Multi-means fusion unmanned aerial vehicle high-precision detection method and system

Abstract

This invention discloses a high-precision detection method and system for unmanned aerial vehicles (UAVs) using a multi-method fusion approach, relating to the field of electronic digital data processing technology. The method includes: acquiring the trace data of the target UAV at various detection time points within the current detection period to generate initial position coordinates; acquiring the altitude data of the target UAV at various detection time points within the current detection period to obtain positioning coordinates to be processed; obtaining the compensation ratio of the target UAV based on real-time wind speed and direction data, and obtaining various corrected position coordinates based on the compensation ratios of the target UAV in each axis and the positioning coordinates to be processed; acquiring the UAV's aircraft type feature vector within the current detection period, matching the aircraft type feature vector with a pre-stored feature fingerprint database to generate an aircraft type confirmation signal, and generating an alarm signal containing the aircraft type confirmation signal based on multiple corrected position coordinates within the current detection period. This invention has the advantages of data collaborative fusion, anti-interference, and high detection accuracy.

Description

Technical Field

[0001] This invention relates to the field of electronic digital data processing technology, specifically to a high-precision detection method and system for unmanned aerial vehicles that integrates multiple methods. Background Technology

[0002] In the field of low-altitude UAV detection, existing radar equipment only provides distance, azimuth, and elevation information in polar coordinates when outputting target point data, failing to directly generate three-dimensional spatial coordinates. Furthermore, it lacks co-correction with other sensor data, leading to positioning drift under complex weather conditions (such as strong winds). Simultaneously, existing detection methods relying on single sensors (such as only spectral or acoustic features) for aircraft identification suffer from unstable feature extraction and high misjudgment rates due to the randomness of UAV frequency hopping behavior and environmental interference from rotor noise (especially when the similarity threshold is below 90%). Existing technologies lack a multi-modal feature joint verification mechanism. Moreover, while 5G-A sensing base stations can provide altitude data and three-dimensional motion vectors, they are not fused with radar planar coordinates in real time, nor are weather compensation mechanisms (such as the influence of wind speed and direction on the axial offset of the flight trajectory) introduced, resulting in accumulated trajectory reconstruction errors. Especially when the target suddenly maneuvers, existing methods, due to insufficient sampling frequency or the lack of a trajectory change rate assessment model, cannot trigger timely tiered alarms.

[0003] In summary, the current solution suffers from a gap in data collaboration, resulting in a decrease in the accuracy of the final target positioning and making it difficult to meet the demand for high-precision real-time monitoring of sensitive areas by UAVs. Summary of the Invention

[0004] To address the technical problems mentioned above, this invention provides a high-precision detection method and system for unmanned aerial vehicles (UAVs) that integrates multiple methods.

[0005] A high-precision detection method for unmanned aerial vehicles (UAVs) using a multi-method fusion approach includes: acquiring point data of the target UAV at various detection time points within the current detection period using radar equipment, converting the point data from a polar coordinate system to a three-dimensional coordinate system, and generating initial position coordinates; acquiring altitude data of the target UAV at various detection time points within the current detection period using a sensor-integrated base station, and obtaining the corresponding positioning coordinates to be processed based on the altitude data and initial position coordinates at the same detection time point; acquiring real-time wind speed and direction data based on meteorological information, obtaining the compensation ratio of the target UAV in each axis based on the real-time wind speed and direction data, and obtaining corrected position coordinates based on the compensation ratio of the target UAV in each axis and the positioning coordinates to be processed; acquiring the UAV's model feature vector within the current detection period, matching the model feature vector with a pre-stored feature fingerprint database to generate a model confirmation signal, and generating an alarm signal containing the model confirmation signal based on multiple corrected position coordinates within the current detection period.

[0006] Optionally, obtaining the positioning coordinates to be processed at a given time point based on the height data and initial position coordinates at the same time point includes: decomposing the initial position coordinates output by the radar device at the i-th detection time point into the i-th planar coordinates; extracting the height data output by the integrated sensing base station at the i-th detection time point; and combining the i-th planar coordinates and the height data output at the i-th detection time point to generate the positioning coordinates to be processed.

[0007] Optionally, obtaining the compensation ratio of the target UAV in each axis based on real-time wind speed and wind direction data includes: obtaining the angle between the wind direction and the radar equipment beam based on the wind direction data; and obtaining the compensation ratio of the target UAV in each axis based on the beam angle.

[0008] Optionally, obtaining the corrected position coordinates based on the compensation ratio of the target UAV in each axis and each positioning coordinate to be processed includes: obtaining the actual adjustment amount of each axis based on the adjustable amount and the compensation ratio of each axis; and obtaining the corrected position coordinates based on the actual adjustment amount of each axis and the positioning coordinates to be processed.

[0009] Optionally, obtaining the aircraft type feature vector of the target UAV in the current detection period includes: obtaining the frequency hopping period feature output in the current detection period based on the spectrum detection device; obtaining the rotor main frequency feature value output in the current detection period based on the acoustic sensor array; and integrating the frequency hopping period feature and the rotor main frequency feature value into the aircraft type feature vector.

[0010] Optionally, generating an alarm signal containing an aircraft type confirmation signal based on multiple corrected position coordinates within the current detection period includes: obtaining the trajectory mutation rate of the current detection period based on multiple corrected position coordinates within the current detection period; if the trajectory mutation rate of the current detection period exceeds a preset threshold, obtaining an alarm level based on the trajectory mutation rate, and integrating the aircraft type confirmation signal and the alarm level into an alarm signal.

[0011] A high-precision UAV detection system integrating multiple methods is also provided, comprising: an acquisition module for acquiring point data of the target UAV at various detection time points within the current detection period based on radar equipment, converting the point data from polar coordinates to a three-dimensional coordinate system and generating initial position coordinates; a fusion module for acquiring altitude data of the target UAV at various detection time points within the current detection period based on a sensor-integrated base station, and acquiring the corresponding positioning coordinates to be processed at the same time point based on the altitude data and initial position coordinates; a correction module for acquiring real-time wind speed and direction data based on meteorological information, acquiring the compensation ratio of the target UAV in each axis based on the real-time wind speed and direction data, and acquiring corrected position coordinates based on the compensation ratio of the target UAV in each axis and the positioning coordinates to be processed; and a detection data processing module for acquiring the UAV model feature vector within the current detection period, matching the model feature vector with a pre-stored feature fingerprint database to generate a model confirmation signal, and generating an alarm signal containing the model confirmation signal based on multiple corrected position coordinates within the current detection period.

[0012] Optionally, the fusion module is also used to: decompose the initial position coordinates output by the radar device at the i-th detection time point into the i-th planar coordinates; extract the height data output by the integrated sensing base station at the i-th detection time point; and combine the i-th planar coordinates and the height data output at the i-th detection time point to generate the positioning coordinates to be processed.

[0013] Optionally, the correction module is also used to: obtain the angle between the wind direction and the radar equipment beam based on the wind direction data; and obtain the compensation ratio of the target UAV in each axis based on the beam angle.

[0014] Optionally, the correction module is also used to: obtain the actual adjustment amount of each axis based on the adjustable amount and the compensation ratio of each axis; and obtain the corrected position coordinates based on the actual adjustment amount of each axis and the positioning coordinates to be processed.

