Dynamic grouping-based multi-station unmanned aerial vehicle detection method and system, and storage medium

Abstract

The application discloses a kind of multi-station unmanned aerial vehicle detection methods, systems and storage medium based on dynamic grouping, comprising steps S1, the position information of the multiple detection equipment deployed in monitoring area and the detection data of target unmanned aerial vehicle are dynamically acquired as the characteristic information of each detection equipment;Step S2, based on the similarity of each detection equipment characteristic information, all detection equipment is divided into several equipment groups;Step S3, the subsequent detection data of detection equipment in the same equipment group is compared, and each detection equipment is judged to be normal or abnormal state;Step S4, the detection data of the detection equipment determined to be normal state is used to fuse the positioning of target unmanned aerial vehicle.The application ensures that the communication environment of equipment in group is similar through dynamic grouping, and solves the problems of misjudgment of edge equipment, large amount of calculation and difficulty of adaptive target dynamic change of fixed setting of cooperative group by consistency evaluation combined with numerical reference and change reference, to improve the accuracy and stability of unmanned aerial vehicle detection positioning.

Description

Technical Field

[0001] This application relates to the field of unmanned aerial vehicle (UAV) detection and positioning technology, and in particular to a multi-station UAV detection method, system and storage medium based on dynamic grouping. Background Technology

[0002] In existing technologies, the positioning accuracy of single-station UAV detection is insufficient, therefore multi-station positioning schemes are commonly adopted, including gridded multi-station detection schemes and distributed cooperative group multi-station detection schemes. To ensure detection stability and prevent abnormal positioning jumps, it is necessary to evaluate the detection quality of each detection device. Existing gridded schemes include all detection devices in a lateral comparison for consistency evaluation. However, for detection devices on the opposite edge of the grid, their communication environment differs significantly from that of the target UAV, resulting in inconsistent detection data. Including all detection devices in a lateral comparison consistency evaluation is unreasonable, prone to misjudgment, and computationally intensive. Furthermore, the cooperative groups in existing distributed schemes are usually fixed during initial deployment and cannot be reasonably grouped according to the actual position of the target UAV, making it difficult to adapt to dynamic changes in the target. Summary of the Invention

[0003] The purpose of this invention is to provide a multi-station UAV detection method, system, and storage medium based on dynamic grouping, which can dynamically adjust the grouping of detection devices according to the actual location of the target UAV, perform consistency evaluation within the group, accurately and efficiently identify abnormal detection devices, thereby improving the accuracy and efficiency of target UAV detection and positioning.

[0004] To achieve the above objectives, the present invention adopts the following technical solution: This invention provides a multi-station UAV detection method based on dynamic grouping, comprising the following steps: Step S1: Dynamically acquire the location information of multiple detection devices deployed in the monitoring area, as well as the detection data of each detection device on the target UAV, as the feature information of each detection device; Step S2: Based on the similarity of the feature information of each of the detection devices, all the detection devices are divided into several device groups; Step S3: Compare the subsequent detection data of the detection devices in the same device group to determine whether each detection device is in a normal or abnormal state. Step S4: Use the detection data of the detection device that is determined to be in normal condition to perform fusion positioning of the target UAV.

[0005] Further, step S3 includes: S31, For each of the detection devices in the same device group, acquire their subsequent detection data of the same target UAV, and perform normalization preprocessing on the subsequent detection data; S32, calculate the deviation of the detection data of each detection device in the same device group from the consistency benchmark within the group; S33, based on the comparison result between the deviation amount and the preset deviation threshold, determine whether each of the detection devices is in a normal state or an abnormal state.

[0006] Furthermore, in step S32, the intra-group consistency benchmark includes a numerical benchmark and / or a change benchmark, and the deviation includes the deviation of each detection index in the detection data of the detection device from the intra-group consistency benchmark. When one or more deviations exceed the preset deviation threshold range, the device is determined to be in an abnormal state.

[0007] Furthermore, the numerical benchmark includes the median and / or weighted mean of each detection index of all detection data from the detection devices within the group; the variation benchmark includes the first-order difference median of each detection index of all detection data from the detection devices within the group.

