Low-altitude trajectory fusion method and device

Abstract

The embodiment of the application provides a low-altitude track fusion method and device, the method comprises the following steps: realizing unified access of multi-source data through a protocol adaptation gateway; performing space-time alignment, abnormal filtering and cleaning on the data; classifying track points into two categories of analysis and non-analysis according to whether the track points carry a unique code of a unmanned aerial vehicle, and calculating the confidence thereof; adopting a direct matching and weighted point supplement fusion strategy for the analysis track, and adopting a combination strategy of DBSCAN clustering, nearest neighbor association and Kalman filter updating for track generation and association updating of the non-analysis track; and finally outputting accurate unified track data, so that the precision and reliability of low-altitude track reconstruction and fusion can be improved.

Description

Technical Field

[0001] This application relates to the field of data processing, specifically to a method and apparatus for low-altitude trajectory fusion. Background Technology

[0002] Currently, with the widespread application of drone technology, their flight activities in low-altitude airspace are becoming increasingly frequent, placing higher demands on safety supervision and trajectory monitoring. Existing technologies mainly rely on single sensors (such as analytical devices, radar, spectrum monitoring, or 5G-A) to collect and present drone trajectories, or perform preliminary integration through simple data overlay. However, these methods have significant shortcomings in multi-source heterogeneous data fusion: they lack a systematic fusion framework, making it difficult to achieve complementary correlation between analytical and non-analyzed trajectories; due to the lack of an effective spatiotemporal collaborative matching mechanism, trajectory confusion and identity binding errors are prone to occur in multi-target scenarios; at the same time, due to factors such as sensor detection gaps and obstructions, trajectories are prone to breakage, failing to form a continuous and complete flight trajectory chain. In addition, existing solutions often fail to effectively calibrate the measurement errors and format differences of multi-source data, resulting in low positioning accuracy and confidence of the fused trajectory, and weak system adaptability, making it difficult to adapt to the dynamic needs of different monitoring scenarios.

[0003] Therefore, there is an urgent need for a low-altitude trajectory fusion method that can improve the accuracy and reliability of low-altitude trajectory reconstruction and fusion. Summary of the Invention

[0004] To address the problems in the prior art, this application provides a low-altitude trajectory fusion method and apparatus, which can improve the accuracy and reliability of low-altitude trajectory reconstruction and fusion.

[0005] To solve at least one of the above problems, this application provides the following technical solution: Firstly, this application provides a low-altitude trajectory fusion method, including: The system receives multi-source trajectory data from a UAV, cleans and performs spatiotemporal alignment on the multi-source trajectory data, determines the corresponding standardized trajectory data, extracts trajectory points from the standardized trajectory data, classifies the trajectory points into parsed trajectory points and non-parsed trajectory points based on whether the trajectory points carry a unique UAV code, and evaluates the confidence level of each trajectory point according to a set confidence parameter. For the parsed trajectory points, each parsed trajectory point is associated with the corresponding parsed trajectory according to the preset trajectory matching engine. If there is a blank parsed trajectory, multiple non-parsed trajectory points around the parsed trajectory that meet the preset conditions are associated according to the nearest neighbor method. The parsed trajectory is then filled and fused according to the non-parsed trajectory points to determine the corresponding fused parsed trajectory. For the non-analytical trajectory points, a preset clustering algorithm is used to cluster each non-analytical trajectory point. Non-analytical trajectories are created based on the effective clusters after clustering. The non-analytical trajectories are then merged and updated based on newly received non-analytical trajectory points to determine the corresponding merged non-analytical trajectory. After fusing the obtained analytical and non-analytical trajectories, the corresponding unified trajectory data is output. The unified trajectory data includes trajectory confidence, which is calculated based on the confidence of each trajectory point.

[0006] Furthermore, the process of receiving multi-source trajectory data from the UAV, cleaning and spatiotemporally aligning the multi-source trajectory data, and determining the corresponding standardized trajectory data includes: The system receives multi-source trajectory data from the UAV and performs basic rule verification. The multi-source trajectory data includes data from the parsing device, spectrum monitoring data, and radar data. For the data that passes the basic rule verification, the mean and standard deviation of each data field are calculated using a sliding window algorithm. Data that deviates from the preset threshold is marked as abnormal noise. Abnormal noise data is then removed and adjacent data is marked to determine the corresponding cleaned data. The cleaned data is subjected to a coordinate system operation, and the data after the coordinate system operation is time-aligned by time zone offset correction and interpolation to determine the corresponding standardized trajectory data.

[0007] Further, the step of evaluating the confidence level of each trajectory point according to the set confidence level parameter includes: Set confidence parameters, including base confidence, dimension correction factor, and error deduction term; The basic confidence level is set according to the sensor type of the collected trajectory data, with the basic confidence level decreasing sequentially for analysis equipment, radar equipment, and spectrum monitoring equipment. The dimension correction coefficients include data integrity correction coefficients and preprocessing quality correction coefficients, which are used to perform positive gain corrections on data integrity and preprocessing quality. The error deduction items include position deviation deduction items, time synchronization deviation deduction items, and data anomaly deduction items, which are used to deduct data deviations and data anomalies in reverse. A confidence quantification model is constructed based on the confidence parameters, and the confidence of each trajectory point is evaluated using the confidence quantification model to determine the confidence of each trajectory point.

[0008] Furthermore, if a blank in the analytical trajectory occurs, multiple non-analytical trajectory points around the analytical trajectory that meet preset conditions are associated according to the nearest neighbor method. The analytical trajectory is then filled and fused based on these non-analytical trajectory points to determine the corresponding fused analytical trajectory, including: When a detection gap appears in the parsed trajectory, multiple non-parsed trajectory points around the parsed trajectory that meet the preset filtering conditions are selected according to the nearest neighbor method and stored in the parsed trajectory's dedicated cache pool. The filtering conditions include time difference conditions and spatial distance conditions. When the number of non-analyzed trajectory points in the dedicated cache pool of the parsed trajectory reaches a preset number, the non-analyzed trajectory points are fused and calculated according to the weighted average algorithm to determine the corresponding supplementary trajectory points. The weights of the weighted average algorithm are allocated according to the confidence level of each non-analyzed trajectory point. The supplementary trajectory points are added to the analytical trajectory as supplementary points for the detected blank areas. At the same time, the points are marked as supplementary points, and the overall confidence of the analytical trajectory after adding supplementary points is reduced accordingly.

[0009] Further, the step of clustering each of the non-analytical trajectory points using a preset clustering algorithm, and creating a non-analytical trajectory based on the effective clusters after clustering, includes: Each of the non-analyzed trajectory points is stored in the clustering cache pool and sorted according to timestamp; The DBSCAN clustering algorithm is used to perform clustering operations on the non-parse trajectory points in the cache pool to be clustered, and the corresponding effective clusters are determined. If the effective cluster is not associated with any existing trajectory data, a unique trajectory identifier is assigned to the effective cluster, and the non-analyzable trajectory points in the effective cluster are sorted by time to determine the corresponding non-analyzable trajectory. At the same time, the non-analyzable trajectory is marked as active.

[0010] Further, the step of fusing and updating the non-analyzable trajectory based on the newly received non-analyzable trajectory points to determine the corresponding fused non-analyzable trajectory includes: The newly received non-analyzable trajectory points are matched with all currently marked as active non-analyzable trajectories using the nearest neighbor method; If a match is successful, the non-analyzed trajectory point is stored in the corresponding non-analyzed trajectory dedicated cache pool. When the number of newly added non-analyzed trajectory points in the non-analyzed trajectory dedicated cache pool reaches the preset number of points, the trajectory points in the cache pool are fused according to the Kalman filter algorithm to determine the corresponding fused non-analyzed trajectory and update the confidence of the non-analyzed trajectory. If the match fails, the historical trajectory backtracking match is performed, and the newly received non-parsed trajectory points are matched with all terminated non-parsed trajectories in the historical trajectory database. If a matching historical trajectory exists, the status of the historical trajectory is marked as active. The non-analyzable trajectory point and the historical trajectory are fused together to determine the corresponding fused non-analyzable trajectory, and the confidence of the non-analyzable trajectory is updated. If no matching historical trajectory exists, the unparsed trajectory point is stored in the clustering cache pool for later re-execution of the clustering process.

[0011] Furthermore, the process of fusing the obtained analytical and non-analytical trajectories to output corresponding unified trajectory data includes: The obtained parsed and non-parsed trajectories are fused in the data fusion layer. The fused data includes the unique identifier of the trajectory, the latest trajectory point, the trajectory type, the real-time status, the current trajectory confidence level, and the list of associated devices. The unified trajectory data obtained after fusion will be output, including trajectory data and statistical indicators derived from the trajectory data.