[0015] The beneficial effects of this invention are reflected in:

[0016] In the multi-method fusion method for high-precision UAV detection, firstly, after establishing an initial three-dimensional reference framework using radar polar coordinate data conversion, the altitude component calculated by radar elevation angle is replaced by independent altitude measurements from a sensing base station, which to some extent solves the problem of altitude error accumulation in the vertical direction caused by multipath effects or long-range detection. Furthermore, the three-dimensional axial compensation ratio is calculated by combining real-time meteorological data, and dynamic corrections are applied to the horizontal and vertical positions based on wind direction differences, significantly reducing positioning drift caused by strong wind interference. Further, by fusing spectral frequency hopping periodic features and acoustic rotor main frequency features to form a composite aircraft feature vector, the limitation of single sensor features being susceptible to environmental interference is overcome. Combined with a two-layer matching mechanism of photoelectric contour verification, misjudgment of aircraft type caused by frequency hopping randomness or noise pollution is effectively suppressed. Finally, the trajectory mutation rate is calculated based on the high-precision position sequence after meteorological compensation, and the target maneuvering is identified by combining multi-dimensional motion state, realizing a quantitative correlation between alarm level and threat behavior. Attached Figure Description

[0017] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.

[0018] Figure 1 This is a schematic diagram illustrating the steps of the multi-method fusion UAV high-precision detection system of the present invention;

[0019] Figure 2 This is a schematic diagram showing the conversion of point data in S1 from polar coordinates to three-dimensional coordinates in the multi-method fusion UAV high-precision detection system of the present invention.

[0020] Figure 3 This is a schematic diagram of part of step S2 in the multi-means fusion UAV high-precision detection system of the present invention;

[0021] Figure 4 This is a schematic diagram of part of step S3 in the multi-means fusion UAV high-precision detection system of the present invention;

[0022] Figure 5 This is a schematic diagram of another part of the steps in S3 of the multi-means fusion UAV high-precision detection system of the present invention;

[0023] Figure 6 This is a schematic diagram of part of step S4 in the multi-means fusion UAV high-precision detection system of the present invention;

[0024] Figure 7 This is a schematic diagram of another part of the steps in S4 of the multi-means fusion UAV high-precision detection system of the present invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0026] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0027] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, the terms "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0028] like Figure 1 As shown, a high-precision detection method for unmanned aerial vehicles (UAVs) using a multi-method fusion approach is provided. In one embodiment, the method includes:

[0029] S1. Based on radar equipment, acquire the point data of the target UAV at each detection time point in the current detection cycle, convert each point data from the polar coordinate system to the three-dimensional coordinate system, and generate each initial position coordinate.

[0030] S2. Based on the integrated sensor base station, obtain the altitude data of the target UAV at each detection time point within the current detection cycle, and obtain the corresponding positioning coordinates to be processed at the same detection time point based on the altitude data and initial position coordinates at the same detection time point.

[0031] S3. Obtain real-time wind speed and direction data based on meteorological information, and obtain the compensation ratio of the target UAV in each axis based on the real-time wind speed and direction data, and obtain the corrected position coordinates based on the compensation ratio of the target UAV in each axis and each positioning coordinate to be processed.

[0032] S4. Obtain the aircraft type feature vector of the target UAV within the current detection period, match the aircraft type feature vector with the pre-stored feature fingerprint database and generate an aircraft type confirmation signal, and generate an alarm signal containing the aircraft type confirmation signal based on multiple corrected position coordinates within the current detection period.

[0033] In this embodiment, it should be noted that in S1, a pre-deployed dedicated radar device is first used. This device operates continuously within the set current detection cycle for the target UAV, capturing its reflected signals at various discrete detection time points. The radar device outputs the raw form of sensing information—point data. Its core parameters are the distance value of the target UAV relative to the radar station, the azimuth angle value on the horizontal plane, and the elevation angle value in the vertical direction. These data are essentially expressed based on the polar coordinate system of the radar beam itself.

[0034] Furthermore, to enable multi-source data fusion, the polar coordinate point data obtained at each detection time point, existing in a ternary form (range, azimuth, elevation), are uniformly mapped to a standard three-dimensional Cartesian coordinate system. This transformation process applies to each independent detection time point within the current detection cycle. The result is a preliminary coordinate description of the target UAV's specific location in three-dimensional space at that moment, generating a corresponding three-dimensional spatial coordinate point, which is defined as the initial position coordinates at each detection time point. This step transforms the low-level information acquired by the single radar sensing device into the starting point for the spatial position representation required for subsequent processing, and creates the possible preconditions for spatiotemporal alignment and fusion with additional information from the sensing base station.

[0035] Furthermore, the three-dimensional initial position coordinates established by S1 form the basis for the entire subsequent high-precision positioning. These initial position coordinates, converted from radar points, have planar components (usually understood as X-axis and Y-axis coordinates) derived from radar measurements of distance and azimuth, while their Z-axis coordinate component primarily originates from elevation measurements. This data is essentially based on relative position positioning from a single radar sensing source, and does not yet incorporate information such as absolute altitude, motion vectors, and the effects of weather disturbances. Therefore, its accuracy is affected by radar measurement errors and environmental factors, especially under complex weather conditions or strong reflection backgrounds.

[0036] Furthermore, its Z-axis altitude information mainly relies on elevation angle measurements for estimation. At long distances or when the target is at a low elevation angle, the altitude estimation accuracy is often poor and susceptible to multipath effects. Therefore, the initial position coordinates output by S1 are essentially a raw positioning starting point containing errors, especially potentially inaccurate altitude information. It provides a spatial reference and timestamp sequence for the entire process, but the data itself has not undergone any compensation or enhancement processing. The improvement of its accuracy depends on the data introduced in subsequent steps and the compensation mechanism.

[0037] Furthermore, for example, the current detection cycle is set to 10 seconds, with a detection time point every 0.5 seconds, for a total of 21 time points. The radar equipment captures the trace data of a target UAV at time point t5 (2.5 seconds after detection begins), such as... Figure 2 As shown, this data is expressed in polar coordinates: distance from radar station 2500 meters (r=2500m), azimuth angle relative to true north 35° (θ=35°), elevation angle 15° (φ=15°). This point data is a description of the target's original position in three-dimensional space relative to the radar.

[0038] Then, in order to coordinate spatially with data from other sensors, S1 transforms (2500, 35°, 15°) into a Cartesian coordinate system according to a preset transformation relationship. For example... Figure 2 As shown, firstly, the planar displacement component of the target relative to the radar is calculated on the horizontal plane based on the distance and azimuth angle; secondly, its height component is calculated using the elevation angle and distance; finally, the initial position coordinates of the target at time point t5 are obtained, for example, it may be (X≈1788.5m, Y≈1467.8m, Z≈647.4m), where the Z value (height) is entirely dependent on the radar elevation angle measurement and trigonometric function calculation. It is worth noting that the accuracy of the radar elevation angle is easily affected by the target distance and environmental factors. For example, at the detection time point t15 (the distance may be farther or the elevation angle lower), the radar measured an elevation angle of only 8°, and the point data was (3000m, 40°, 8°), which yielded an initial height Z≈418.4m.

[0039] However, this calculated altitude may be lower than the actual flight altitude due to radar angular deviations caused by multipath effects or changes in air refractive index. Step S1 performs this point acquisition and coordinate transformation operation at each sampling time (e.g., t1, t2, t3...t21) within the current detection cycle, generating an initial position coordinate sequence of 21 position points, providing a basic spatial position framework and timestamp reference for the entire detection process, but its altitude information contains potential sources of measurement error.

[0040] In S2, higher-precision and independently sourced altitude information is introduced to improve the vertical positioning accuracy of the target location. This step relies on a pre-deployed 5G-A sensing base station, which has independent sensing capabilities and can synchronously acquire accurate altitude measurements of the same target UAV at discrete detection time points within the same detection cycle (typically derived from base station signal reflection or calculations based on specific sensing patterns).