[0008] Furthermore, the preset deviation threshold in step S33 is dynamically adjusted based on the number of detection devices in the group and / or the motion state of the target UAV. When the number of detection devices in the group is less than the set value or the target UAV is maneuvering violently, the preset deviation threshold is relaxed.

[0009] Furthermore, the detection indicators include at least one of the following indicators: angle of arrival, azimuth, elevation, received signal strength, signal-to-noise ratio, distance, and radial velocity.

[0010] Furthermore, the similarity of the feature information of the detection device is calculated using Euclidean distance, cosine similarity, or Mahalanobis distance.

[0011] Furthermore, the dynamic acquisition in step S1 includes timed acquisition and triggered acquisition. The triggering conditions for the triggered acquisition include: the number of detection devices determined to be in an abnormal state in step S3 is higher than a set number threshold, and / or the difference between the fused positioning and historical positioning of the target UAV in step S4 exceeds a set range.

[0012] The present invention also provides a multi-station UAV detection system based on dynamic grouping, comprising multiple UAV detection devices and a controller communicatively connected to the detection devices, the controller comprising: The data acquisition module is used to dynamically acquire the location information of multiple detection devices deployed in the monitoring area, as well as the detection data of each detection device on the target UAV, as the feature information of the detection device, and transmit the feature information to the grouping module through a data connection; The grouping module, which is connected to the data acquisition module, is used to receive the feature information and divide all the detection devices into several device groups based on the similarity of the feature information of each detection device, and transmit the grouping results to the quality assessment module through the data connection. The quality assessment module is connected to the grouping module and the data acquisition module. It is used to receive grouping results and subsequent detection data, compare the subsequent detection data of each detection device in the same device group, determine whether each detection device is in a normal or abnormal state, and transmit the list of detection devices in a normal state to the fusion positioning module through the data connection. The fusion positioning module is connected to the quality assessment module and the data acquisition module, and is used to receive the list of normal state detection devices and their subsequent detection data and perform fusion positioning of the target UAV.

[0013] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of a multi-station UAV detection method based on dynamic grouping.

[0014] Compared with the prior art, the present invention has the following beneficial effects: By employing a dynamic grouping mechanism based on feature information similarity, the detection devices are divided into several groups according to the actual location of the target UAV and the similarity of the detection data. This ensures that the detection devices in the same group have similar communication environments and data characteristics, enabling more accurate identification of abnormal devices during subsequent quality assessment. This solves the misjudgment problem caused by including devices with large differences in communication environments in opposite directions in the global comparison in the gridded scheme, significantly improving the accuracy of detection quality judgment. Compared with the scheme of comparing all detection devices pairwise, data comparison is only performed within the same group of devices, reducing the computational complexity and the total amount of computation significantly, thus improving the real-time performance and computational efficiency of the system.

[0015] By adopting a dynamic mechanism that combines timed acquisition and triggered acquisition, when the number of abnormal detection devices exceeds a set threshold or the fusion positioning deviation exceeds a set threshold, the system automatically triggers the reacquisition of feature information and regrouping. This establishes a dynamic feedback mechanism that can respond promptly to changes in the target UAV's position or sudden environmental changes, enabling adaptive tracking of the target UAV's position and motion state. This overcomes the shortcomings of fixed grouping in distributed schemes, which cannot adapt to dynamic changes in the target, and ensures the continuous accuracy and robustness of the detection system. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating a multi-station UAV detection method based on dynamic grouping.

[0017] Figure 2 This is a flowchart illustrating sub-steps S31 to S34 in step S3.

[0018] Figure 3 This is a block diagram of the controller module of a multi-station UAV detection system based on dynamic grouping. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0020] Figure 1 A flowchart of a multi-station UAV detection method based on dynamic grouping provided in an embodiment of the present invention is shown below. Figure 1 As shown, the method includes step S1, dynamically acquiring the location information of multiple detection devices deployed in the monitoring area, as well as the detection data of each detection device on the target UAV, as the feature information of each detection device.