[0012] Secondly, this application provides a low-altitude trajectory fusion device, comprising: The multi-source trajectory data processing module is used to receive multi-source trajectory data from UAVs, clean and perform spatiotemporal alignment processing on the multi-source trajectory data, determine the corresponding standardized trajectory data, extract trajectory points from the standardized trajectory data, classify the trajectory points into parsed trajectory points and non-parsed trajectory points according to whether the trajectory points carry the UAV's unique code, and evaluate the confidence of each trajectory point according to the set confidence parameters. The analytical trajectory fusion module is used to associate each analytical trajectory point with the corresponding analytical trajectory according to a preset trajectory matching engine. If there is a blank analytical trajectory, it associates multiple non-analytical trajectory points around the analytical trajectory that meet the preset conditions according to the nearest neighbor method, and performs point filling and fusion on the analytical trajectory according to the non-analytical trajectory points to determine the corresponding fused analytical trajectory. The non-analytical trajectory fusion module is used to cluster the non-analytical trajectory points using a preset clustering algorithm, create non-analytical trajectories based on the effective clusters after clustering, and fuse and update the non-analytical trajectories based on newly received non-analytical trajectory points to determine the corresponding fused non-analytical trajectory. The unified trajectory determination module is used to fuse the obtained analytical trajectory and non-analytical trajectory and output the corresponding unified trajectory data. The unified trajectory data includes trajectory confidence, which is calculated based on the confidence of each trajectory point.

[0013] Thirdly, this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the low-altitude trajectory fusion method.

[0014] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the low-altitude trajectory fusion method described above.

[0015] Fifthly, this application provides a computer program product, including a computer program / instructions that, when executed by a processor, implement the steps of the low-altitude trajectory fusion method described above.

[0016] As can be seen from the above technical solution, this application provides a low-altitude trajectory fusion method and apparatus, which realizes unified access of multi-source data through a protocol adaptation gateway; performs spatiotemporal alignment, anomaly filtering and cleaning on the data; classifies trajectory points into two categories, analytical and non-analytical, based on whether they carry a unique UAV code, and calculates their confidence level; adopts a direct matching and weighted point supplementation fusion strategy for analytical trajectories, and adopts a combination strategy of DBSCAN clustering, nearest neighbor association and Kalman filter update for non-analytical trajectories to generate trajectory and update association; finally outputs accurate unified trajectory data, thereby improving the accuracy and reliability of low-altitude trajectory reconstruction and fusion. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is one of the flowcharts illustrating the low-altitude trajectory fusion method in the embodiments of this application; Figure 2 This is a structural diagram of the low-altitude trajectory fusion device in the embodiments of this application; Figure 3 This is a schematic diagram of the structure of the electronic device in the embodiments of this application; Figure 4 This is the second flowchart illustrating the low-altitude trajectory fusion method in the embodiments of this application.

[0019] Figure label: Electronic device 9600, central processing unit 9100, memory 9140, communication module 9110, input unit 9120, audio processor 9130, display 9160, power supply 9170, buffer memory 9141, application / function storage unit 9142, data storage unit 9143, driver storage unit 9144, antenna 9111, speaker 9131, microphone 9132. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0021] The acquisition, storage, use, and processing of data in this application comply with relevant laws and regulations.

[0022] Considering the lack of a systematic fusion framework and an effective spatiotemporal matching mechanism for UAV multi-source trajectory data fusion, this application provides a low-altitude trajectory fusion method and apparatus. It achieves unified access to multi-source data through a protocol-adaptive gateway; performs spatiotemporal alignment, anomaly filtering, and cleaning on the data; classifies trajectory points into analytical and non-analytical categories based on whether they carry a unique UAV code, and calculates their confidence levels; employs a direct matching and weighted point-filling fusion strategy for analytical trajectories, and a combined strategy of DBSCAN clustering, nearest neighbor association, and Kalman filter update for non-analytical trajectories to generate and update trajectories; finally, it outputs accurate unified trajectory data, thereby improving the accuracy and reliability of low-altitude trajectory reconstruction and fusion.

[0023] For the complete process of this solution, please refer to [link / reference]. Figure 4 This includes constructing a six-level hierarchical fusion system consisting of "data acquisition - data preprocessing - dynamic parameter configuration - trajectory classification - core fusion - trajectory output". It integrates trajectory data from multiple sources such as analysis equipment, spectrum monitoring, radar, and 5G-A. By leveraging standardized parameter configuration, combined algorithm applications, and differentiated fusion rules, it achieves complementary correlation between non-analyzed and analyzed trajectories, as well as correlation and merging between non-analyzed trajectories, ultimately generating high-precision, continuous, and unified UAV trajectories. This solves the problems of poor fusion effect and incomplete trajectory in existing technologies.

[0024] First, the structure of the system used to implement this method is described below: Data acquisition layer: Accesses and analyzes data from various sensors such as equipment, spectrum monitoring, 5G-A, and radar.

[0025] Data preprocessing layer: unified coordinate system (WGS84), time synchronization (UTC), and abnormal data filtering.

[0026] Dynamic parameter configuration layer: presets various fusion parameters and dynamically adjusts parameters during the process.

[0027] Trajectory point classification layer: The trajectory points are divided into two categories: analytical and non-analytical.

[0028] Trajectory fusion core layer: executes matching, clustering, association, and fusion algorithms.

[0029] Trajectory matching engine: responsible for the association and matching of trajectory points with existing trajectories, the core of which is to execute the nearest neighbor method matching logic to provide the association basis for fusion; Cluster Processor: Focuses on cluster analysis of non-analytical trajectory points, generating effective clusters through the DBSCAN clustering algorithm to filter false alarm data; Cache Pool Manager: Creates dedicated cache pools for parsed trajectories, unparsed trajectories, and trajectory points to be clustered, storing data to be merged and supporting temporary data storage and fast retrieval; Fusion and Point Filling Processor: For the detection gaps in the analytical trajectory, it performs point filling and fusion operations on non-analytical trajectory points to fill the trajectory gaps; Trajectory generator: Based on the fused trajectory points, it constructs a complete trajectory model, calculates trajectory confidence, and updates trajectory status; Scheduled task scheduler: Schedules various fusion tasks to be executed according to preset time parameters (such as data fallback time window) to ensure the timing and stability of the fusion process.

[0030] Trajectory Output Layer: Outputs a unified trajectory, supporting real-time push and historical query.

[0031] To improve the accuracy and reliability of low-altitude trajectory reconstruction and fusion, this application provides an embodiment of a low-altitude trajectory fusion method, see [link to embodiment]. Figure 1 The low-altitude trajectory fusion method specifically includes the following: Step S101: Receive multi-source trajectory data of UAV, clean and spatiotemporally align the multi-source trajectory data, determine the corresponding standardized trajectory data, extract trajectory points from the standardized trajectory data, classify the trajectory points into parsed trajectory points and non-parsed trajectory points according to whether the trajectory points carry the UAV's unique code, and evaluate the confidence of each trajectory point according to the set confidence parameter. Optionally, in this embodiment, multi-source trajectory data is first acquired through a data acquisition layer.

[0032] The data acquisition layer, as the fundamental support for multi-source trajectory fusion, aims to achieve unified access, complete aggregation, and efficient forwarding of raw data from analysis equipment, spectrum monitoring equipment, 5G-A equipment, radar, and other extended sensors. This provides comprehensive and high-quality data source support for subsequent preprocessing and fusion processes. The specific implementation is as follows: To address the differences in communication protocols among various sensors, a design combining a "protocol adaptation gateway and standardized conversion interface" is adopted. This supports mainstream communication protocols such as TCP / IP, UDP, HTTP, and MQTT, while also being compatible with various device-defined private protocols. The protocol adaptation gateway parses and encapsulates the raw data from different protocols, converting heterogeneous protocol data into a unified transmission format within the system. This ensures seamless access for devices such as analysis equipment, spectrum monitoring equipment, 5G-A sensors, and radar, preventing data loss or access failures due to protocol incompatibility.

[0033] Optionally, in this embodiment, the collected data is then preprocessed.

[0034] As a crucial pre-processing step in multi-source trajectory fusion, the core objective of the data preprocessing module is to eliminate noise interference and fill data gaps in multi-source sensor data through two core operations: data cleaning and spatiotemporal alignment. This achieves standardization and unification of multi-source data, including analysis equipment, spectrum monitoring, 5G-A, and radar, in both temporal and spatial dimensions, providing high-quality and highly consistent data source support for subsequent trajectory classification and core fusion. The specific technical solution is as follows: Based on sensor type, preset basic verification rules are used: For parsed device data, the integrity of the SN encoding format (conforming to preset unique encoding rules) and the rationality of the position coordinates (falling within the preset monitoring area) are verified; for radar data, the speed value range (0-200 m / s, adapting to common UAV flight speeds) and the validity of the RCS parameter (non-negative value) are verified; for spectrum monitoring data, the signal frequency range (conforming to the commonly used UAV communication frequency bands 2.4 GHz and 5.8 GHz) and the signal strength stability (fluctuation amplitude does not exceed the preset threshold) are verified.