[0041] Specifically, for the same detection time point i, the three-dimensional initial position coordinates generated by S1 for that time point are first decomposed to extract their two-dimensional coordinate information on the ground plane or a specific reference plane, i.e., the i-th plane coordinate (this usually includes the values ​​of the X and Y axes, representing the horizontal position of the target, and this position information is relatively more reliable). Then, the target height data (i.e., the Z-axis value) captured at the exact same time point i is extracted from the sensing data of the integrated sensing base station. Finally, the decomposed two-dimensional plane coordinates (Xi, Yi) are directly combined and paired with the independent height measurement value (Zi) provided by the sensing base station.

[0042] Specifically, this process essentially replaces the relatively inaccurate Z-axis component in the initial position coordinates of S1, which is calculated solely based on the radar elevation angle, with height measurements from the sensing base station. This generates a new three-dimensional spatial point information—the i-th unprocessed positioning coordinate (Xi, Yi, Zi'), where Zi' directly originates from the sensing base station. This step fully leverages the high-precision sensing advantage of the integrated sensing base station in the vertical direction, directly enhancing and correcting the main weaknesses in the height dimension of the initial radar positioning.

[0043] Furthermore, step S2 significantly improves the reliability and accuracy of the position coordinates in the vertical dimension by introducing the altitude data from the sensing base station, thus solving the core limitation in S1 where the initial position coordinate Z-axis information (altitude) is susceptible to interference from radar elevation angle measurement errors, distance effects, and multipath effects. It provides a more balanced and accurate set of basic coordinate points in three-dimensional space—the positioning coordinates to be processed—for subsequent position compensation and trajectory generation.

[0044] Furthermore, this coordinate system is not the final corrected result, as it only addresses the bottleneck of vertical accuracy but does not consider another crucial external interference factor: the axial offset effect of weather conditions on the actual flight trajectory of the UAV (especially strong wind disturbances). The horizontal component (X, Y) of this positioning coordinate system still originates solely from the initial transformation result of radar measurements. Although the planar position accuracy is relatively small compared to the height error problem, in real-world scenarios, wind speed and direction continuously exert dynamic, directional thrust or drag on the UAV's movement direction in three-dimensional space (including horizontal movement and vertical speed). This wind-induced displacement cannot be effectively identified and compensated for in raw position measurements relying solely on radar or base stations.

[0045] Therefore, although the height data of the generated positioning coordinates has been enhanced by high-quality synsensory measurements, its overall positional information (especially its planar position) still implicitly contains inherent errors caused by wind forces that have not been corrected in the current step. To address this new challenge, meteorological information was specifically introduced in subsequent step S3, and a dynamic correction mechanism based on axial compensation ratio was designed to further optimize the positional accuracy of the positioning coordinates, thereby more realistically reflecting the target's three-dimensional spatial position under the influence of wind disturbance.

[0046] Furthermore, for example, at the same time point t5 in S1, the 5G-A integrated sensing base station obtains the actual height measurement of the target at that moment through its sensing capabilities. For example, the base station obtains a precise height value of 732.6m through signal delay difference or Doppler effect analysis. S2 first extracts the initial position coordinates generated by S1 for this point, such as (1788.5m, 1467.8m, 647.4m), decomposes them, retains its planar coordinate information (1788.5m, 1467.8m), and discards its radar-calculated Z component value (647.4m); then it loads the height data of 732.6m measured by the sensing base station at the exact same detection time point t5; this height data is combined with the aforementioned planar coordinates to generate a new three-dimensional coordinate point at time point t5, i.e., the positioning coordinates to be processed (1788.5m, 1467.8m, 732.6m).

[0047] Here we can see that the horizontal component (X, Y) of this point is inherited from the initial radar coordinates of S1, but the altitude value (Z) has been replaced by the precise value independently measured by the sensing base station. The same operation is performed for other times within the detection period: for example, at time t15, although the initial coordinate altitude displayed by S1 is approximately 418.4m, the actual altitude measured by the base station is 612.2m. Therefore, the horizontal component of the coordinates to be processed remains unchanged, but the altitude is updated to the actual value measured by the base station. S2 outputs a sequence of the positioning coordinates to be processed at all time points within the current detection period. This sequence improves the accuracy of the original radar data in the vertical direction (Z-axis), solving the problem of inaccurate altitude calculation at long distances or low elevation angles. However, the horizontal component of the coordinates to be processed is still based solely on the radar's measurement and transformation results, without considering horizontal offsets caused by external meteorological factors.

[0048] In S3, a key environmental interference factor in low-altitude UAV detection is addressed—real-time weather conditions, particularly the dynamic and directional offsets caused by wind speed and direction on the UAV's flight trajectory. These offsets can lead to errors in position coordinates relying solely on raw sensor measurements (such as radar and sensor base stations). This step first accesses the real-time weather information stream, continuously acquiring the current environmental wind speed (representing wind strength) and wind direction (representing the primary direction of wind action). Instead of simply compensating for fixed position coordinates, it analyzes the relationship between the direction of wind action and the possible direction of displacement of the UAV based on physical mechanisms.

[0049] Specifically, it first calculates the spatial angle between the current wind direction and the main beam direction of the deployed radar equipment. This angle is a key input for understanding how wind force is decomposed in the UAV's three-dimensional motion coordinate system (usually defined as the X, Y, and Z axes). Based on this beam angle, step S3 further derives the relative weights of the compensation required for the target UAV along each independent coordinate axis (X, Y, and Z axes), i.e., the compensation ratio. This ratio is direction-dependent and axis-differentiated, reflecting the intensity of the push / pull effect of wind force on the UAV in a specific direction (e.g., tailwinds or headwinds mainly affect X-axis movement, crosswinds mainly affect Y-axis movement, while vertical gusts affect Z-axis altitude). Obtaining this compensation ratio is the core calculation step of S3, providing a differential adjustment basis for the axial dimensions for subsequent precise position fine-tuning.

[0050] Furthermore, after obtaining the compensation ratios for each axis reflecting the intensity of the three-dimensional directional influence of wind, step S3 then performs fine-tuning on the "positioning coordinates to be processed" at each detection time point generated and input by S2. This correction process first requires a "total adjustable amount" (usually positively correlated with wind speed) characterizing the expected deviation of the UAV's position due to the overall wind intensity. Then, the total adjustable amount is multiplied by the specific compensation ratio calculated on each of the X, Y, and Z axes, thus obtaining the specific position adjustment range applicable to each coordinate axis—that is, the actual adjustment amount for each axis (for example, if the compensation ratio for the X axis is large, it indicates that the wind force is strong in that direction, and its allocated actual adjustment amount is also correspondingly larger).

[0051] Finally, step S3 uses the actual adjustment amounts for each axis as fine-tuning components, directly superimposing or applying them to the corresponding axis's positioning coordinate values ​​to be processed. This superposition operation, based on the direction of the coordinate axes and the push / pull effect of wind (increase or decrease), ultimately generates the "corrected position coordinates" for each detection time point. This step profoundly reflects the method's adaptability to complex environmental dynamics: it not only identifies the general interference of weather on flight but also precisely quantifies and decomposes the differential effects of this interference in different directions in three-dimensional space, transforming this quantitative analysis into specific position correction values ​​for each spatial dimension.