[0021] Specifically, each detection device has a built-in GPS / BeiDou positioning module, which automatically acquires its own latitude and longitude coordinates and altitude after deployment and reports them to the controller via a communication network. In another embodiment, the location information of each detection device is manually measured and entered into the system database during installation, and the location information of each device is read from the database during initialization. Preferably, the location information of each detection device is stored in the form of a geographic coordinate system (longitude, latitude, altitude) or a local coordinate system (x, y, z).

[0022] Each detection device continuously detects target UAVs within the monitoring area at a preset sampling frequency, acquiring at least one of the following detection data indicators: angle of arrival, azimuth, pitch, received signal strength, signal-to-noise ratio (SNR), distance, and radial velocity. The angle of arrival reflects the direction from which the target UAV signal reaches the detection device; the azimuth represents the target UAV's angle in the horizontal plane; and the SNR reflects the quality of the received signal. These parameters comprehensively reflect the spatial relationship and signal propagation characteristics between the detection device and the target UAV, providing a data foundation for accurate grouping and quality assessment.

[0023] The location information of each detection device is combined with real-time detection data to form the feature information of each detection device. Specifically, the feature information of the i-th detection device can be represented as Fi={Pi,Di(t)}, where Pi=(xi,yi,zi) are the three-dimensional position coordinates of the i-th detection device, and Di(t) is the detection data vector acquired by the i-th detection device at time t, which includes one or more of the following indicators: angle of arrival θi, azimuth ϕi, elevation angle φi, received signal strength RSSIi, signal-to-noise ratio SNRi, distance ri, and radial velocity vi.

[0024] The dynamic acquisition mechanism enables real-time reflection of the target UAV's position changes. In some embodiments of the present invention, the dynamic acquisition in step S1 is timed acquisition, where data acquisition commands are sent to each detection device according to a preset time period. Each detection device responds to the command and reports its current detection data and position information. The time period for timed acquisition can be dynamically adjusted according to the target UAV's movement speed. When the target UAV is moving at high speed, the time period is shortened to improve the positioning refresh rate; when the target UAV is moving at low speed or hovering, the time period is appropriately extended to reduce system overhead.

[0025] The method embodiment of the present invention further includes step S2, which divides all the detection devices into several device groups based on the similarity of the feature information of each detection device.

[0026] The first step in this step is to normalize the feature information of the detection device, such as through Z-score normalization. Then, similarity is calculated, which can be done using Euclidean distance, cosine similarity, or Mahalanobis distance. Preferably, in some embodiments, Mahalanobis distance is used to calculate similarity. Mahalanobis distance considers the correlation between the dimensions of the feature vectors and is defined as... Where Σ is the covariance matrix of the eigenvectors of all detection devices. Mahalanobis distance can effectively eliminate the influence of different dimensions of different detection indicators, and is especially suitable for scenarios where the detection data contains multiple different types of indicators.

[0027] After calculating the pairwise similarity between all detection devices, a clustering algorithm is used to divide all detection devices into several device groups. In some embodiments of the present invention, the K-means clustering algorithm is used to divide N detection devices into K device groups (K is a preset value), thereby maximizing the similarity of devices within the same group and minimizing the similarity of devices between different groups. The specific steps are as follows: Step S21: Initialize K cluster centers C1, C2, ..., CK. This can be done by randomly selecting the feature vectors of K detection devices as the initial cluster centers, or by using the K-means++ algorithm for optimization and initialization. Step S22: Assign each detection device to the group to which the nearest cluster center belongs, and use the similarity measurement method selected in step S2 to measure the distance. Step S23: Recalculate the cluster center of each group, which is the mean of the feature vectors of all detection devices in the group; Step S24: Repeat steps S22 and S23 until the cluster centers no longer change or the preset maximum number of iterations (e.g., 100 times) is reached.

[0028] In one embodiment, the K value is dynamically determined based on the total number of detection devices in the monitoring area. For example, K = ⌈N / 5⌉, meaning that each group contains approximately 5 detection devices, to ensure that there are enough devices in each group for subsequent consistency assessment.