[0035] Dynamic threshold calculation and application: The mean and standard deviation of each data field are calculated using a sliding window algorithm (window size set to 5 consecutive data points). Data that deviates from the mean by ±3 times the standard deviation are marked as abnormal noise. At the same time, combined with the historical data accuracy model of the equipment, differentiated thresholds are set for sensors with different accuracy levels (e.g., the radar data position error threshold is set to 5 meters, and the 5G-A data position error threshold is set to 3 meters) to ensure that the filtering rules are adapted to the characteristics of the equipment.

[0036] Noise filtering execution: For the marked noisy data, the strategy of "direct removal + adjacent data marking" is adopted. This not only removes invalid noise points directly, but also marks one data point before and after the noise point as "suspicious data". In the subsequent fusion process, the weight ratio of these data points is reduced to avoid a single noise point affecting the overall data quality.

[0037] Coordinate system transformation: The WGS84 coordinate system is adopted as the standard coordinate system of the system, covering three dimensions: longitude (range -180~180°), latitude (range -90~90°), and altitude (unit: meters, based on the geoid). A standardized transformation model is established for the original coordinate systems of different sensors.

[0038] Time synchronization calibration: UTC time is uniformly adopted as the system standard time, accurate to the millisecond level (timestamp format is 13-bit number). The raw time reported by each sensor is converted to UTC time through time zone offset correction. At the same time, for high-frequency acquisition equipment (such as radar, acquisition frequency ≥10Hz), interpolation method is used to correct the timestamp dispersion problem, ensuring that the time interval between adjacent data points is uniform (error ≤10ms).

[0039] Optionally, in this embodiment, the trajectory points in the preprocessed trajectory data are classified and confidence scores are calculated.

[0040] The trajectory classification module, as a key intermediate link in multi-source trajectory fusion, aims to accurately classify trajectory points and trajectories based on the identification characteristics of multi-source sensor data and standardized information after data preprocessing. This provides a classification basis for subsequent differentiated fusion strategies for analyzed and non-analyzed trajectories, ensuring the targeting and efficiency of the fusion process. The specific technical solution is as follows: Track point analysis: It can directly identify and verify track points through the drone's unique serial number (SN). The data mainly comes from the analysis equipment and has 100% identity accuracy.

[0041] Unresolved trajectory points: Trajectory points whose unique serial number (SN) cannot be directly identified. The data comes from spectrum monitoring equipment, radar, 5G-A equipment and other extended sensors, and only contains device-defined codes (UAV ID) or no identification information.

[0042] Calculate the confidence level of the trajectory points: Score = Base confidence level S0 × (1 + Σ Dimensional correction coefficient C) n - Σ Error deduction item D k The calculation and value rules for each parameter are as follows: 1. Basic credibility S0 Analysis device: S0=1 (possesses full SN identification capability, with high data accuracy); 5G-A equipment: S0=0.90 (good accuracy of location and communication data); Radar equipment: S0=0.60 (unaffected by the environment, stable motion parameter measurement, no direct identification capability); Spectrum monitoring equipment: S0=0.85 (signal characteristic data is reliable, but location estimation accuracy is slightly low); 2. Dimensional correction factor C n (Positive gain, cumulative summation) Data integrity correction C1: When no core fields (deviceType, deviceId, uavId, latitude and longitude, altitude, timestamp) are missing, C1=0.05; when only non-core fields (speed, heading, metadata) are missing, C1=0.03; when some core fields are missing, C1=0. Preprocessing quality correction C2: When the preprocessed data is determined to be clean data without noise, C2=0.05; when the data is marked as suspicious but passes the second verification, C2=0.01; when the data fails the second verification, C2=0.

[0043] 3. Error deduction item D k (Reverse deduction, cumulative summation) Deductions are applied to address issues such as measurement errors and time deviations, thereby reducing the confidence level of low-quality data. Position deviation deduction D1: Based on the inherent accuracy of the sensor, D1=0 for analytical devices, D1=0.02 for 5G-A devices, D1=0.03 for radar devices, D1=0.04 for spectrum monitoring devices, and D1=0.05 for extended sensors; Time synchronization deviation deduction D2: When the time difference with UTC standard time after synchronization is ≤10ms, D2=0; when 10ms < time difference ≤50ms, D2=0.02; when time difference >50ms, D2=0.05. Data anomalies are deducted by D3: When data values ​​exceed a reasonable range (e.g., speed > 200 m / s, altitude < -100 m), D3 = 0.10; when data fluctuations exceed the mean of similar data ± 3 times the standard deviation, D3 = 0.08.

[0044] Step S102: For the parsed trajectory points, each parsed trajectory point is associated with the corresponding parsed trajectory according to the preset trajectory matching engine. If there is a blank parsed trajectory, multiple non-parsed trajectory points around the parsed trajectory that meet the preset conditions are associated according to the nearest neighbor method. The parsed trajectory is then filled and fused according to the non-parsed trajectory points to determine the corresponding fused parsed trajectory. Optionally, in this embodiment, the steps for obtaining the analytical trajectory by fusing the analytical trajectory points with extremely high reliability are as follows: Based on the unique serial number (SN) encoding and 100% data accuracy of the analyzed trajectory points, a dual-rule fusion strategy of "direct matching + point fusion" is adopted. Rule A - Direct Match: Triggering condition: The analyzed trajectory is in an "active" state. Execution logic: The trajectory matching engine directly associates the parsed trajectory point with the corresponding parsed trajectory, the cache pool manager updates the trajectory cache, the trajectory generator adds the point as the latest trajectory point to the trajectory list, and synchronously updates the trajectory status and confidence (confidence is maintained at 1.0).

[0045] Rule B - Merge and fill points: Triggering condition: If no new parsing trajectory points are received within the parsing trajectory prediction time window (10s), it is determined that a detection blank has occurred; Association filtering: The trajectory matching engine calls the nearest neighbor method to filter non-analyzed trajectory points with a time difference ≤ the parsed trajectory association time window (10s) and a spatial distance ≤ the parsed trajectory nearest neighbor method distance (150 meters), and pushes them to the parsed trajectory dedicated cache pool; Point fusion: When the number of non-analyzed trajectory points in the cache pool reaches the minimum number of points for parsing fusion (5), the fusion point processor uses a weighted average algorithm (weights are assigned according to the sensor confidence: the confidence of the trajectory points calculated in step S3) to fuse and generate supplementary trajectory points; Track update: The track generator adds the supplementary track points to the parsed track, marks the supplementary point attributes, and adjusts the track confidence to 0.9 (lower than the confidence of pure parsed track points) to maintain track continuity.

[0046] Step S103: For the non-analytical trajectory points, cluster each non-analytical trajectory point using a preset clustering algorithm, create non-analytical trajectories based on the effective clusters after clustering, and merge and update the non-analytical trajectories based on the newly received non-analytical trajectory points to determine the corresponding merged non-analytical trajectory. Optionally, in this embodiment, the steps for obtaining a non-analytical trajectory by fusing non-analytical trajectory points that are prone to generating false points are as follows: To address the issue of non-analyzable trajectory points lacking unique serial number (SN) codes and being prone to false alarms, a full-process fusion strategy of "clustering initiation - association update - termination determination" is adopted. 1. Trajectory point clustering Cache storage: Unresolved trajectory points first enter the clustering cache pool, and the cache pool manager stores them in sorted order by collection timestamp; Clustering operation: The clustering processor calls the DBSCAN clustering algorithm, with DBSCAN clustering distance (100 meters), DBSCAN clustering time window (10s), and DBSCAN minimum number of clustering points (5) as core parameters, to cluster trajectory points that are similar in space and time into clusters and filter out noise points (non-core points and non-boundary points); if the number of trajectory points in a cluster exceeds the minimum number of points 5, a new fused trajectory will be generated and will follow the "trajectory start".

[0047] Cluster validity determination: DBSCAN clustering sets an expiration timeout policy, each cluster is only retained for 30 seconds, and those that expire will be cleaned up to prevent memory overflow.

[0048] 2. Trajectory Start Triggering condition: A valid cluster is generated, and the cluster is not associated with any existing trajectory; Track creation: The track generator assigns a unique track identifier (fusedTargetId) to the cluster, sorts the track points in the cluster by timestamp as the initial track point list, marks the track type as "non-parsed track", initializes the status to "active", and records the sensor types involved in the cluster in the associated device list.

[0049] 3. Track association update: Association Matching: Newly received non-analyzable trajectory points are matched with existing active non-analyzable trajectories by the trajectory matching engine using the nearest neighbor method (spatial distance ≤ 150 meters for the nearest neighbor method of non-analyzable trajectory, time difference ≤ 10 seconds for the association time window of non-analyzable trajectory). Successful matching process: Add the trajectory point to the cache pool of the corresponding trajectory. When the number of newly added trajectory points in the cache pool is greater than or equal to the minimum number of points to be fused (3), call the Kalman filter algorithm, combine the previous state of the trajectory with the motion model, fuse the observation data to generate the optimal trajectory point, and update the trajectory list and confidence (the confidence is calculated in step S3). Historical trajectory matching: If no active trajectory is matched for the current trajectory point according to the current threshold, it will be matched from the historical trajectory. If the proportion of UAVID appearing in a certain trajectory is greater than or equal to 0.70, the current threshold will be dynamically doubled and the matching will be performed again. If a match is found, the processing will continue as in the previous step.

