspace is constructed, enabling real-time, all-domain dynamic monitoring of low-altitude aircraft; radio and motion characteristics captured by multiple sensors are intelligently analyzed to synthesize a unique and reliable output result, thereby improving the accuracy and efficiency of the monitoring system and accurately identifying potential threats from legal and illegal flights. Using the above methods and systems, the needs of urban low-altitude safety supervision can be met, achieving high-precision tracking, identification, and trajectory prediction of legal / illegal drones at all times, effectively improving low-altitude threat early warning capabilities and airspace collaborative management efficiency.

[0111] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0112] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods of various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0113] The above description is only a part of the embodiments of the present invention and does not limit the scope of protection of the present invention. Any equivalent device or equivalent process transformation made based on the content of the present invention specification and drawings, or direct or indirect application in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for managing unmanned aerial vehicles (UAVs) by fusing multi-source data, characterized in that, include: The system acquires raw monitoring data from various drone detection devices, including positioning data collected by 5G-A sensing devices, identity data acquired by RID devices, radio frequency data captured by passive spectrum detection devices, and azimuth data measured by AOA monitoring devices. The original monitoring data is subjected to spatiotemporal standardization processing to generate standardized monitoring data with a unified spatiotemporal reference. The standardized monitoring data is used for target association matching to generate the association mapping relationship of UAVs; Multi-source data fusion is performed on the standardized monitoring data after correlation mapping to generate comprehensive UAV status data; Based on the comprehensive status data of the UAV, airspace supervision analysis is performed to generate an airspace supervision decision report, which includes monitoring information, authentication status, and early warning prompts. The target association matching includes direct association based on identifier consistency and indirect association based on spatiotemporal similarity. The dynamic time warping algorithm is used to calculate trajectory similarity, and the correspondence between multi-source data and UAV targets is established by combining motion feature matching. Target association matching is performed on the standardized monitoring data to generate association mapping relationships for UAVs, including: The standardized monitoring data is received, and a first-level association matching is performed based on identifier consistency. When the identifiers reported by multiple detection devices are consistent and the spatiotemporal conditions meet the preset threshold, a direct association relationship is established and a preliminary association result is generated. When the identifier is unavailable or there is a conflict, the spatiotemporal similarity association algorithm is activated to perform secondary association matching. By calculating the spatial proximity and temporal synchronization of the target location, association hypotheses are established and a candidate association set is generated. The dynamic time warping algorithm is used to calculate the trajectory morphology similarity of the candidate association set. Combined with the consistency analysis of motion feature parameters, the preliminary association results are verified and optimized to generate the optimized association mapping relationship. For ambiguous association mappings, a multi-hypothesis tracking mechanism is established, multiple association schemes are evaluated based on the principle of probabilistic optimality, and the association result with the highest confidence is selected as the final association mapping. Based on the final association mapping relationship, establish a correspondence table between UAV targets and multi-source monitoring data, and output the final association mapping relationship.

2. The UAV management method based on multi-source data fusion according to claim 1, characterized in that, The spatiotemporal standardization process includes establishing a global time reference system to align timestamps from multiple sources, using a sliding window mechanism to achieve time synchronization, and using a spatial indexing algorithm to achieve unified transformation of geographic coordinates. The original monitoring data is subjected to spatiotemporal standardization processing to generate standardized monitoring data with a unified spatiotemporal reference, including: Receive the raw monitoring data, establish a global time reference system, and uniformly convert the timestamps of monitoring data from different detection devices to the standard time coordinate system; The original monitoring data after time conversion is processed by a sliding time window mechanism to generate time-aligned first intermediate data. A spatial grid index is constructed on the first intermediate data, and the geographic coordinates collected by different UAV detection devices are uniformly converted into a standard plane coordinate system to generate a spatially unified second intermediate data; The second intermediate data is subjected to data integrity verification. For missing data fields, a data completion strategy based on device characteristics is used to complete the data and output the standardized monitoring data.

3. The UAV management method based on multi-source data fusion according to claim 2, characterized in that, The second intermediate data undergoes data integrity verification. For missing data fields, a data completion strategy based on device characteristics is used for completion processing, including: Identify missing fields in the second intermediate data; When the missing field is the vertical flight speed field, the vertical flight speed data for the corresponding time window is obtained from the historical data of the RID device. If the RID device data is unavailable, the missing field is estimated by differential calculation based on the historical rate of change of true altitude or altitude. When the missing field is the track angle field, the associated entity obtains historical track angle data. If historical track angle data exists, the missing field is filled in using a linear extrapolation algorithm. If there is no historical track angle data, the missing field is left as null. When the missing field is true height or altitude, the available data of 5G-A sensing devices, RID devices and passive spectrum detection devices are weighted and fused based on the device confidence weight to fill in the missing field. When the missing field is the horizontal flight speed field, the available equipment data is weighted and fused based on equipment reliability, or the missing field is estimated based on the rate of change of position using a continuous position difference algorithm. When the missing field is the product model field, the product model information is inherited from the available data of the passive spectrum detection device. If the passive spectrum detection device data is unavailable, the corresponding field information is inherited from the available data of the AOA monitoring device to complete the missing field.