[0029] In some embodiments of the present invention, the DBSCAN clustering algorithm is used for grouping, and the specific steps are as follows: Step S21: Set two parameters: neighborhood radius ε and minimum number of points MinPts; Step S22: For each detection device, calculate the number of detection devices contained in its ε neighborhood and the number of detection devices whose distance is not greater than the neighborhood radius ε. The distance measurement adopts the similarity measurement method selected in step S2. Step S23: If the number of detection devices in the ε neighborhood of a certain detection device is greater than or equal to MinPts, then the detection device is marked as the core point, and a cluster (device group) is formed by expanding with the detection device as the core. Step S24: Group the detection devices that are directly density-reachable and indirectly density-reachable into the same cluster; the detection devices that are directly density-reachable refer to the detection devices within the neighborhood radius ε of the core point, and the detection devices that are indirectly density-reachable refer to all the detection devices within the ε neighborhood of the core point when a device within the neighborhood radius ε is also the core point.

[0030] In step S25, devices not assigned to any cluster are marked as noise points and either grouped separately or handled manually.

[0031] The advantage of the DBSCAN algorithm is that it does not require a preset number of groups and can automatically identify and remove isolated detection devices that are too different from other devices, thus avoiding the forced grouping of abnormal devices into a certain group and affecting the accuracy of the consistency assessment within the group.

[0032] In some embodiments of the present invention, a hierarchical clustering algorithm is employed to achieve grouping by constructing a tree-like hierarchical clustering structure. The specific steps are as follows: Step S21: Initialize each detection device into an independent group; Step S22: Calculate the distance (similarity) between all pairs of groups, and combine the two closest pairs into a new group; Step S23: Update the distance between the new group and other groups; Step S24: Repeat steps S22 and S23 until the preset number of groups is reached.

[0033] After clustering is completed, each detection device is uniquely assigned to a device group.

[0034] This invention dynamically divides all detection devices into several groups based on the similarity of their feature information. This ensures that detection devices within the same group have similar spatial positions and communication environments relative to the target UAV, making their detection data comparable. Compared to existing grid-based schemes that include all detection devices in a horizontal comparison for consistency evaluation, this invention avoids the unreasonable problem of forcibly including detection devices located on the opposite edge of the grid and with significantly different communication environments from the target UAV in the same comparison system, effectively reducing the misjudgment rate of abnormal devices.

[0035] The method embodiment of the present invention further includes step S3, which compares the subsequent detection data of the detection devices within the same device group to determine whether each detection device is in a normal or abnormal state. By performing a consistency comparison of detection data only within the group, rather than performing a global pairwise comparison of all detection devices, the computational load is significantly reduced. Especially in application scenarios with a large number of detection devices, the present invention can significantly shorten data processing time and improve the system's real-time response capability.

[0036] like Figure 2 As shown, step S3 specifically includes sub-steps S31 to S34.

[0037] S31, acquire subsequent detection data from the detection devices within the group and perform data normalization preprocessing. For each detection device group obtained in step S2, acquire subsequent detection data of all detection devices within the group for the same target UAV. Since the detection data acquired by different detection devices within the same device group may have different dimensions and orders of magnitude (e.g., angle of arrival in degrees, received signal strength in dBm), it is necessary to perform normalization preprocessing on different types of data to make them comparable. In this embodiment of the invention, the Z-score normalization method is used for normalization.

[0038] S32, calculate the deviation of each detection index of the detection data of each detection device within the same device group from the group's consistency benchmark. In this invention, the group's consistency benchmark includes a numerical benchmark and / or a change benchmark, and the deviation includes the deviation of each detection index of the detection data of the detection device relative to the group's consistency benchmark.

[0039] Numerical benchmarks are used to measure the consistency of the detection data of various detection devices within a group at the same time, mainly addressing issues such as signal attenuation caused by detection device gain deviation and obstruction. In one embodiment, the numerical benchmark includes the median benchmark of the same detection index of all detection devices in the group. Taking the angle of arrival as an example, suppose there are n devices in the group, and their measured angles of arrival are θ1, θ2, ..., θn. Then the corresponding median benchmark is the median(θ) of the angles of arrival of the n detection devices. The deviation of detection device i from the median benchmark in the angle of arrival dimension is dnum,i = |θi−median(θ)|. When the deviation exceeds the corresponding preset deviation threshold (e.g., 3σ, where σ is the standard deviation of the angle of arrival), it is determined that the angle of arrival of the detection device deviates from the median benchmark.