[0050] Matching failure handling: Return the trajectory point to the clustering cache pool and re-execute the trajectory point clustering process.

[0051] Step S104: After fusing the obtained analytical trajectory and non-analytical trajectory, output the corresponding unified trajectory data, wherein the unified trajectory data includes trajectory confidence, which is calculated based on the confidence of each trajectory point.

[0052] Optionally, in this embodiment, there is a module for the correlation, fusion, and output of analyzed and non-analyzed trajectories.

[0053] The trajectory output module, as the terminal presentation and data service unit of the multi-source trajectory fusion system, aims to generate high-precision, continuous, and unified trajectory data based on the core layer of trajectory fusion. Through a three-dimensional output system of "real-time push + historical storage + visualization," it achieves multi-scenario adaptability of the fused trajectory, providing standardized and highly available data support for applications such as real-time monitoring of low-altitude targets, historical backtracking analysis, and identity behavior assessment. Its specific technical solution is as follows: Real-time trajectory data includes core fields such as a unique trajectory identifier (fusedTargetId), the latest trajectory point (latitude, longitude, altitude, speed, heading, etc.), trajectory type (parsed / non-parsed), real-time status (active / terminated), current confidence level, and a list of associated devices. The data update frequency is consistent with the processing frequency of the fusion layer (up to 10Hz).

[0054] Historical trajectory data: Complete storage of trajectory lifecycle data, including a list of trajectory points (sorted in ascending order by UTC timestamp, including attributes such as the acquisition device, confidence level, and supplementary point identifier for each trajectory point), trajectory status change records, and key logs of the fusion process (such as clustering association results and supplementary point fusion details). It supports searching by time range, trajectory identifier, UAV SN / UAV ID, and other conditions.

[0055] Statistical analysis data: Statistical indicators derived from trajectory data, including flight duration, flight distance, maximum altitude, average speed, mean trajectory confidence score, contribution ratio of multi-sensor data, etc., support automatic aggregation and generation of statistical reports on a daily / weekly / monthly basis.

[0056] Alarm trigger data: When an abnormal state occurs in the trajectory (such as entering a no-fly zone, sudden speed change, or trajectory interruption and failure to be completed), alarm data is generated, which includes fields such as alarm type, trigger time, associated trajectory identifier, and abnormal parameter details.

[0057] This example demonstrates how this embodiment integrates multi-source trajectory data by constructing a full-process fusion system of "data acquisition - data preprocessing - dynamic parameter configuration - trajectory classification - core fusion - trajectory output". By leveraging standardized parameter configuration and combined algorithm applications, it achieves complementary correlation of multi-source trajectory data and ultimately obtains high-precision UAV trajectory data.

[0058] As described above, the low-altitude trajectory fusion method provided in this application can achieve unified access to multi-source data through a protocol adaptation gateway; perform spatiotemporal alignment, anomaly filtering, and cleaning on the data; classify trajectory points into two categories, analytical and non-analytical, based on whether they carry a unique UAV code, and calculate their confidence level; adopt a direct matching and weighted point supplementation fusion strategy for analytical trajectories, and adopt a combination strategy of DBSCAN clustering, nearest neighbor association, and Kalman filter update for non-analytical trajectories to generate and update trajectories; and finally output accurate unified trajectory data, thereby improving the accuracy and reliability of low-altitude trajectory reconstruction and fusion.

[0059] In one embodiment of the low-altitude trajectory fusion method of this application, it may further include the following: Step S201: Receive multi-source trajectory data of the UAV and perform basic rule verification. The multi-source trajectory data includes parsing device data, spectrum monitoring data, and radar data. Step S202: For the data that passes the basic rule verification, the mean and standard deviation of each data field are calculated by the sliding window algorithm. Data that deviates from the preset threshold is marked as abnormal noise. Abnormal noise data is removed and adjacent data is marked to determine the corresponding cleaned data. Step S203: Perform coordinate system operation on the cleaned data, and perform time alignment on the data after coordinate system operation by time zone offset correction and interpolation method to determine the corresponding standardized trajectory data.

[0060] Optionally, in this embodiment, multi-source trajectory data from different monitoring sources is first received, specifically including parsing device data that can provide a unique identifier for the UAV, spectrum monitoring data that captures target activity based on signal spectrum characteristics, and radar data that measures position and motion parameters through reflected waves.

[0061] For various types of data, pre-defined basic rules are executed for verification. For example, the integrity of the SN encoding format of the parsed device data is verified to ensure that it conforms to the predefined unique encoding specifications; the speed values ​​in radar data are range-determined to ensure that they are within the common flight speeds of UAVs (e.g., 0-200 m / s); and the signal frequency of the spectrum monitoring data is checked for compliance, limiting it to the commonly used communication frequency bands of UAVs (e.g., 2.4 GHz, 5.8 GHz). This filters out obviously non-compliant, incorrectly formatted, or clearly out-of-reasonable invalid data at the source, providing a basic and reliable data foundation for subsequent processing.

[0062] Optionally, in this embodiment, after passing the basic verification, a sliding window algorithm is used to dynamically assess the quality of continuous data points. Using a preset window size (e.g., 5 consecutive data points) as the unit, the mean and standard deviation of each key data field (e.g., position coordinates, velocity, signal strength, etc.) are calculated. Values ​​deviating from the mean ± 3 times the standard deviation are marked as abnormal noise points. Simultaneously, based on the historical accuracy characteristics of different sensors, differentiated dynamic thresholds are set for various types of devices; for example, a radar position error threshold of 5 meters and a 5G-A device threshold of 3 meters, to more accurately identify anomalies. For the marked noise data, a "direct removal + adjacent marking" processing strategy is implemented, that is, not only are abnormal points deleted, but adjacent data points are also marked as "suspicious data" to reduce their weight or perform secondary verification in subsequent fusion stages.

[0063] Optionally, in this embodiment, the cleaned data is first processed by a coordinate system operation. This transforms the original coordinate systems from different sensors (such as local coordinate systems, device-defined coordinate systems, etc.) to the internationally recognized WGS84 geodetic coordinate system, ensuring that all trajectory data have a consistent spatial reference, covering the three dimensions of longitude, latitude, and altitude. Subsequently, the time information is aligned. Time zone offset correction converts the local time of each device to UTC standard time. For high-frequency acquisition devices (such as radar), interpolation is used to smooth the timestamps, making the time intervals between adjacent data points more uniform and ensuring that the time synchronization error is controlled within milliseconds (e.g., ≤10ms).

[0064] Through step S203, this embodiment provides high-quality standardized data with a consistent spatiotemporal reference for subsequent trajectory matching, association, and fusion.

[0065] In one embodiment of the low-altitude trajectory fusion method of this application, it may further include the following: Step S301: Set confidence parameters, including base confidence, dimension correction coefficient, and error deduction item; Step S302: The basic confidence level is set according to the sensor type of the collected trajectory data, wherein the basic confidence level of the analysis device, radar device, and spectrum monitoring device decreases in that order. Step S303: The dimension correction coefficient includes a data integrity correction coefficient and a preprocessing quality correction coefficient, which are used to perform positive gain correction on data integrity and preprocessing quality; Step S304: The error deduction items include position deviation deduction items, time synchronization deviation deduction items, and data anomaly deduction items, which are used to deduct data deviations and data anomalies in reverse. Step S305: Construct a confidence quantification model based on the confidence parameters, and evaluate the confidence of each trajectory point using the confidence quantification model to determine the confidence of each trajectory point.

[0066] Optionally, in this embodiment, the confidence assessment process is based on a multi-dimensional quantification model, which achieves a refined assessment of the quality of trajectory points through parameterized configuration and calculation rules. Specifically, the confidence parameter system includes three core parameters: basic confidence, dimension correction coefficient, and error deduction term.

[0067] 1. The basic reliability S0 is a fixed value preset according to the sensor type, used to reflect the inherent reliability level of various sensors, where: Analysis device: S0=1 (possesses full SN identification capability, with high data accuracy); 5G-A equipment: S0=0.90 (good location and communication data accuracy); Radar equipment: S0=0.60 (unaffected by the environment, stable motion parameter measurement, no direct identification capability); Spectrum monitoring equipment: S0=0.85 (signal characteristic data is reliable, but location estimation accuracy is slightly low); 2. Dimensional correction factor C n Used to apply positive gain correction to data quality. It includes two categories: data integrity correction coefficients and preprocessing quality correction coefficients, among which: Data integrity correction C1: When no core fields (deviceType, deviceId, uavId, latitude and longitude, altitude, timestamp) are missing, C1=0.05; when only non-core fields (speed, heading, metadata) are missing, C1=0.03; when some core fields are missing, C1=0. Preprocessing quality correction C2: When the preprocessed data is determined to be clean data without noise, C2=0.05; when the data is marked as suspicious but passes the second verification, C2=0.01; when the data fails the second verification, C2=0.