4. The UAV management method based on multi-source data fusion according to claim 1, characterized in that, For ambiguous association mappings, a multi-hypothesis tracking mechanism is established. Multiple association schemes are evaluated based on the principle of probabilistic optimality, and the association result with the highest confidence is selected as the final association mapping. This includes: Multiple association hypothesis schemes are generated for the ambiguous association mapping relationship. Each association hypothesis scheme contains different target-data correspondences, including: The location matching probability of each associated hypothesis scheme is calculated based on the location likelihood function, which is constructed according to the spatial proximity of the target location. The motion matching probability of each associated hypothesis scheme is calculated based on the consistency of motion features, wherein the motion features include consistency measures of velocity vector and heading angle; By combining the location matching probability and the motion matching probability, a comprehensive confidence score is calculated for each association hypothesis scheme; The association hypothesis with the highest overall confidence score is selected as the optimal association hypothesis, and the final association mapping relationship is generated.

5. The UAV management method based on multi-source data fusion according to claim 1, characterized in that, The multi-source data fusion includes fusing location information using an adaptive weighting algorithm, estimating motion state using a Kalman filter algorithm, and integrating identity features using a confidence voting mechanism to form a complete state description that includes spatial situation, motion parameters, and identity identifiers. Multi-source data fusion is performed on the standardized monitoring data after correlation mapping to generate comprehensive UAV status data, including: The standardized monitoring data after the association mapping is received, and the location information from different UAV detection devices is fused and calculated using an adaptive weighting algorithm to generate fused spatial location data. The Kalman filter algorithm is used to estimate the motion state of the fused spatial location data, generating motion state data that includes velocity and acceleration. Based on a confidence-based voting mechanism, identity feature data from different UAV detection devices are integrated to generate unified identity identification data; The fused spatial location data, motion state data, and unified identity data are integrated to form a complete state description that includes spatial situation, motion parameters, and identity, and the comprehensive state data of the UAV is output.

6. The UAV management method based on multi-source data fusion according to claim 5, characterized in that, The standardized monitoring data after the association mapping is received, and the location information from different UAV detection devices is fused and calculated using an adaptive weighting algorithm to generate fused spatial location data, including: Obtain the location information of each UAV detection device in the standardized monitoring data after the association mapping, and determine the initial confidence weight of each device based on the device type; The initial confidence weight is dynamically adjusted based on the real-time signal quality parameters of each device's location information to generate dynamic weight coefficients for each device. Based on the dynamic weighting coefficients, the location information of each device is weighted and averaged to generate weighted and fused latitude and longitude coordinates. The weighted and fused latitude and longitude coordinates are subjected to inverse coordinate transformation to convert the plane coordinate system coordinates into geographic coordinate system coordinates, thereby generating the fused spatial location data. The fused spatial location data is subjected to outlier detection and filtering to output the final spatial location data.

7. The UAV management method based on multi-source data fusion according to claim 5, characterized in that, The fused spatial location data, motion state data, and unified identity data are integrated to form a complete state description that includes spatial situation, motion parameters, and identity identifiers, including: Receive fused spatial location data, motion status data, and unified identity data to establish a UAV status description framework; Map the latitude and longitude coordinates, true altitude, and altitude information in the spatial location data to the spatial situation field of the state description framework; Map the horizontal flight speed, vertical flight speed, and track angle information in the motion state data to the motion parameter fields of the state description framework; Map the serial number and product model information in the unified identity data to the identity field of the status description framework; Perform data consistency verification on the mapped state description framework to ensure that the logical relationships between the fields conform to the flight characteristics of the UAV. Based on the verification results, a complete state description including spatial situation, motion parameters and identity identifiers is generated, and the comprehensive state data of the UAV is output.

8. The UAV management method based on multi-source data fusion according to claim 1, characterized in that, The airspace regulatory analysis includes inferring flight intentions through trajectory prediction algorithms, assessing compliance in conjunction with an airspace rule base, and determining threat levels based on a risk assessment model; Based on the comprehensive status data of the aforementioned UAVs, airspace regulatory analysis is performed to generate an airspace regulatory decision report, including: Receive the comprehensive status data of the UAV, and use a trajectory prediction algorithm to make short-term and medium-term predictions of the UAV's flight trajectory to generate flight intention analysis results; The flight intent analysis results are matched and compared with the no-fly zones, restricted flight zones, and special control areas in the airspace rule base to generate flight compliance assessment results; Based on the comprehensive status data and flight intention analysis results of the UAV, a threat level score is calculated through a risk assessment model. The threat level score comprehensively considers flight altitude anomaly, speed anomaly, heading deviation, and airspace sensitivity. Based on the flight compliance assessment results and threat level scores, early warning level information is generated, and an early warning prompt is generated when illegal or high-risk flights are detected. By integrating the flight intent analysis results, flight compliance assessment results, threat level scores, and warning level information, an airspace regulatory decision report is generated that includes monitoring information, certification status, and warning prompts.

9. A drone management system for multi-source data fusion of unmanned aerial vehicles (UAVs), characterized in that, The system for implementing the method according to any one of claims 1 to 8, the system comprising: The data acquisition module is used to acquire raw monitoring data from various UAV detection devices. The raw monitoring data includes positioning data collected by 5G-A sensing devices, identity data acquired by RID devices, radio frequency data captured by passive spectrum detection devices, and azimuth data measured by AOA monitoring devices. The spatiotemporal standardization processing module is used to perform spatiotemporal standardization processing on the original monitoring data to generate standardized monitoring data with a unified spatiotemporal benchmark. The target association matching module is used to perform target association matching on the standardized monitoring data to generate the association mapping relationship of the UAV; The multi-source data fusion module is used to fuse standardized monitoring data after correlation mapping to generate comprehensive UAV status data; The airspace supervision and analysis module is used to perform airspace supervision and analysis based on the comprehensive status data of the UAV and generate an airspace supervision and decision report, which includes monitoring information, authentication status and early warning prompts.

Images