[0040] In another embodiment, to enhance robustness, the deviation of the detection index x of detection device i from the median benchmark is calculated as dnum,i = |xi−median(x)| / MAD(x), where the median absolute deviation MAD(x) = median(|xi−median(x)|). The median and MAD are highly robust to outliers, ensuring that even if a few outlier devices exist within the group, the benchmark value will not be significantly skewed.

[0041] In another embodiment, the numerical benchmark includes a weighted average benchmark of the same detection index across all detection devices within the group. The weights are determined based on the signal-to-noise ratio (SNR) of the current detection data from each detection device; devices with higher SNR have greater weights. Taking the received signal strength index as an example, the corresponding weighted average benchmark... Among them, weight That is, the weight is proportional to the signal-to-noise ratio. When the deviation of a detection indicator of a certain detection device from the corresponding weighted average benchmark exceeds the corresponding preset deviation threshold, it is determined that the detection indicator of that detection device deviates from the weighted average benchmark.

[0042] The variation benchmark is used to measure whether the rate and direction of change of the detection data of the detection devices within the group are consistent over time. When the target UAV is maneuvering, the signal strength or angle change rate received by the detection devices within the group should have a similar increasing or decreasing trend. In one embodiment, the variation benchmark includes the first-order difference median benchmark of the same detection index of all detection devices within the group at adjacent sampling times. The first-order difference of the detection index x of detection device i at the current time t is Δxi(t)=xi(t)−xi(t−1), the first-order difference median benchmark is Δmedian(x(t))=median(Δx1(t),Δx2(t),...,Δxn(t)), and the deviation is dgrad,i=|Δxi(t)−Δmedian(x(t))|. When the deviation exceeds the corresponding preset deviation threshold, it is determined that the detection index of the detection device deviates from the first-order difference median benchmark.

[0043] This invention comprehensively evaluates the status of detection equipment using both numerical and change benchmarks. The numerical benchmark measures the consistency of the absolute values ​​detected by each device within the group at the same time, effectively identifying signal anomalies caused by device gain deviations, obstructions, etc. The change benchmark measures the consistency of the rate and direction of change of the detected data of devices within the group over time, effectively identifying abnormal devices that fail to correctly track target changes during target maneuvers. The two dimensions complement each other and cross-validate each other, reducing the false positive and false negative rates of single-dimensional detection.

[0044] S33, based on the comparison result between the deviation amount and the preset deviation threshold, determine whether each of the detection devices is in a normal state or an abnormal state. When one or more deviation amounts exceed the preset deviation threshold range, the device is determined to be in an abnormal state.

[0045] In one embodiment, if more than half of the detection indicators of a certain detection device deviate from the corresponding value or change benchmark, the detection device is determined to be in an abnormal state; if only a single detection indicator is abnormal, it is marked as a suspicious state, isolated only for this fusion positioning, and continuously observed in subsequent detection times; if all detection indicators are normal, it is determined to be in a normal state.

[0046] In another embodiment, a continuous anomaly accumulation strategy is adopted. Only when a detection device is determined to be abnormal in multiple consecutive detection times (e.g., 3 or 5) will it be finally marked as an abnormal state. If it is only abnormal in a single cycle, it will not be determined as an abnormal state for the time being, but will be marked as a suspicious state. Only the current fusion positioning is isolated to avoid misjudgment caused by instantaneous interference.

[0047] In some embodiments of the present invention, the preset deviation threshold in step S34 is dynamically adjusted based on the number of detection devices in the group and / or the motion state of the target UAV. When the number of detection devices in the group is small (e.g., less than 3), the deviation threshold is appropriately widened to avoid misjudging normal fluctuations as abnormalities. When the number of detection devices in the group is large, the deviation threshold can be appropriately tightened. When the target UAV is maneuvering violently, the fluctuation of the detection data itself increases, and the deviation threshold is appropriately widened; when the target UAV is flying steadily, the deviation threshold is appropriately tightened. The motion state of the target UAV can be determined by the historical flight path obtained by fusion positioning in step S4. For example, the average speed and acceleration of the target at the most recent 5 positioning points are calculated. If the speed exceeds 30 m / s or the acceleration exceeds 5 m / s², it is determined to be a "violent maneuvering" state.