[0052] Therefore, compared to the "position coordinates to be processed" output by S2 (altitude enhanced but wind disturbance not compensated), the "corrected position coordinates" output by S3 have received further physical-law-compliant compensation and optimization in terms of horizontal movement (affected by crosswinds and headwinds) and altitude stability (affected by vertical wind shear). This lays a high-precision position data foundation for the final generation of a high-fidelity flight trajectory (the basis for calculating trajectory change rate in S4) and the triggering of accurate alarms.

[0053] Further, for example, current environmental monitoring data shows a wind speed of 6 m / s and a northeast wind direction (angle of 60°, with due north as the baseline 0°). The radar equipment is deployed with its main beam due north (0°). First, the spatial angle between this wind direction (60°) and the radar beam direction (0°) is calculated to be 60°. Then, based on this angle, the compensation ratio distribution along the three-dimensional axes (X, Y, Z) is derived: the wind force component is stronger on the X-axis (due east direction), so a higher compensation ratio is assigned, for example, Kx=0.45; the component is weaker on the Y-axis (due north direction), so the ratio is correspondingly lower, for example, Ky=0.30; on the Z-axis (height direction), there may be updrafts or vortex effects, so the ratio is set to Kz=0.25. Subsequently, based on the wind speed (6 m / s), an overall adjustable amount is calculated (this value is related to the wind speed), for example, this total amount is set as Δ_total=1.8 (unit represents the offset estimate). The actual adjustment amounts for each axis are obtained by multiplying the total adjustable amount by the corresponding axial ratio: X-axis adjustment Δ_X = 1.8 * 0.45 ≈ 0.81; Y-axis adjustment Δ_Y = 1.8 * 0.30 ≈ 0.54; Z-axis adjustment Δ_Z = 1.8 * 0.25 ≈ 0.45. Finally, considering the wind direction is northeast, a thrust is applied to the target in the X and Y axes, and the above positive adjustment amounts are superimposed on the unprocessed positioning coordinates generated by S2 at time t5: X_c = 1788.5m + 0.81m ≈ 1789.31m; Y_c = 1467.8m + 0.54m ≈ 1468.34m; Z_c = 732.6m + 0.45m ≈ 733.05m (assuming a slight updraft effect), finally generating the corrected position coordinates of the target at that moment. This compensation calculation is performed at every moment within the detection period (including when there are other wind direction and speed combinations), and a high-precision three-dimensional coordinate sequence with axial correction for meteorological disturbances is output. For example, the coordinates at time point t15 also undergo similar compensation processing, thereby obtaining a more physically accurate representation in terms of the plane movement and height stability of the trajectory.

[0054] In S4, the inherent correlation of multimodal sensing data is used to extract the most model-distinguishing features from two independent physical domains—radio frequency spectrum and acoustic noise.

[0055] Specifically, it uses pre-deployed spectrum detection equipment to capture and analyze the communication signal patterns of the target UAV throughout the current detection period, extracting key radio frequency fingerprint features, particularly the frequency hopping cycle characteristics of its frequency hopping behavior (i.e., the time pattern of signal switching between multiple frequencies). Simultaneously, it collects unique noise signals generated by the target UAV's rotor using a deployed acoustic sensor array and calculates and extracts its core rotor frequency characteristic value (an inherent acoustic feature closely related to the number, rotational speed, and structure of rotor blades). Then, it integrates these two heterogeneous features—the frequency hopping cycle characteristic and the rotor frequency characteristic value—acquired within the same detection period to form a comprehensive descriptor representing the target UAV's identity, namely, the aircraft type feature vector. This feature vector converges the target's "fingerprint" information at both the electromagnetic spectrum and physical acoustic levels.

[0056] Next, the real-time generated aircraft model feature vector is matched with a pre-established and stored feature fingerprint database (containing standard feature vectors of various known drone models) for refined matching calculation. Only when the matching similarity reaches a preset high confidence standard (≥90%), a preliminary aircraft model identification result is output. To further eliminate false positives and improve credibility, this preliminary identification result is not the final conclusion and needs to be visually verified by the real-time drone image contour captured by the electro-optical tracking device: only when the contour matching degree also reaches a preset standard (≥85%), a final aircraft model confirmation signal with high confidence is output.

[0057] Furthermore, while obtaining a high-confidence aircraft type confirmation signal, step S4 needs to simultaneously utilize the high-precision sequence position information output by the aforementioned steps (especially S3), which has undergone weather compensation—that is, the corrected position coordinates of each discrete detection time point within the current detection cycle—to perform flight behavior analysis, focusing on detecting whether the target has any sudden maneuvers that may threaten its safety.

[0058] Specifically, its core behavior recognition mechanism calculates the trajectory mutation rate during the current detection cycle. This rate is obtained by analyzing the changing trends of multiple consecutive corrected position coordinates in three-dimensional space to obtain the subsequent predicted trajectory. This trajectory is then compared with a predetermined trajectory, and the difference is divided by the upper limit of trajectory change to obtain the trajectory mutation rate, thus quantitatively assessing the intensity of the target's movement within the detection cycle. Once the current trajectory mutation rate exceeds a preset safety threshold, it is determined that the target is engaging in high-risk or unintended abnormal maneuvering behavior. Depending on the degree to which the trajectory mutation rate exceeds the threshold (e.g., slight exceedance, significant exceedance, or severe exceedance), different alarm levels are calculated, quantifying the potential threat level of the maneuvering behavior. Finally, step S4 organically integrates these two core outputs—the aircraft type confirmation signal representing the target's identity and the alarm level representing the current dynamic threat level—to form a comprehensive and complete alarm signal. This alarm signal not only informs the target's aircraft type but also clearly indicates the target's actions and provides the anomaly level, meeting the real-time monitoring needs of sensitive areas for comprehensive and graded responses to intrusion targets.

[0059] Furthermore, for example, within the current detection period (10 seconds, 21 sampling points), the spectrum detection device continuously monitors the target signal, captures its characteristic frequency hopping behavior pattern, and extracts the frequency hopping period as 120ms. The acoustic sensor array collects the target rotor noise spectrum within the same period, identifying the dominant frequency characteristic value as 215Hz. This radio frequency hopping period characteristic (120ms) is combined with the acoustic rotor dominant frequency characteristic (215Hz) to construct a feature vector (e.g., represented as [120, 215]). This vector is matched against a pre-stored fingerprint database: the similarity between vector [120, 215] and the pre-stored standard vector [122, 216] for model C in the database is calculated to be 92% (exceeding the 90% threshold), and the preliminary identification result is output as model C.

[0060] Subsequently, the photoelectric tracking device captures the target's image contour at key moments. The feature matching algorithm calculates that this contour matches the standard model contour of Model C with an 87% match rate (exceeding the 85% threshold). After verification, the final model confirmation signal is output as Model C. Simultaneously, using the weather-compensated corrected position coordinate sequence (21 points) output by S3, S44 calculates the trajectory abrupt change rate for the entire detection cycle. By analyzing indicators such as the rate of change of direction and velocity between consecutive position points, it is found that in the final stage (t17 to t21), the target performs a sharp turning maneuver and climb, resulting in a calculated abrupt change rate of 14%. The preset safety threshold (set according to the scenario) is assumed to be 8%, therefore S45 determines that this abnormal maneuver exceeds the limit, and further classifies the alarm level as a medium threat based on the degree of exceedance (14% - 8% = 6 percentage points).

[0061] Ultimately, S4 integrates the model confirmation signal (Model C) with the alarm level (intermediate threat) to generate a structured alarm signal output containing identity information and dynamic risk assessment results. This provides a reliable basis for target identification and quantitative classification information of behavioral risks for monitoring, completing the entire closed-loop processing from multi-source data acquisition, high-precision location estimation, reliable model identification to intelligent behavior assessment.