[0068] 3. Error deduction item D k This is used to apply reverse penalties for data deviations and anomalies, including three deductions: location deviation deduction, time synchronization deviation deduction, and data anomaly deduction. Position deviation deduction D1: Based on the inherent accuracy of the sensor, D1=0 for analytical devices, D1=0.02 for 5G-A devices, D1=0.03 for radar devices, D1=0.04 for spectrum monitoring devices, and D1=0.05 for extended sensors; Time synchronization deviation deduction D2: When the time difference with UTC standard time after synchronization is ≤10ms, D2=0; when 10ms < time difference ≤50ms, D2=0.02; when time difference >50ms, D2=0.05. Data anomalies are deducted by D3: When data values ​​exceed a reasonable range (e.g., speed > 200 m / s, altitude < -100 m), D3 = 0.10; when data fluctuations exceed the mean of similar data ± 3 times the standard deviation, D3 = 0.08.

[0069] Based on the complete parameter settings, construct a quantitative calculation model for confidence metrics: Score = Basic credibility S0 × (1 + Σ dimensional correction coefficient C) n -Σ Error Deduction Item D k This model integrates basic credibility, quality gain, and error penalty to achieve a multi-dimensional fusion evaluation of trajectory point quality.

[0070] Finally, the sensor type, data integrity status, preprocessing results, position error, time deviation, and data anomalies corresponding to each trajectory point are substituted into the model to calculate the final confidence score for that trajectory point. This score serves as the core basis for subsequent trajectory classification, fusion weight allocation, and trajectory quality assessment.

[0071] Through step S305, this embodiment achieves a quantitative assessment of the quality of multi-source trajectory points, providing a precise data quality basis for subsequent differentiated fusion strategies.

[0072] In one embodiment of the low-altitude trajectory fusion method of this application, it may further include the following: Step S401: When a detection blank appears in the parsed trajectory, multiple non-parsed trajectory points around the parsed trajectory that meet the preset filtering conditions are filtered according to the nearest neighbor method, and stored in the parsed trajectory dedicated cache pool. The filtering conditions include time difference conditions and spatial distance conditions. Step S402: When the number of non-analyzed trajectory points in the dedicated cache pool of the parsed trajectory reaches a preset number of points, the non-analyzed trajectory points are fused and calculated according to the weighted average algorithm to determine the corresponding supplementary trajectory points. The weights of the weighted average algorithm are allocated according to the confidence level of each non-analyzed trajectory point. Step S403: Add the supplementary trajectory point as a supplementary point to the analytical trajectory as a supplementary point for the detection blank area, and mark the point as a supplementary point attribute, and reduce the overall confidence of the analytical trajectory after adding the supplementary point accordingly.

[0073] Optionally, in this embodiment, in the fusion processing of the parsed trajectory, a point-filling mechanism based on nearest neighbor matching and weighted fusion is designed to address the trajectory gaps caused by detection interruptions.

[0074] Specifically, if no new parsed trajectory points are received within a preset time window (10 seconds), the system determines that the trajectory has a detection gap. At this time, the trajectory matching engine filters the non-parsed trajectory points cached in the current system based on the nearest neighbor method.

[0075] The filtering criteria include: the time difference does not exceed the preset associated time window, which is usually set to 10 seconds; the spatial distance does not exceed the preset distance threshold, which is typically 150 meters.

[0076] Non-analytical trajectory points that meet the above spatiotemporal constraints will be extracted and stored in the dedicated cache pool of the analytical trajectory.

[0077] Next, the system continuously monitors the number of unresolved trajectory points in the cache pool. When the cumulative number of points reaches the preset minimum number of fusion points, such as 5 points, the fusion and point-filling processor starts a weighted average algorithm to perform point fusion calculations. The confidence level of each unresolved trajectory point serves as the basis for its weight allocation; the higher the confidence level, the greater the weight in the fusion calculation. Through weighted averaging, the system generates a more reliable and representative supplementary trajectory point in terms of spatial location and motion state.

[0078] Finally, the trajectory generator inserts this supplementary trajectory point into the corresponding time position of the original analytical trajectory as a supplement to the detection gaps. This point is explicitly marked with the "supplementary point" attribute to distinguish it from the directly detected analytical trajectory points. Meanwhile, since the supplementary point data comes from non-analyzed sensors, its reliability is lower than that of fully analyzed data. Therefore, the system accordingly lowers the overall confidence level of this analytical trajectory, for example, from 1.0 to 0.9.

[0079] Through step S403, this embodiment successfully fills in the gaps in the analytical trajectory and merges them, maintaining the continuity of the spatial trajectory while reflecting the source and quality changes of the trajectory data through confidence adjustment.

[0080] In one embodiment of the low-altitude trajectory fusion method of this application, it may further include the following: Step S501: Store each of the non-analyzed trajectory points into the clustering cache pool and sort them according to their timestamps; Step S502: Perform clustering operations on the non-parse trajectory points in the cache pool to be clustered according to the DBSCAN clustering algorithm to determine the corresponding effective clusters; Step S503: If the effective cluster is not associated with any existing trajectory data, then assign a unique trajectory identifier to the effective cluster, sort the non-analyzable trajectory points in the effective cluster by time, determine the corresponding non-analyzable trajectory, and mark the non-analyzable trajectory as active.

[0081] Optionally, in this embodiment, the classified non-analyzed trajectory points are first uniformly stored in a dedicated clustering cache pool, and then sorted in ascending order according to the timestamps reported by each trajectory point, thereby ensuring the temporal consistency and data orderliness during subsequent clustering processing.

[0082] Secondly, the DBSCAN clustering algorithm is used to perform clustering operations on the non-analytical trajectory points stored in the clustering cache pool. Based on preset clustering distance thresholds, time window thresholds, and minimum point count thresholds, this algorithm aggregates spatially and temporally close trajectory points into clusters, while filtering out discrete noise points to form several effective clusters. Specifically, the clustering distance threshold limits the maximum allowable spatial spacing between trajectory points, the time window threshold constrains the maximum time span of trajectory points within the same cluster, and the minimum point count threshold specifies the minimum number of trajectory points required to form an effective cluster. These three factors together ensure the reasonableness and reliability of the clustering results in both the spatiotemporal dimensions.

[0083] Finally, when the system detects the generation of a valid cluster, and this cluster has not yet been associated with any existing trajectory in the system, a globally unique trajectory identifier is assigned to the cluster. All trajectory points within the cluster are then organized into an initial trajectory point list in chronological order, thereby generating a new non-parsed trajectory. This trajectory is initialized to an "active" state, and a list of sensor types associated with this trajectory is recorded for use in subsequent trajectory association and update processes.

[0084] Through step S503, this embodiment successfully achieved automatic clustering and trajectory generation of non-analyzed trajectory points, avoiding the isolation and loss of trajectory points without identity information.

[0085] In one embodiment of the low-altitude trajectory fusion method of this application, it may further include the following: Step S601: Match the newly received non-analytical trajectory points with all currently marked as active non-analytical trajectories using the nearest neighbor method; Step S602: If the match is successful, the non-analyzed trajectory point is stored in the corresponding non-analyzed trajectory dedicated cache pool. When the number of newly added non-analyzed trajectory points in the non-analyzed trajectory dedicated cache pool reaches the preset number of points, the trajectory points in the cache pool are fused according to the Kalman filter algorithm to determine the corresponding fused non-analyzed trajectory and update the confidence of the non-analyzed trajectory. Step S603: If the matching fails, perform historical trajectory backtracking matching, and match the newly received non-parsed trajectory points with all terminated non-parsed trajectories in the historical trajectory database; Step S604: If a matching historical trajectory exists, mark the status of the historical trajectory as active, perform fusion processing based on the non-analyzable trajectory point and the historical trajectory, determine the corresponding fused non-analyzable trajectory, and update the confidence of the non-analyzable trajectory. Step S605: If no matching historical trajectory exists, store the unparsed trajectory point in the clustering cache pool for re-execution of the clustering process.

[0086] Optionally, in this embodiment, this step uses the nearest neighbor method to match the newly received non-parsed trajectory point with all currently marked "active" non-parsed trajectories. The core basis of the matching is to find the closest existing trajectory within a preset spatial distance threshold and time window, thereby initially determining which continuously updating trajectory the trajectory point may belong to. This step establishes the spatial and temporal correlation between trajectories and trajectory points, providing a basic judgment basis for subsequent fusion.

[0087] For successful matches, the non-parsed trajectory point is first temporarily stored in a dedicated cache pool for the corresponding trajectory. Once the number of newly accumulated trajectory points in the cache pool reaches a preset minimum, the Kalman filter algorithm is invoked to perform batch fusion processing on these trajectory points. Based on the trajectory's state estimate from the previous moment and multi-point information from current observations, the Kalman filter calculates the optimal estimated position and motion parameters of the trajectory at the current moment through two stages: state prediction and measurement update. This generates the fused trajectory points and updates the confidence level of the entire trajectory. This step, through multi-point smoothing and filtering, effectively reduces single-point observation noise and improves the accuracy and stability of trajectory positioning.