[0048] For each detection device in each device group, a status flag is output. The detection data of the detection device in normal state can be used for subsequent fusion positioning; the detection data of the detection device in abnormal state cannot be used for subsequent fusion positioning; the detection device in suspicious state is temporarily excluded from the current fusion positioning and will be continuously monitored in subsequent cycles.

[0049] In some embodiments of the present invention, the dynamic acquisition in step S1 also includes a triggered acquisition mode. When the number of detection devices determined to be in an abnormal state in step S3 is higher than a set threshold (for example, the number of abnormal devices exceeds 30% of the total number of devices in the group or exceeds 3 devices), it indicates that the current grouping may no longer be suitable for the actual position changes of the target UAV, and it is necessary to reacquire the feature information of the detection devices and regroup them.

[0050] The method embodiment of the present invention further includes step S4, which uses the detection data of the detection device determined to be in a normal state to perform fusion positioning of the target UAV.

[0051] In some embodiments, the present invention employs cross-positioning based on angle of arrival. Cross-positioning is used when at least two operational detection devices can provide angle of arrival measurements. Specifically, starting from the position of each operational detection device, rays are drawn along the direction of their measured angle of arrival; the intersection of these rays represents the estimated position of the target UAV. In practical applications, due to measurement errors, the rays typically do not intersect precisely at a single point, but rather form a convergence area. The optimal position estimate is then solved using the least squares method or weighted least squares method.

[0052] In some embodiments, the present invention employs positioning based on received signal strength. When the detection device in normal state can provide a measurement of the received signal strength, a positioning method based on a signal propagation model is used. According to a signal propagation loss model (such as a free space propagation model or a logarithmic distance path loss model), the received signal strength is converted into a distance estimate between the detection device and the target UAV, and then the target position is solved using trilateration or multilateral measurement methods.

[0053] In some embodiments, the present invention employs multi-source fusion positioning, which comprehensively utilizes various detection data such as angle of arrival and received signal strength, and uses nonlinear filtering methods such as extended Kalman filtering, unscented Kalman filtering, or particle filtering to perform multi-source fusion positioning, thereby improving positioning accuracy and robustness.

[0054] In one embodiment, step S4 also calculates the difference between the current positioning result and the historical positioning result (positioning deviation). When the positioning deviation exceeds a set threshold, it indicates that the positioning result has changed abnormally, indicating that there may be a problem with the data of the detection device currently used for positioning. At this time, the trigger acquisition mode of step S1 is triggered to reacquire the detection data of each detection device and regroup and reassess the quality.

[0055] The present invention also provides a multi-station UAV detection system based on dynamic grouping, including multiple UAV detection devices and a controller that is communicatively connected to the detection devices.

[0056] like Figure 3As shown, the controller includes the following modules: The data acquisition module 201 is used to dynamically acquire the location information of multiple detection devices deployed within the monitoring area, as well as the detection data of each detection device on the target UAV, as feature information of the detection devices, and transmit the feature information to the grouping module 202 via a data connection. The data acquisition module 201 performs data collection operations according to a preset time period (timed acquisition) or when a trigger condition is met (triggered acquisition).

[0057] The grouping module 202, connected to the data acquisition module 201, receives feature information and divides all detection devices into several groups based on the similarity of their feature information. The grouping results are then transmitted to the quality assessment module 203 via the data connection. The grouping module uses K-means clustering, DBSCAN clustering, or hierarchical clustering algorithms to perform dynamic grouping.

[0058] The quality assessment module 203, data-connected to the grouping module 202 and the data acquisition module 201, receives grouping results and subsequent detection data. It compares the subsequent detection data of each detection device within the same device group to determine whether each device is in a normal or abnormal state. The list of detection devices in a normal state is then transmitted to the fusion positioning module 204 via the data connection. The quality assessment module performs the operations described in step S3, including data normalization, deviation calculation, and state determination.