[0062] In summary, in this multi-method fusion approach for high-precision UAV detection, firstly, after establishing an initial three-dimensional reference framework using radar polar coordinate data conversion, the altitude component calculated from the radar elevation angle is replaced by independent altitude measurements from a sensing base station, which to some extent solves the problem of accumulated altitude errors in the vertical direction caused by multipath effects or long-range detection. Furthermore, by combining real-time meteorological data to calculate the three-dimensional axial compensation ratio, dynamic corrections are applied to the horizontal and vertical positions based on wind direction differences, significantly reducing positioning drift caused by strong wind interference. Further, by fusing spectral frequency hopping periodic features with acoustic rotor frequency characteristics to form a composite aircraft feature vector, the limitations of single sensor features being susceptible to environmental interference are overcome. Combined with a two-layer matching mechanism of photoelectric contour verification, misjudgment of aircraft type caused by frequency hopping randomness or noise pollution is effectively suppressed. Finally, based on the high-precision position sequence after meteorological compensation, the trajectory mutation rate is calculated, and by combining multi-dimensional motion state identification of target maneuvers, a quantitative correlation between alarm levels and threat behaviors is achieved.

[0063] like Figure 3 As shown, in one embodiment, S2, obtaining the positioning coordinates to be processed at the same time point based on the height data and initial position coordinates at the same time point includes:

[0064] S21. Decompose the initial position coordinates output by the radar equipment at the i-th detection time point into the i-th planar coordinates;

[0065] S22. Extract the altitude data output by the integrated sensing base station at the i-th detection time point;

[0066] S23. Combine the i-th planar coordinates and the height data output at the i-th detection time point to generate the positioning coordinates to be processed.

[0067] In this embodiment, it should be noted that in S21, after obtaining the initial position coordinates generated by S1 at the i-th detection time point (these coordinates have been converted from the radar polar coordinate system to a three-dimensional rectangular coordinate system), step S21 performs a spatial decomposition operation to extract specific information. Specifically, this involves analyzing the three-dimensional coordinates of the point and separating its components on the ground projection plane, i.e., the i-th plane coordinates. These plane coordinates are typically composed of the X-axis and Y-axis values ​​from the initial position coordinates, representing the target's horizontal position relative to the reference point at that time point. The purpose of this decomposition is to retain the relatively reliable parts of the radar's planar positioning (X and Y components), while preparing for the subsequent replacement of the Z-axis component (height), as the radar-calculated height is considered insufficiently accurate.

[0068] Further, for example, at the 5th detection time point (t=2.5 seconds), S1 outputs the initial position coordinates (1788.5m, 1467.8m, 647.4m). Step S21 analyzes these three-dimensional coordinates, separating their horizontal components: the X-axis value of 1788.5m represents the target's eastward offset, and the Y-axis value of 1467.8m represents the northward offset (using the radar station as the origin to construct a northeast-northeast coordinate system). By retaining these two-dimensional plane coordinates (1788.5, 1467.8) and discarding the altitude component 647.4m, the plane coordinates at the 5th time point are formed. This operation focuses on the radar's advantageous positioning dimensions, avoiding the altitude error calculated from its elevation angle, and preparing for the introduction of base station altitude data. In this example, the plane coordinates maintain the geometric relationship of the original radar measurement, ensuring the continuity of the horizontal position reference.

[0069] In step S22, crucial altitude information is collected from an independent 5G-A integrated sensing base station. Its input is the identifier i of the same detection time point used in step S1. The core operation is to query the sensing data output by the base station at that specific time i and extract the altitude data value (i.e., Z-axis position) of the target UAV, obtained directly from the base station's measurement or calculation. This altitude data originates from sensing mechanisms such as base station signal propagation time, phase difference, or multi-base station cooperative positioning, and possesses higher accuracy and anti-interference capabilities than radar elevation angle estimation.

[0070] Furthermore, for example, at the same time point i=5, the integrated sensing base station measures the target height based on the signal propagation time difference. The timestamp of the base station receiving the target's reflected signal is T+12.6ms. Combining this with the electromagnetic wave propagation speed, the straight-line distance is calculated to be 2532m. Then, using the geometric relationship of an elevation angle of 45°, the absolute height value of 732.6m is calculated (the formula is omitted, the result is directly quoted). Step S22 retrieves the precise height data at this moment from the base station database. This value is independent of the radar elevation angle measurement, avoiding multipath interference. The example shows that the base station height is 85.2m higher than the radar-estimated value, reflecting the difference in height measurement between the two technologies, thus confirming the necessity of correcting the height data in S1.

[0071] In step S23, the i-th planar coordinate (Xi, Yi) output from S21 and the i-th time-point base station height data (Zi) output from S22 are received. This step directly combines or "assembles" these two sets of data to generate a new three-dimensional spatial position coordinate, namely the i-th positioning coordinate to be processed (Xi, Yi, Zi). The combination process essentially replaces the radar-calculated Z component value in the initial position coordinate of S1 with a high-precision independent height measurement value Zi from the base station. The resulting new coordinate point still has a horizontal position (X, Y) based on the original radar planar data, but its vertical position (Z) has been enhanced by higher-precision base station data, thereby optimizing the vertical dimension accuracy of the entire position.

[0072] Further, for example, consider the planar coordinates (1788.5, 1467.8) output by S21 and the height of 732.6m output by S22. Through direct assignment, the base station height data is bound to the Z-axis component of the planar coordinates, generating a new 3D point (1788.5, 1467.8, 732.6) as the positioning coordinates to be processed. This operation achieves physical data replacement rather than mathematical transformation, preserving the integrity of the radar's horizontal positioning. The same operation is performed at t=7.5 seconds (i=15): the initial coordinates of S1 (2189.3m, 1876.2m, 418.4m) are decomposed to obtain planar coordinates (2189.3, 1876.2), and the base station height of 612.2m is used to generate (2189.3, 1876.2, 612.2). Data from both times verify the time synchronization of the S2 mechanism throughout the entire cycle, resolving the issue of a height error of 193.8m at low radar elevation angles (8°).

[0073] like Figure 4 As shown, in one embodiment, S3, obtaining the compensation ratio of the target UAV in each axis based on real-time wind speed and direction data includes:

[0074] S31. Obtain the angle between the wind direction and the radar equipment beam based on the wind direction data;

[0075] S32. Obtain the compensation ratio of the target UAV in each axis based on the beam angle.

[0076] In this embodiment, it should be noted that in S31, meteorological information and equipment deployment orientation knowledge are incorporated for spatial geometric analysis. The core input is real-time acquired environmental wind direction data (usually an angle value, such as 0 degrees for true north, increasing clockwise). The operation involves calculating the three-dimensional spatial angle between this real-time wind direction vector and the transmission direction of the radar equipment's main detection beam (also a vector direction). This angle value quantifies the degree of deviation of the wind's direction of action relative to the radar observation axis (usually the radar coordinate system reference).

[0077] Further, for example, consider real-time meteorological data: wind speed 6 m / s, wind direction 60° (due north reference). The radar deployment azimuth is 0° (due north) for the main beam. Through vector geometry calculation, the spatial angle θ between the wind direction vector (cos60°, sin60°) and the radar beam vector (1, 0) is 60°. This angle represents the deviation of the wind's direction from the observation reference. The anemometer data is updated every 500 ms at t = 2.5 seconds. The example shows that when the wind direction changes to 120° (southeast), the angle value is updated synchronously to 120°, ensuring real-time performance in dynamic environments. The angle calculation does not involve UAV parameters, relying only on equipment deployment and meteorological data to maintain model versatility.