[0088] In cases of failed matching, a historical trajectory backtracking match is performed, comparing the current trajectory point with all terminated (i.e., previously ended) non-parsed trajectories in the historical trajectory database. The aim is to identify whether the point belongs to a trajectory that was previously observed but temporarily lost due to intermittent detection, thereby restoring and continuing the tracking of that trajectory. This mechanism enhances the system's ability to re-associate after trajectory interruption and reduces trajectory splitting caused by brief signal loss.

[0089] If a match is found in the historical trajectory, the status of that historical trajectory is remarked as "active," indicating that trajectory tracking has resumed. Next, the system fuses the newly received trajectory point with the existing trajectory point information from the historical trajectory to generate an updated continuous trajectory. This process not only reactivates the trajectory but also maintains continuity and smoothness during the recovery process by fusing new and old information.

[0090] If both active trajectory matching and historical trajectory backtracking fail, the trajectory point is stored in the clustering cache pool, awaiting subsequent trajectory re-analysis using a clustering algorithm (such as DBSCAN). This ensures that unmatched discrete trajectory points are not lost and still have the opportunity to form new trajectories through clustering, thereby improving the system's ability to capture new targets and the completeness of trajectory generation.

[0091] Through step S605, this embodiment successfully performs adaptive fusion and update of the non-analytical trajectory based on the newly received non-analytical trajectory points, ensuring the continuity of the spatial trajectory.

[0092] In one embodiment of the low-altitude trajectory fusion method of this application, it may further include the following: Step S701: The obtained parsed trajectory and non-parsed trajectory are fused in the data fusion layer. The fused data includes the unique trajectory identifier, the latest trajectory point, the trajectory type, the real-time status, the current trajectory confidence level, and the list of associated devices. Step S702: Output the unified trajectory data obtained after fusion. The output includes trajectory data and statistical indicators derived from the trajectory data.

[0093] Optionally, in this embodiment, at the data fusion layer, the system performs a final fusion process on the parsed trajectory and the non-parsed trajectory generated in the aforementioned steps.

[0094] The fusion process uses the unique identifier of the trajectory as the core index, integrating key information from both types of trajectories one by one: First, the latest trajectory points of each trajectory are extracted, including real-time status parameters such as latitude and longitude, altitude, speed, and heading angle; then, the trajectory type is identified and distinguished as "analyzed trajectory" or "non-analyzed trajectory"; at the same time, the real-time running status of the trajectory is recorded, such as "active", "to be completed", or "terminated"; the overall confidence of the current trajectory is calculated and updated, which integrates the weighted average confidence of trajectory points, trajectory continuity coefficient, and sensor data consistency assessment; finally, a list of associated devices is established, recording all sensor types involved in the generation process of the trajectory and their data contribution weights.

[0095] The unified trajectory data after fusion is output through a standardized interface. The output includes complete trajectory sequence data; each trajectory contains all trajectory points and their attribute information in chronological order, supporting millisecond-level timestamp retrieval. Simultaneously, the system generates multi-dimensional statistical indicators in real time based on the trajectory data, including core indicators such as total flight time, cumulative flight distance, maximum flight altitude, average speed, and trajectory confidence distribution curve. It can also analyze the contribution ratio of sensor data by device type, examining the participation of each sensor in trajectory formation. The output mechanism combines streaming push and batch storage; real-time push supports protocols such as WebSocket, historical data is stored in a time-series database, and a visual query interface is provided.

[0096] Through step S702, this embodiment successfully obtained a unified fused trajectory through the final trajectory fusion layer, which has high precision and high continuity, enabling the UAV monitoring system to simultaneously meet the needs of multiple scenarios such as real-time situational awareness, historical behavior backtracking, flight pattern analysis, and abnormal pattern recognition.

[0097] To improve the accuracy and reliability of low-altitude trajectory reconstruction and fusion, this application provides an embodiment of a low-altitude trajectory fusion apparatus for implementing all or part of the aforementioned low-altitude trajectory fusion method. See [link to embodiment]. Figure 2 The low-altitude trajectory fusion device specifically includes the following components: The multi-source trajectory data processing module 10 is used to receive multi-source trajectory data of UAV, clean and perform spatiotemporal alignment processing on the multi-source trajectory data, determine the corresponding standardized trajectory data, extract trajectory points from the standardized trajectory data, classify the trajectory points into parsed trajectory points and non-parsed trajectory points according to whether the trajectory points carry the UAV's unique code, and evaluate the confidence of each trajectory point according to the set confidence parameters. The parsing trajectory fusion module 20 is used to associate each parsing trajectory point with the corresponding parsing trajectory according to a preset trajectory matching engine. If there is a blank parsing trajectory, it associates multiple non-parsing trajectory points around the parsing trajectory that meet the preset conditions according to the nearest neighbor method, and performs point filling and fusion on the parsing trajectory according to the non-parsing trajectory points to determine the corresponding fused parsing trajectory. The non-analytical trajectory fusion module 30 is used to cluster the non-analytical trajectory points using a preset clustering algorithm, create non-analytical trajectories based on the effective clusters after clustering, and fuse and update the non-analytical trajectories based on newly received non-analytical trajectory points to determine the corresponding fused non-analytical trajectory. The unified trajectory determination module 40 is used to fuse the obtained analytical trajectory and non-analytical trajectory and output the corresponding unified trajectory data. The unified trajectory data includes trajectory confidence, which is calculated based on the confidence of each trajectory point.

[0098] As described above, the low-altitude trajectory fusion device provided in this application embodiment can achieve unified access to multi-source data through a protocol adaptation gateway; perform spatiotemporal alignment, anomaly filtering, and cleaning on the data; classify trajectory points into two categories, analytical and non-analytical, based on whether they carry a unique UAV code, and calculate their confidence level; adopt a direct matching and weighted point supplementation fusion strategy for analytical trajectories, and adopt a combination strategy of DBSCAN clustering, nearest neighbor association, and Kalman filter update for non-analytical trajectories to generate and update trajectories; and finally output accurate unified trajectory data, thereby improving the accuracy and reliability of low-altitude trajectory reconstruction and fusion.

[0099] From a hardware perspective, in order to improve the accuracy and reliability of low-altitude trajectory reconstruction and fusion, this application provides an embodiment of an electronic device for implementing all or part of the low-altitude trajectory fusion method, wherein the electronic device specifically includes the following: The system comprises a processor, memory, a communications interface, and a bus; wherein the processor, memory, and communications interface communicate with each other via the bus; the communications interface is used to realize information transmission between the low-altitude trajectory fusion method and core business systems, user terminals, and related databases and other related devices; the logic controller can be a desktop computer, tablet computer, or mobile terminal, etc., and this embodiment is not limited to these. In this embodiment, the logic controller can be implemented with reference to the embodiments of the low-altitude trajectory fusion method in the present embodiment, and the contents of the embodiments of the low-altitude trajectory fusion method are incorporated herein, and repeated details will not be described again.

[0100] It is understood that the user terminal may include smartphones, tablet computers, network set-top boxes, portable computers, desktop computers, personal digital assistants (PDAs), in-vehicle devices, smart wearable devices, etc. Among these, the smart wearable devices may include smart glasses, smartwatches, smart bracelets, etc.

[0101] In practical applications, parts of the low-altitude trajectory fusion method can be executed on the electronic device side as described above, or all operations can be completed in the client device. The choice can be made based on the processing power of the client device and the limitations of the user's usage scenario. This application does not impose any limitations in this regard. If all operations are completed in the client device, the client device may further include a processor.

[0102] The aforementioned client device may have a communication module (i.e., a communication unit) that can communicate with a remote server to achieve data transmission with the server. The server may include a server on the task scheduling center side; in other implementation scenarios, it may also include a server on an intermediate platform, such as a server on a third-party server platform that has a communication link with the task scheduling center server. The server may include a single computer device, a server cluster consisting of multiple servers, or a distributed server structure.

[0103] Figure 3 This is a schematic block diagram illustrating the system configuration of the electronic device 9600 according to an embodiment of this application. Figure 3 As shown, the electronic device 9600 may include a central processing unit 9100 and a memory 9140; the memory 9140 is coupled to the central processing unit 9100. It is worth noting that... Figure 3 This is an example; other types of structures can also be used to supplement or replace this structure to achieve telecommunications functions or other functions.