[0059] The fusion positioning module 204, connected to the quality assessment module 203 and the data acquisition module 201, receives the list of normal-state detection devices and their detection data, performs fusion positioning of the target UAV, outputs the fusion positioning result to the trigger control module 205, and can be presented to the user through a display terminal or other output device. The fusion positioning module 204 uses angle-of-arrival cross-positioning, time-of-arrival difference positioning, received signal strength positioning, or a combination of the above methods for fusion positioning.

[0060] The trigger control module 205 is data-connected to the quality assessment module 203, the fusion positioning module 204, the data acquisition module 201, and the grouping module 202. It is used to monitor the number of abnormal state detection devices and the fusion positioning deviation. When the number of abnormal state detection devices exceeds a set threshold or the fusion positioning deviation exceeds a set threshold, the trigger control module 205 sends a trigger signal to the data acquisition module 201 and the grouping module 202 via the data connection, triggering the reacquisition of feature information and regrouping, thus achieving dynamic feedback adjustment.

[0061] The system works as follows: the data acquisition module 201 collects the location information and detection data of the detection devices, constructs feature vectors, and transmits them to the grouping module 202; the grouping module 202 executes a clustering algorithm, calculates the similarity, and outputs the grouping results to the quality assessment module 203; the quality assessment module 203 compares and analyzes the detection data of the same group of detection devices, determines the status of the detection devices, and transmits the list of normal detection devices to the fusion positioning module 204; the fusion positioning module 204 acquires the detection data of the normal detection devices, executes a weighted fusion algorithm, calculates and outputs the fused position.

[0062] The trigger control module 205 monitors the number of abnormal devices and the deviation of the fused positioning in real time. When the monitored value exceeds the threshold, it sends a trigger signal to the data acquisition module 201 and the grouping module 202 to start the regrouping process and form a closed-loop control. The entire system realizes fully automated processing from data acquisition, dynamic grouping, quality assessment to fused positioning, and can respond to changes in the target UAV's position in real time.

[0063] In one embodiment, the controller is a server deployed at a ground command center, including a processor, memory, storage devices, and communication interfaces. The data acquisition module, grouping module, quality assessment module, and fusion positioning module are software modules running on the processor, implementing their respective functions through software programs.

[0064] In another embodiment, the controller is a cloud server, and each detection device is connected to the cloud server through a 4G / 5G communication network to upload the detection data to the cloud in real time for centralized processing.

[0065] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the above-described multi-station UAV detection method based on dynamic grouping.

[0066] Computer-readable storage media can be any tangible medium capable of storing program code, including but not limited to: USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0067] It is understood that the detection data in the above embodiments is not limited to angle of arrival, azimuth, pitch, signal-to-noise ratio, received signal strength (RSSI), distance, and radial velocity. It may also include parameters such as Doppler shift, as long as they reflect the spatial relationship and signal propagation characteristics between the detection device and the target UAV. The clustering algorithm is not limited to K-means, DBSCAN, or hierarchical clustering algorithms; spectral clustering algorithms, mean-shift algorithms, etc., may also be used, as long as they can group devices based on feature similarity.

[0068] Clearly, similarity metrics are not limited to Euclidean distance, cosine similarity, and Mahalanobis distance; Manhattan distance, Chebyshev distance, etc., can also be used. The appropriate metric should be chosen based on the dimension of the feature vector and the characteristics of the data distribution. Similarly, fusion localization algorithms are not limited to weighted least squares or Kalman filtering; the choice can be made based on the target motion model and system complexity.

[0069] Understandably, for devices deemed suspicious, instead of complete exclusion, a weighting strategy can be adopted, such as reducing their fusion weight to 50% of that of normal devices. This utilizes some of their useful information while reducing their adverse impact on the fusion result. Triggering conditions are not limited to the number of abnormal devices and the fusion positioning deviation, but can also include sudden changes in the target UAV's speed, fault detection of the detection equipment itself, sudden changes in environmental interference intensity, etc.