[0078] In S32, the weights of the potential impact of wind on the UAV's flight trajectory in different spatial directions (typically defined as the X, Y, and Z axes) are determined. The internal logic analyzes the mapping components or projection characteristics of this included angle vector along preset three-dimensional axes (such as X, Y, and Z). The core calculation outputs the required "compensation ratio" values ​​for a target UAV in each coordinate axis (X, Y, Z). These ratio values ​​are three independent values ​​within a certain range (summed to a common value), representing the relative offset influence of wind disturbance on the target in the X-axis movement, Y-axis movement, and Z-axis elevation / reflection directions, respectively.

[0079] Furthermore, for example, based on an included angle θ=60°, the three-dimensional influence weights are decomposed according to a preset projection rule: the cosine component of the X-axis (eastward) is cos60°=0.5, the sine component of the Y-axis (northward) is sin60°≈0.866, and the empirical coefficient of the Z-axis (vertical) is 0.2 (considering vertical turbulence). After normalization: the total component sum is 0.5+0.866+0.2=1.566, then K_x=0.5 / 1.566≈0.319, K_y=0.866 / 1.566≈0.553, K_z=0.2 / 1.566≈0.128. This ratio reflects that the northeasterly wind mainly affects the Y-axis direction. At t=7.5 seconds and the wind direction is 120°, K_x increases to 0.553, and K_y decreases to 0.319, proving that the axial weights dynamically adjust with the wind direction.

[0080] like Figure 5 As shown, in one embodiment, S3, obtaining the corrected position coordinates based on the compensation ratio of the target UAV in each axis and each positioning coordinate to be processed includes:

[0081] S33. Obtain the actual adjustment amount for each axis based on the adjustable amount and the compensation ratio for each axis.

[0082] S34. Obtain the corrected position coordinates based on the actual adjustment amount of each axis and the positioning coordinates to be processed.

[0083] In this embodiment, it should be noted that in S33, the total wind force influence intensity is distributed to different directions in three-dimensional space. The adjustable value input is a total parameter value related to real-time wind speed data. It represents the estimated total displacement amplitude that the target location is expected to compensate for under the current wind speed level (generally, the higher the wind speed, the larger this value). The specific value depends on the pre-built parameter mapping table (aircraft wind resistance library, environmental calibration table, and local historical library). Specifically, when determining the adjustable value, it is based on the wind speed value, and then based on the historical average position drift of a finite number of times under the wind speed value in the local historical library, the average position drift is used as the adjustable value. For example, if the measured wind speed is 8 m / s and the historical average position drift is 1.2 meters, then the adjustable value is taken as 1.2 meters.

[0084] Specifically, based on the three axial compensation ratios (e.g., Px, Py, Pz) obtained from S32, independent adjustment components corresponding to each axis (X, Y, Z) are calculated: the adjustable amount is multiplied by the X-axis compensation ratio Px to obtain the actual adjustment amount (ΔX) in the X-axis, multiplied by the Y-axis compensation ratio Py to obtain the actual adjustment amount (ΔY) in the Y-axis, and multiplied by the Z-axis compensation ratio Pz to obtain the actual adjustment amount (ΔZ) in the Z-axis. The calculation results ΔX, ΔY, and ΔZ are the specific fine-tuning values ​​that will be applied to the positioning coordinates to be processed in their respective axes, with the direction (positive or negative) determined by the direction of wind action.

[0085] Furthermore, for example, the adjustable amount Δ_total is 1.8m based on "measured wind speed 6m / s → historical offset table" (6m / s corresponds to an average offset of 1.2-2.4m, taking the median value). Combining this with the S32 ratio: Δ_x = 1.8 × 0.319 ≈ 0.574m, Δ_y = 1.8 × 0.553 ≈ 0.995m, Δ_z = 1.8 × 0.128 ≈ 0.230m. Parameter source explanation: The historical database stores the average offset of 100 drone sets under the same wind speed over the past 30 days. If the wind speed increases to 8m / s, the adjustable amount increases to 2.4m, reflecting the dynamic correlation of the parameters. The example shows the largest adjustment on the Y-axis, consistent with the strong influence of northeasterly winds on northward movement.

[0086] In step S34, the system receives the location coordinates (Xi, Yi, Zi) to be processed at the i-th time point from step S2 and the actual adjustment amounts (ΔX, ΔY, ΔZ) of the X, Y, and Z axes at that time point calculated from step S33. The operation is clear and direct: the axial adjustment amounts are superimposed on the corresponding components of the location coordinates to be processed. That is, the corrected X coordinate = original X coordinate to be processed + ΔX, the corrected Y coordinate = original Y coordinate to be processed + ΔY, and the corrected Z coordinate = original Z coordinate to be processed + ΔZ. This generates the final corrected position coordinates (X_c, Y_c, Z_c) at the i-th time point. This coordinate point integrates radar planar information, the precise altitude of the base station, and dynamic compensation for meteorological wind disturbance in three dimensions. It is a high-precision position measurement value after data fusion, providing a reliable spatial data foundation for subsequent trajectory analysis and alarm generation.

[0087] Further, for example, the coordinates to be processed (1788.5, 1467.8, 732.6) and adjustment amounts (0.574, 0.995, 0.230) are received. Based on the direction of the northeasterly wind, the coordinate overlay rules are: X_c = 1788.5 + 0.574 = 1789.074m (eastward shift due to easterly wind), Y_c = 1467.8 + 0.995 = 1468.795m (northward shift due to northerly wind), Z_c = 732.6 + 0.230 = 732.830m (fine-tuning of vertical airflow). Corrected coordinates (1789.074, 1468.795, 732.830) are generated. Comparing the coordinates to be processed in S2, it can be seen that the maximum horizontal adjustment is 1.0m, verifying the optimization effect of meteorological compensation on planar positioning. Similarly, at t=7.5 seconds, the output is (2191.253, 1877.219, 612.430), demonstrating consistency throughout the entire cycle.

[0088] like Figure 6 As shown, in one embodiment, obtaining the target UAV's model feature vector within the current detection period in S4 includes:

[0089] S41. Obtain the frequency hopping period characteristics of the output within the current detection period based on the spectrum detection device;

[0090] S42. Obtain the rotor main frequency characteristic value output within the current detection period based on the acoustic sensor array;

[0091] S43. Integrate the frequency hopping cycle characteristics and rotor main frequency characteristics into an aircraft feature vector.

[0092] In this embodiment, it should be noted that step S41 focuses on extracting key identification features from the target UAV's radio communication signals. This operation relies on a spectrum detection device continuously capturing the radio frequency signals emitted by the target during the current detection period. The core processing involves analyzing the captured signal sequence, identifying and quantifying the temporal pattern of the signal's periodic switching between multiple operating frequencies, i.e., accurately extracting the target UAV's frequency-hopping periodic characteristics. This periodic characteristic represents its fixed frequency-hopping behavior pattern and is one of the important radio frequency fingerprints.

[0093] For example, a particular model of drone may switch between several specific frequencies in an orderly manner at a stable rhythm. This rhythm (cycle) is the key feature that distinguishes it from other models. Different designs and communication protocols will lead to different frequency hopping behavior patterns.

[0094] Furthermore, as an example, the spectrum device captures the target communication signal within a 10-second period and identifies 6 frequency hopping events: a timestamp sequence [1.2s, 3.4s, 5.1s, 6.9s, 8.3s, 9.7s].