[0104] In one embodiment, the low-altitude trajectory fusion method functionality can be integrated into the central processing unit 9100. The central processing unit 9100 can be configured to perform the following control: Step S101: Receive multi-source trajectory data of UAV, clean and spatiotemporally align the multi-source trajectory data, determine the corresponding standardized trajectory data, extract trajectory points from the standardized trajectory data, classify the trajectory points into parsed trajectory points and non-parsed trajectory points according to whether the trajectory points carry the UAV's unique code, and evaluate the confidence of each trajectory point according to the set confidence parameter. Step S102: For the parsed trajectory points, each parsed trajectory point is associated with the corresponding parsed trajectory according to the preset trajectory matching engine. If there is a blank parsed trajectory, multiple non-parsed trajectory points around the parsed trajectory that meet the preset conditions are associated according to the nearest neighbor method. The parsed trajectory is then filled and fused according to the non-parsed trajectory points to determine the corresponding fused parsed trajectory. Step S103: For the non-analytical trajectory points, cluster each non-analytical trajectory point using a preset clustering algorithm, create non-analytical trajectories based on the effective clusters after clustering, and merge and update the non-analytical trajectories based on the newly received non-analytical trajectory points to determine the corresponding merged non-analytical trajectory. Step S104: After fusing the obtained analytical trajectory and non-analytical trajectory, output the corresponding unified trajectory data, wherein the unified trajectory data includes trajectory confidence, which is calculated based on the confidence of each trajectory point.

[0105] As described above, the electronic device provided in this application embodiment achieves unified access to multi-source data through a protocol adaptation gateway; performs spatiotemporal alignment, anomaly filtering, and cleaning on the data; classifies trajectory points into two categories, analytical and non-analytical, based on whether they carry a unique UAV code, and calculates their confidence level; adopts a direct matching and weighted point-filling fusion strategy for analytical trajectories, and a combination strategy of DBSCAN clustering, nearest neighbor association, and Kalman filter update for non-analytical trajectories to generate and update trajectories; and finally outputs accurate unified trajectory data, thereby improving the accuracy and reliability of low-altitude trajectory reconstruction and fusion.

[0106] In another embodiment, the low-altitude trajectory fusion method can be configured separately from the central processing unit 9100. For example, the low-altitude trajectory fusion method can be configured as a chip connected to the central processing unit 9100, and the low-altitude trajectory fusion method function can be implemented through the control of the central processing unit.

[0107] like Figure 3 As shown, the electronic device 9600 may further include: a communication module 9110, an input unit 9120, an audio processor 9130, a display 9160, and a power supply 9170. It is worth noting that the electronic device 9600 does not necessarily need to include these components. Figure 3 All components shown; in addition, the electronic device 9600 may also include Figure 3 For components not shown, please refer to existing technologies.

[0108] like Figure 3 As shown, the central processing unit 9100, sometimes also referred to as a controller or operating control, may include a microprocessor or other processor device and / or logic device, which receives inputs and controls the operation of various components of the electronic device 9600.

[0109] The memory 9140 may be, for example, one or more of a cache, flash memory, hard drive, removable media, volatile memory, non-volatile memory, or other suitable devices. It may store the aforementioned failure-related information, and also store a program for executing that information. The central processing unit 9100 may execute the program stored in the memory 9140 to perform information storage or processing, etc.

[0110] Input unit 9120 provides input to central processing unit 9100. Input unit 9120 may be, for example, a keypad or touch input device. Power supply 9170 provides power to electronic device 9600. Display 9160 displays images and text. Display may be, for example, an LCD display, but is not limited thereto.

[0111] The memory 9140 can be a solid-state memory, such as a read-only memory (ROM), random access memory (RAM), a SIM card, etc. It can also be a memory that retains information even when power is off, can be selectively erased, and contains more data; examples of this type of memory are sometimes referred to as EPROMs. The memory 9140 can also be some other type of device. The memory 9140 includes a buffer memory 9141 (sometimes referred to as a buffer). The memory 9140 may include an application / function storage unit 9142 for storing application programs and function programs or processes for executing the operation of the electronic device 9600 via the central processing unit 9100.

[0112] The memory 9140 may also include a data storage unit 9143 for storing data, such as contacts, digital data, pictures, sounds, and / or any other data used by the electronic device. The driver storage unit 9144 of the memory 9140 may include various drivers for the electronic device for communication functions and / or for performing other functions of the electronic device (such as messaging applications, address book applications, etc.).

[0113] The communication module 9110 is a transmitter / receiver that sends and receives signals via the antenna 9111. The communication module 9110 is coupled to the central processing unit 9100 to provide input signals and receive output signals, which is the same as in a conventional mobile communication terminal.

[0114] Based on different communication technologies, multiple communication modules 9110 can be configured in the same electronic device, such as cellular network modules, Bluetooth modules, and / or wireless LAN modules. The communication module 9110 is also coupled to a speaker 9131 and a microphone 9132 via an audio processor 9130 to provide audio output via the speaker 9131 and receive audio input from the microphone 9132, thereby realizing typical telecommunications functions. The audio processor 9130 may include any suitable buffer, decoder, amplifier, etc. Furthermore, the audio processor 9130 is also coupled to a central processing unit 9100, enabling on-device recording via the microphone 9132 and on-device playback of stored sound via the speaker 9131.

[0115] Embodiments of this application also provide a computer-readable storage medium capable of implementing all steps of the low-altitude trajectory fusion method with a server or client as the execution subject in the above embodiments. The computer-readable storage medium stores a computer program that, when executed by a processor, implements all steps of the low-altitude trajectory fusion method with a server or client as the execution subject in the above embodiments. For example, when the processor executes the computer program, it implements the following steps: Step S101: Receive multi-source trajectory data of UAV, clean and spatiotemporally align the multi-source trajectory data, determine the corresponding standardized trajectory data, extract trajectory points from the standardized trajectory data, classify the trajectory points into parsed trajectory points and non-parsed trajectory points according to whether the trajectory points carry the UAV's unique code, and evaluate the confidence of each trajectory point according to the set confidence parameter. Step S102: For the parsed trajectory points, each parsed trajectory point is associated with the corresponding parsed trajectory according to the preset trajectory matching engine. If there is a blank parsed trajectory, multiple non-parsed trajectory points around the parsed trajectory that meet the preset conditions are associated according to the nearest neighbor method. The parsed trajectory is then filled and fused according to the non-parsed trajectory points to determine the corresponding fused parsed trajectory. Step S103: For the non-analytical trajectory points, cluster each non-analytical trajectory point using a preset clustering algorithm, create non-analytical trajectories based on the effective clusters after clustering, and merge and update the non-analytical trajectories based on the newly received non-analytical trajectory points to determine the corresponding merged non-analytical trajectory. Step S104: After fusing the obtained analytical trajectory and non-analytical trajectory, output the corresponding unified trajectory data, wherein the unified trajectory data includes trajectory confidence, which is calculated based on the confidence of each trajectory point.

[0116] As described above, the computer-readable storage medium provided in this application embodiment enables unified access to multi-source data through a protocol adaptation gateway; performs spatiotemporal alignment, anomaly filtering, and cleaning on the data; classifies trajectory points into two categories, analytical and non-analytical, based on whether they carry a unique UAV code, and calculates their confidence level; adopts a direct matching and weighted point-filling fusion strategy for analytical trajectories, and a combination strategy of DBSCAN clustering, nearest neighbor association, and Kalman filter update for non-analytical trajectories to generate and update trajectories; and finally outputs accurate unified trajectory data, thereby improving the accuracy and reliability of low-altitude trajectory reconstruction and fusion.

[0117] Embodiments of this application also provide a computer program product capable of implementing all steps of the low-altitude trajectory fusion method in the above embodiments, where the execution subject is a server or a client. When executed by a processor, this computer program / instruction implements the steps of the low-altitude trajectory fusion method. For example, the computer program / instruction implements the following steps: Step S101: Receive multi-source trajectory data of UAV, clean and spatiotemporally align the multi-source trajectory data, determine the corresponding standardized trajectory data, extract trajectory points from the standardized trajectory data, classify the trajectory points into parsed trajectory points and non-parsed trajectory points according to whether the trajectory points carry the UAV's unique code, and evaluate the confidence of each trajectory point according to the set confidence parameter. Step S102: For the parsed trajectory points, each parsed trajectory point is associated with the corresponding parsed trajectory according to the preset trajectory matching engine. If there is a blank parsed trajectory, multiple non-parsed trajectory points around the parsed trajectory that meet the preset conditions are associated according to the nearest neighbor method. The parsed trajectory is then filled and fused according to the non-parsed trajectory points to determine the corresponding fused parsed trajectory. Step S103: For the non-analytical trajectory points, cluster each non-analytical trajectory point using a preset clustering algorithm, create non-analytical trajectories based on the effective clusters after clustering, and merge and update the non-analytical trajectories based on the newly received non-analytical trajectory points to determine the corresponding merged non-analytical trajectory. Step S104: After fusing the obtained analytical trajectory and non-analytical trajectory, output the corresponding unified trajectory data, wherein the unified trajectory data includes trajectory confidence, which is calculated based on the confidence of each trajectory point.