[0070] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A multi-station UAV detection method based on dynamic grouping, characterized in that, include: Step S1: Dynamically acquire the location information of multiple detection devices deployed in the monitoring area, as well as the detection data of each detection device on the target UAV, as the feature information of each detection device; Step S2: Based on the similarity of the feature information of each of the detection devices, all the detection devices are divided into several device groups; Step S3: Compare the subsequent detection data of the detection devices in the same device group to determine whether each detection device is in a normal or abnormal state. Step S4: Use the detection data of the detection device that is determined to be in normal condition to perform fusion positioning of the target UAV.

2. The multi-station UAV detection method based on dynamic grouping according to claim 1, characterized in that, Step S3 includes: S31, For each of the detection devices in the same device group, acquire their subsequent detection data of the same target UAV, and perform normalization preprocessing on the subsequent detection data; S32, calculate the deviation of the detection data of each detection device in the same device group from the consistency benchmark within the group; S33, based on the comparison result between the deviation amount and the preset deviation threshold, determine whether each of the detection devices is in a normal state or an abnormal state.

3. The multi-station UAV detection method based on dynamic grouping according to claim 2, characterized in that, In step S32, the intra-group consistency benchmark includes a numerical benchmark and / or a change benchmark, and the deviation includes the deviation of each detection index in the detection data of the detection device from the intra-group consistency benchmark. When one or more deviations exceed the preset deviation threshold range, the device is determined to be in an abnormal state.

4. The multi-station UAV detection method based on dynamic grouping according to claim 3, characterized in that, The numerical benchmark includes the median and / or weighted mean of each detection index of all detection data from the detection devices within the group; the variation benchmark includes the first-order difference median of each detection index of all detection data from the detection devices within the group.

5. The multi-station UAV detection method based on dynamic grouping according to claim 3, characterized in that, The preset deviation threshold in step S33 is dynamically adjusted based on the number of detection devices in the group and / or the motion state of the target UAV. When the number of detection devices in the group is less than the set value or the target UAV is maneuvering violently, the preset deviation threshold is relaxed.

6. The multi-station UAV detection method based on dynamic grouping according to any one of claims 1-5, characterized in that, The detection indicators include at least one of the following: angle of arrival, azimuth, elevation, received signal strength, signal-to-noise ratio, distance, and radial velocity.

7. The multi-station UAV detection method based on dynamic grouping according to any one of claims 1-5, characterized in that, The similarity of the feature information of the detection devices is calculated using Euclidean distance, cosine similarity, or Mahalanobis distance.

8. The multi-station UAV detection method based on dynamic grouping according to any one of claims 1-5, characterized in that, The dynamic acquisition in step S1 includes timed acquisition and triggered acquisition. The triggering conditions for the triggered acquisition include: the number of detection devices determined to be in an abnormal state in step S3 is higher than a set number threshold, and / or the difference between the fused positioning and historical positioning of the target UAV in step S4 exceeds a set range.

9. A multi-station UAV detection system based on dynamic grouping, characterized in that, It includes multiple drone detection devices and a controller communicatively connected to the detection devices, the controller comprising: The data acquisition module is used to dynamically acquire the location information of multiple detection devices deployed in the monitoring area, as well as the detection data of each detection device on the target UAV, as the feature information of the detection device, and transmit the feature information to the grouping module through a data connection; The grouping module, which is connected to the data acquisition module, is used to receive the feature information and divide all the detection devices into several device groups based on the similarity of the feature information of each detection device, and transmit the grouping results to the quality assessment module through the data connection. The quality assessment module is connected to the grouping module and the data acquisition module. It is used to receive grouping results and subsequent detection data, compare the subsequent detection data of each detection device in the same device group, determine whether each detection device is in a normal or abnormal state, and transmit the list of detection devices in a normal state to the fusion positioning module through the data connection. The fusion positioning module is connected to the quality assessment module and the data acquisition module, and is used to receive the list of normal state detection devices and their subsequent detection data and perform fusion positioning of the target UAV.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the multi-station UAV detection method based on dynamic grouping as described in any one of claims 1 to 8.

Images