[0095] First, the adjacent hopping intervals are calculated: [2.2s, 1.7s, 1.8s, 1.4s, 1.4s]. After excluding outliers at the beginning and end, the average value of 1.72s is taken. The final frequency hopping period characteristic is 1720ms (rounded to the nearest integer). An example demonstrating actual data processing shows that 3.4s - 1.2s = 2.2s is the initial interval. Due to potentially high channel contention, it is filtered by median filtering to retain valid data. This value reflects the temporal regularity of the target communication protocol, providing a feature anchor for subsequent identification.

[0096] In S42, the focus is on the acoustic characteristics generated by the physical structure of the target UAV. The operational basis is the sound signals collected by an acoustic sensor array, generated by the high-speed rotation of the UAV's rotor (or multi-rotor) cutting through the air. The core processing involves spectral analysis and feature identification of these sound signals, with the goal of accurately extracting the dominant fundamental frequency in the noise signal, i.e., the rotor's main frequency characteristic value. This frequency value is strongly correlated with structural parameters such as the rotor's physical dimensions, number of blades, and motor speed, and is the "acoustic signature" of the target UAV's power.

[0097] For example, the differences in rotor size combinations and typical rotational speeds between quadcopters and hexcopters are directly reflected in the different main frequency values, becoming an important basis for identification.

[0098] Furthermore, for example, an acoustic array acquires 10 seconds of audio. FFT spectral analysis reveals a fundamental frequency peak of 215Hz lasting 8 seconds (80% of the period), with a brief harmonic at 430Hz (blade resonance). The main processor filters transient noise using a power spectral density threshold, determining the fundamental frequency characteristic value to be 215Hz. In this example, this frequency matches the typical rotational speed of a quadcopter at 12800rpm (calculated: 12800rpm / 60 = 213.3Hz). This verifies the strong correlation between acoustic characteristics and physical structure; environmental wind noise interference is suppressed by a 10-200Hz bandpass filter.

[0099] S43 is the key operation point for achieving multimodal feature fusion. It receives the frequency hopping periodic features (characterizing electromagnetic domain behavior) output from S41 and the rotor dominant frequency feature value (characterizing physical domain structure) output from S42. The core operation is to organically integrate these two complementary heterogeneous feature parameters from different physical domains, combining them into a comprehensive feature description—the aircraft feature vector. This feature vector is a structured data object (such as a vector containing two elements) that unifies the unique communication behavior pattern of the target object with its physical characteristics of propulsion within a mathematical expression framework.

[0100] Furthermore, for example, the frequency hopping period of 1720ms output from S41 is integrated with the rotor main frequency of 215Hz output from S42. The data structure is defined according to the protocol: Feature Vector = [Communication Feature, Acoustic Feature] = [1720, 215]. Numerical units are standardized: period is converted to milliseconds, and frequency is preserved as an integer Hertz. This vector serves as the target fingerprint input to the recognition module. The example demonstrates that differences in dimensions do not affect the matching calculation (data homogenization within the database).

[0101] like Figure 7 As shown, in one embodiment, S4 generates an alarm signal containing an aircraft type confirmation signal based on multiple corrected position coordinates within the current detection period, including:

[0102] S44. Obtain the trajectory mutation rate of the current detection period based on multiple corrected position coordinates within the current detection period;

[0103] S45. If the trajectory mutation rate of the current detection cycle exceeds the preset threshold, the alarm level is obtained based on the trajectory mutation rate, and the model confirmation signal and the alarm level are integrated into an alarm signal.

[0104] In this embodiment, it should be noted that in S44, the changing trends of multiple consecutive corrected position coordinates in three-dimensional space are analyzed to obtain the subsequent predicted trajectory. This trajectory is then compared with the predetermined trajectory, and the difference is divided by the upper limit of trajectory change (determined according to the aircraft model) to obtain the trajectory abrupt change rate. This allows for a quantitative assessment of the target's motion intensity within the detection period. A high trajectory abrupt change rate directly indicates that the target is currently performing a high-speed maneuver with unclear intentions or potential threats.

[0105] Furthermore, for example, input the S3 corrected coordinate sequence (21 points). Calculate the position difference: from point t20 (1988.7, 1323.4, 645.2) to point t21 (2021.3, 1298.6, 662.8), the three-dimensional displacement Δs = 40.5m, and the time difference 0.5 seconds. The instantaneous speed is 81 m / s. Compared to the average speed of 25 m / s in the preceding phase, the speed abrupt change rate is 224%. The azimuth angle changes abruptly from 56° to 312°, with an angular velocity of 512° / s. The overall weighted trajectory abrupt change rate is 14%. The example shows that the final abrupt change rate is calculated by fusing 10 dynamic indicators, focusing on capturing terminal maneuvering behavior. The preset safety threshold of 8% is derived from historical safe flight data (the maximum abrupt change rate for takeoff and landing according to civil aviation standards is 7.2%).

[0106] In S45, the final aircraft type confirmation signal (containing a high-confidence aircraft type identification result verified by multiple methods including radio frequency, acoustic and photoelectric verification) confirmed in the aircraft type identification process of S4 is received, along with the current trajectory mutation rate calculated in S44 and its comparison with a preset threshold. The preset threshold is determined based on historical safe flight data (such as standard civil aviation takeoffs and landings, and regular routes for logistics drones). It constructs a distribution model of the trajectory mutation rate under normal flight conditions by statistically analyzing the position change rate (acceleration, turning angular velocity, and climb rate) of a limited number of legal flight trajectories. The preset threshold is set as the upper limit of the confidence interval (e.g., the 99th percentile) of this model.

[0107] First, it is determined whether the trajectory mutation rate exceeds a preset threshold. Once it is confirmed that it exceeds the threshold, the corresponding alarm level is determined based on the specific magnitude of the exceedance (e.g., exceeding by less than 1 time, more than 1 time but less than 2 times, or more than 2 times). This quantifies the potential risk level (e.g., "alert," "high threat," "serious intrusion," etc.) posed by the identified abnormal maneuvering behavior. Finally, the aircraft type confirmation signal representing the target's identity and the alarm level representing the current dynamic threat status are logically combined or packaged to generate a comprehensive alarm signal that is complete in information, clearly structured, and can guide subsequent response decisions.

[0108] Furthermore, for example, consider the input model confirmation signal (Model C) and a trajectory mutation rate of 14% (exceeding the threshold of 8%). Based on the degree of exceedance, a 6% exceedance is classified as a moderate threat (rule: 0-3% low, 3-6% medium, >6% high). The alarm level is set to 2 levels (0 no risk, 1 low, 2 medium, 3 high). Thus, the final alarm signal is encapsulated as {ID:C, AlertLv:2, Pos:[2021.3, 1298.6, 662.8]}. This example demonstrates multi-dimensional information integration: a 92% confidence level for model identification and 87% for contour matching ensure identity reliability; trajectory mutation quantification supports risk classification; and subsequent responses can be based on this to initiate corresponding countermeasures.

[0109] A high-precision UAV detection system integrating multiple methods is also provided, the system including:

[0110] The acquisition module is used to acquire the point data of the target UAV at each detection time point in the current detection cycle based on the radar equipment, convert each point data from the polar coordinate system to the three-dimensional coordinate system and generate each initial position coordinate;

[0111] The fusion module is used to acquire the altitude data of the target UAV at each detection time point within the current detection cycle based on the integrated sensing base station, and to acquire the corresponding positioning coordinates to be processed at the same detection time point based on the altitude data and the initial position coordinates at the same detection time point.