[0118] As described above, the computer program product provided in this application embodiment enables unified access to multi-source data through a protocol adaptation gateway; performs spatiotemporal alignment, anomaly filtering, and cleaning on the data; classifies trajectory points into two categories, analytical and non-analytical, based on whether they carry a unique UAV code, and calculates their confidence level; adopts a direct matching and weighted point-filling fusion strategy for analytical trajectories, and a combination strategy of DBSCAN clustering, nearest neighbor association, and Kalman filter update for non-analytical trajectories to generate and update trajectories; and finally outputs accurate unified trajectory data, thereby improving the accuracy and reliability of low-altitude trajectory reconstruction and fusion.

[0119] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, apparatus, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0120] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (devices), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0121] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0122] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0123] Specific embodiments have been used to illustrate the principles and implementation methods of this invention. The descriptions of the embodiments above are only for the purpose of helping to understand the method and core ideas of this invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this invention. Therefore, the content of this specification should not be construed as a limitation of this invention.

Claims

1. A low-altitude trajectory fusion method, characterized in that, The method includes: The system receives multi-source trajectory data from a UAV, cleans and performs spatiotemporal alignment on the multi-source trajectory data, determines the corresponding standardized trajectory data, extracts trajectory points from the standardized trajectory data, classifies the trajectory points into parsed trajectory points and non-parsed trajectory points based on whether the trajectory points carry a unique UAV code, and evaluates the confidence level of each trajectory point according to a set confidence parameter. For the parsed trajectory points, each parsed trajectory point is associated with the corresponding parsed trajectory according to the preset trajectory matching engine. If there is a blank parsed trajectory, multiple non-parsed trajectory points around the parsed trajectory that meet the preset conditions are associated according to the nearest neighbor method. The parsed trajectory is then filled and fused according to the non-parsed trajectory points to determine the corresponding fused parsed trajectory. For the non-analytical trajectory points, a preset clustering algorithm is used to cluster each non-analytical trajectory point. Non-analytical trajectories are created based on the effective clusters after clustering. The non-analytical trajectories are then merged and updated based on newly received non-analytical trajectory points to determine the corresponding merged non-analytical trajectory. After fusing the obtained analytical and non-analytical trajectories, the corresponding unified trajectory data is output. The unified trajectory data includes trajectory confidence, which is calculated based on the confidence of each trajectory point.

2. The low-altitude trajectory fusion method according to claim 1, characterized in that, The process of receiving multi-source trajectory data from the UAV, cleaning and spatiotemporally aligning the multi-source trajectory data, and determining the corresponding standardized trajectory data includes: The system receives multi-source trajectory data from the UAV and performs basic rule verification. The multi-source trajectory data includes data from the parsing device, spectrum monitoring data, and radar data. For the data that passes the basic rule verification, the mean and standard deviation of each data field are calculated using a sliding window algorithm. Data that deviates from the preset threshold is marked as abnormal noise. Abnormal noise data is then removed and adjacent data is marked to determine the corresponding cleaned data. The cleaned data is subjected to a coordinate system operation, and the data after the coordinate system operation is time-aligned by time zone offset correction and interpolation to determine the corresponding standardized trajectory data.

3. The low-altitude trajectory fusion method according to claim 2, characterized in that, The step of evaluating the confidence level of each trajectory point according to the set confidence level parameter includes: Set confidence parameters, including base confidence, dimension correction factor, and error deduction term; The basic confidence level is set according to the sensor type of the collected trajectory data, with the basic confidence level decreasing sequentially for analysis equipment, radar equipment, and spectrum monitoring equipment. The dimension correction coefficients include data integrity correction coefficients and preprocessing quality correction coefficients, which are used to perform positive gain corrections on data integrity and preprocessing quality. The error deduction items include position deviation deduction items, time synchronization deviation deduction items, and data anomaly deduction items, which are used to deduct data deviations and data anomalies in reverse. A confidence quantification model is constructed based on the confidence parameters, and the confidence of each trajectory point is evaluated using the confidence quantification model to determine the confidence of each trajectory point.

4. The low-altitude trajectory fusion method according to claim 1, characterized in that, If a blank in the analytical trajectory occurs, multiple non-analytical trajectory points around the analytical trajectory that meet preset conditions are associated according to the nearest neighbor method. The analytical trajectory is then filled and fused based on the non-analytical trajectory points to determine the corresponding fused analytical trajectory, including: When a detection gap appears in the parsed trajectory, multiple non-parsed trajectory points around the parsed trajectory that meet the preset filtering conditions are selected according to the nearest neighbor method and stored in the parsed trajectory's dedicated cache pool. The filtering conditions include time difference conditions and spatial distance conditions. When the number of non-analyzed trajectory points in the dedicated cache pool of the parsed trajectory reaches a preset number, the non-analyzed trajectory points are fused and calculated according to the weighted average algorithm to determine the corresponding supplementary trajectory points. The weights of the weighted average algorithm are allocated according to the confidence level of each non-analyzed trajectory point. The supplementary trajectory points are added to the analytical trajectory as supplementary points for the detected blank areas. At the same time, the points are marked as supplementary points, and the overall confidence of the analytical trajectory after adding supplementary points is reduced accordingly.

5. The low-altitude trajectory fusion method according to claim 1, characterized in that, The step of clustering the non-analytical trajectory points using a preset clustering algorithm and creating non-analytical trajectories based on the effective clusters after clustering includes: Each of the non-analyzed trajectory points is stored in the clustering cache pool and sorted according to timestamp; The DBSCAN clustering algorithm is used to perform clustering operations on the non-parse trajectory points in the cache pool to be clustered, and the corresponding effective clusters are determined. If the effective cluster is not associated with any existing trajectory data, a unique trajectory identifier is assigned to the effective cluster, and the non-analyzable trajectory points in the effective cluster are sorted by time to determine the corresponding non-analyzable trajectory. At the same time, the non-analyzable trajectory is marked as active.

6. The low-altitude trajectory fusion method according to claim 5, characterized in that, The step of fusing and updating the non-analyzable trajectory based on the newly received non-analyzable trajectory points to determine the corresponding fused non-analyzable trajectory includes: The newly received non-analyzable trajectory points are matched with all currently marked as active non-analyzable trajectories using the nearest neighbor method; If a match is successful, the non-analyzed trajectory point is stored in the corresponding non-analyzed trajectory dedicated cache pool. When the number of newly added non-analyzed trajectory points in the non-analyzed trajectory dedicated cache pool reaches the preset number of points, the trajectory points in the cache pool are fused according to the Kalman filter algorithm to determine the corresponding fused non-analyzed trajectory and update the confidence of the non-analyzed trajectory. If the match fails, the historical trajectory backtracking match is performed, and the newly received non-parsed trajectory points are matched with all terminated non-parsed trajectories in the historical trajectory database. If a matching historical trajectory exists, the status of the historical trajectory is marked as active. The non-analyzable trajectory point and the historical trajectory are fused together to determine the corresponding fused non-analyzable trajectory, and the confidence of the non-analyzable trajectory is updated. If no matching historical trajectory exists, the unparsed trajectory point is stored in the clustering cache pool for later re-execution of the clustering process.

7. The low-altitude trajectory fusion method according to claim 1, characterized in that, The process of fusing the obtained analytical trajectory and non-analytical trajectory to output corresponding unified trajectory data includes: The obtained parsed and non-parsed trajectories are fused in the data fusion layer. The fused data includes the unique identifier of the trajectory, the latest trajectory point, the trajectory type, the real-time status, the current trajectory confidence level, and the list of associated devices. The unified trajectory data obtained after fusion will be output, including trajectory data and statistical indicators derived from the trajectory data.

8. A low-altitude trajectory fusion device, characterized in that, The device includes: The multi-source trajectory data processing module is used to receive multi-source trajectory data from UAVs, clean and perform spatiotemporal alignment processing on the multi-source trajectory data, determine the corresponding standardized trajectory data, extract trajectory points from the standardized trajectory data, classify the trajectory points into parsed trajectory points and non-parsed trajectory points according to whether the trajectory points carry the UAV's unique code, and evaluate the confidence of each trajectory point according to the set confidence parameters. The analytical trajectory fusion module is used to associate each analytical trajectory point with the corresponding analytical trajectory according to a preset trajectory matching engine. If there is a blank analytical trajectory, it associates multiple non-analytical trajectory points around the analytical trajectory that meet the preset conditions according to the nearest neighbor method, and performs point filling and fusion on the analytical trajectory according to the non-analytical trajectory points to determine the corresponding fused analytical trajectory. The non-analytical trajectory fusion module is used to cluster the non-analytical trajectory points using a preset clustering algorithm, create non-analytical trajectories based on the effective clusters after clustering, and fuse and update the non-analytical trajectories based on newly received non-analytical trajectory points to determine the corresponding fused non-analytical trajectory. The unified trajectory determination module is used to fuse the obtained analytical trajectory and non-analytical trajectory and output the corresponding unified trajectory data. The unified trajectory data includes trajectory confidence, which is calculated based on the confidence of each trajectory point.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the low-altitude trajectory fusion method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the steps of the low-altitude trajectory fusion method according to any one of claims 1 to 7.

Images