[0112] The correction module is used to obtain real-time wind speed and direction data based on meteorological information, obtain the compensation ratio of the target UAV in each axis based on the real-time wind speed and direction data, and obtain the coordinates of each correction position based on the compensation ratio of the target UAV in each axis and each positioning coordinate to be processed.

[0113] The detection data processing module is used to acquire the aircraft type feature vector of the target UAV within the current detection period, match the aircraft type feature vector with the pre-stored feature fingerprint database and generate an aircraft type confirmation signal, and generate an alarm signal containing the aircraft type confirmation signal based on multiple corrected position coordinates within the current detection period.

[0114] In one embodiment, the fusion module is further configured to: decompose the initial position coordinates output by the radar device at the i-th detection time point into the i-th planar coordinates; extract the height data output by the integrated sensing base station at the i-th detection time point; and combine the i-th planar coordinates and the height data output at the i-th detection time point to generate the positioning coordinates to be processed.

[0115] In one implementation, the correction module is further configured to: obtain the angle between the wind direction and the radar equipment beam based on wind direction data; and obtain the compensation ratio of the target UAV in each axis based on the beam angle.

[0116] In one embodiment, the correction module is further configured to: obtain the compensation ratio of each axis of the positioning coordinate to be processed based on the beam angle; obtain the adjustable amount and the compensation ratio of each axis to obtain the actual adjustment amount of each axis; and obtain the corrected position coordinates based on the actual adjustment amount of each axis and the positioning coordinate to be processed.

[0117] In this embodiment, it should be noted that the specific methods of performing operations in the above-mentioned multi-means fusion UAV high-precision detection system have been described in detail in the embodiments of the multi-means fusion UAV high-precision detection method, and will not be elaborated here.

[0118] The preferred embodiments of this disclosure have been described in detail above with reference to the accompanying drawings. However, this disclosure is not limited to the specific details of the above embodiments. Within the scope of the technical concept of this disclosure, various simple modifications can be made to the technical solutions of this disclosure, and these simple modifications all fall within the protection scope of this disclosure.

[0119] It should also be noted that the various specific technical features described in the above embodiments can be combined in any suitable manner without contradiction. To avoid unnecessary repetition, this disclosure will not describe the various possible combinations separately.

[0120] Furthermore, various different embodiments of this disclosure can be combined in any way, as long as they do not violate the spirit of this disclosure, they should also be regarded as the content disclosed in this disclosure.

[0121] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the claims and specification of the present invention.

Claims

1. A high-precision detection method for unmanned aerial vehicles (UAVs) that integrates multiple methods, characterized in that, The methods include: Based on radar equipment, the target UAV is acquired at each detection time point within the current detection cycle. The data of each point is converted from the polar coordinate system to the three-dimensional coordinate system and the initial position coordinates are generated. The altitude data of the target UAV at various detection time points within the current detection cycle is obtained based on the integrated sensor base station. The initial position coordinates output by the radar device at the i-th detection time point are decomposed into the i-th planar coordinates; the height data output by the integrated sensing base station at the i-th detection time point is extracted. The i-th planar coordinates and the height data output at the i-th detection time point are combined to generate the positioning coordinates to be processed; Real-time wind speed and direction data are obtained based on meteorological information. The compensation ratio of the target UAV in each axis is obtained based on the real-time wind speed and direction data. The corrected position coordinates are obtained based on the compensation ratio of the target UAV in each axis and each positioning coordinate to be processed. The system acquires the aircraft type feature vector of the target UAV within the current detection period, matches the aircraft type feature vector with the pre-stored feature fingerprint database to generate an aircraft type confirmation signal, and generates an alarm signal containing the aircraft type confirmation signal based on multiple corrected position coordinates within the current detection period.

2. The high-precision UAV detection method based on multi-method fusion according to claim 1, characterized in that, The method of obtaining the compensation ratio of the target UAV in each axis based on real-time wind speed and direction data includes: The angle between the wind direction and the radar equipment beam is obtained based on wind direction data; The compensation ratio of the target UAV in each axis is obtained based on the beam angle.

3. The high-precision UAV detection method based on multi-method fusion according to claim 2, characterized in that, The process of obtaining the corrected position coordinates based on the compensation ratio of the target UAV in each axis and each positioning coordinate to be processed includes: The actual adjustment amount for each axis is obtained based on the adjustable amount and the compensation ratio for each axis. The corrected position coordinates are obtained based on the actual adjustment amount of each axis and the positioning coordinates to be processed.

4. The high-precision UAV detection method based on multi-method fusion according to claim 1, characterized in that, The acquisition of the target UAV's model feature vector within the current detection period includes: The frequency hopping cycle characteristics of the output within the current detection period are obtained based on the spectrum detection equipment; The rotor main frequency characteristic value output during the current detection period is obtained based on the acoustic sensor array; The frequency hopping cycle characteristics and rotor main frequency characteristics are integrated into the aircraft feature vector.

5. The high-precision UAV detection method based on multi-method fusion according to claim 1, characterized in that, The generation of an alarm signal containing an aircraft type confirmation signal based on multiple corrected position coordinates within the current detection period includes: The trajectory mutation rate for the current detection period is obtained based on multiple corrected position coordinates within the current detection period. If the trajectory mutation rate of the current detection cycle exceeds the preset threshold, the alarm level is obtained based on the trajectory mutation rate, and the model confirmation signal and the alarm level are integrated into an alarm signal.

6. A high-precision detection system for unmanned aerial vehicles (UAVs) that integrates multiple methods, characterized in that: The system includes: The acquisition module is used to acquire the point data of the target UAV at each detection time point in the current detection cycle based on the radar equipment, convert each point data from the polar coordinate system to the three-dimensional coordinate system and generate each initial position coordinate; The fusion module is used to acquire the altitude data of the target UAV at each detection time point within the current detection cycle based on the integrated sensing base station, and to acquire the corresponding positioning coordinates to be processed at the same detection time point based on the altitude data and initial position coordinates at the same detection time point. The fusion module is also used to: decompose the initial position coordinates output by the radar device at the i-th detection time point into the i-th planar coordinates; and extract the height data output by the integrated sensing base station at the i-th detection time point; The i-th planar coordinates and the height data output at the i-th detection time point are combined to generate the positioning coordinates to be processed; The correction module is used to obtain real-time wind speed and direction data based on meteorological information, obtain the compensation ratio of the target UAV in each axis based on the real-time wind speed and direction data, and obtain the coordinates of each correction position based on the compensation ratio of the target UAV in each axis and each positioning coordinate to be processed. The detection data processing module is used to acquire the aircraft type feature vector of the target UAV within the current detection period, match the aircraft type feature vector with the pre-stored feature fingerprint database and generate an aircraft type confirmation signal, and generate an alarm signal containing the aircraft type confirmation signal based on multiple corrected position coordinates within the current detection period.

7. The high-precision UAV detection system based on multi-method fusion according to claim 6, characterized in that, The correction module is also used for: The angle between the wind direction and the radar equipment beam is obtained based on wind direction data; The compensation ratio of the target UAV in each axis is obtained based on the beam angle.

8. The high-precision UAV detection system based on multi-method fusion according to claim 6, characterized in that, The correction module is also used for: The actual adjustment amount for each axis is obtained based on the adjustable amount and the compensation ratio for each axis. The corrected position coordinates are obtained based on the actual adjustment amount of each axis and the positioning coordinates to be processed.

Images