A method and system for identifying hidden drones in cities

Abstract

The present application provides a method and system for identifying concealed urban drones. The method obtains the geometric shape and material dielectric constant data of urban buildings and generates a set of building material reflection parameters. Secondly, based on the reflection parameters, the method selects a millimeter wave frequency band combination that is adapted to the building attenuation characteristics, directionally transmits millimeter wave signals, and separates the reflection components of metal parts and drone composite materials to generate a three-dimensional millimeter wave imaging spectrum that integrates multipath interference suppression parameters. Radio frequency signals are collected, and the path loss is compensated using the scattering phase offset. The propeller motor harmonic components and the communication module frequency hopping residual components are extracted to generate a time-space synchronized radio frequency fingerprint coding sequence. By constructing a coupling model of millimeter waves and radio frequencies, the reflection components and harmonic components are cross-validated to achieve accurate identification of concealed drones in the shielded area of ​​urban buildings. The present application improves the accuracy of identifying concealed urban drones.

Description

Technical Field

[0001] The present application relates to the field of drone identification technology, and in particular to a method and system for identifying hidden drones in cities. Background Art

[0002] Detecting covert drone activity in densely populated urban areas presents multiple challenges. The complex geometry and diverse materials of building structures can significantly alter the propagation characteristics of electromagnetic waves, leading to problems such as signal attenuation, multipath interference, and scattering phase shift. Furthermore, metal components within building-shaded areas and drone composite materials can produce similar reflection characteristics, further complicating target identification. Therefore, a technical solution is urgently needed that can separate multi-band signals in real time in a dynamic environment, suppress multipath effects, and accurately extract drone RF harmonic characteristics to achieve reliable detection of concealed drones.

[0003] The current mainstream solution is based on a detection framework that integrates millimeter-wave radar and radio frequency signals. This involves deploying multiple millimeter-wave radar arrays to transmit high-frequency signals, combining them with inertial measurement units and backscatter tags for assisted positioning. Polarization modulation is used to separate environmental reflections. For example, dual-polarization radar (horizontal and vertical polarization) combined with frequency modulation can distinguish tag echoes from environmental clutter. Inertial data is then used to compensate for changes in the drone's posture, generating six-degree-of-freedom spatial positioning information.

[0004] Existing solutions have limited adaptability in complex building environments. Single-band millimeter-wave signals are difficult to match the dynamic reflection attenuation characteristics of building materials, resulting in the inability to effectively separate the reflection components of metal parts and drone composite materials; relying on tag-assisted positioning requires pre-deployed hardware, and the tag backscatter signal is easily affected by the differences in dielectric constants of multiple materials, resulting in insufficient path loss compensation accuracy; RF harmonic component extraction relies on fixed-band filtering and cannot dynamically adapt to the time-varying characteristics of the communication module's frequency hopping residual component, resulting in a high RF fingerprint miss rate for covert drones and low identification accuracy for urban covert drones. Summary of the Invention

[0005] The present application provides a method and system for identifying concealed urban drones, which are used to solve the problem of low recognition accuracy of concealed urban drones in the prior art.

[0006] In a first aspect, the present application provides a method for identifying concealed urban drones, comprising:

[0007] In urban building clusters covered by drone activity, the geometric morphology and dielectric constant distribution data of building structures are obtained, and a set of building material reflection parameters including multi-band attenuation gradients and scattering phase offsets is generated using a dynamic reflection attenuation model.

[0008] Based on the set of building material reflection parameters, a millimeter wave frequency band combination that matches the dynamic reflection attenuation characteristics is selected, and a millimeter wave signal is transmitted to the area shielded by the building complex. The reflection components of the metal parts and the UAV composite material are separated according to the echo signal, and a three-dimensional millimeter wave imaging map containing multipath interference suppression parameters is generated;

[0009] Synchronously collecting radio frequency signals from the area shielded by the building complex, compensating for path loss using the scattering phase offset, extracting harmonic components of the drone's propeller motor and residual frequency hopping components of the communication module, and combining the radio frequency signals with multipath interference suppression parameters in the three-dimensional millimeter wave imaging atlas to generate a spatiotemporally synchronized radio frequency fingerprint coding sequence;

[0010] A millimeter wave and radio frequency coupling model is constructed based on the building material reflection parameter set, and hidden drones in the shielded area of ​​urban buildings are identified through cross-validation of the reflection component, the harmonic component and the radio frequency fingerprint coding sequence.

[0011] Optionally, separating the reflection components of the metal parts and the UAV composite material according to the echo signal to generate a three-dimensional millimeter wave imaging map containing multipath interference suppression parameters includes:

[0012] Performing a joint time-frequency domain analysis on the echo signal to separate the steady-state reflection component of the metal parts from the transient scattering component of the UAV composite material, and constructing multipath interference suppression parameters based on the attenuation gradient distribution in the building material reflection parameter set;

[0013] The transient scattering component is parameterized and filtered using the multipath interference suppression parameters to eliminate aliasing reflections caused by multipath propagation in the building complex shielding area. The spatial distribution characteristics of the multipath interference suppression parameters and the UAV composite material are visualized to generate a three-dimensional millimeter wave imaging map.

[0014] Optionally, the synchronously collecting the radio frequency signal in the area shielded by the building complex, compensating for the path loss by the scattering phase offset, and extracting the harmonic component of the drone propeller motor and the frequency hopping residual component of the communication module include:

[0015] Compensating for path loss by using the scattering phase offset, constructing a dynamic attenuation compensation function for the radio frequency signal propagation path, reconstructing a multipath propagation sequence for the radio frequency signal collected in the area shielded by the building complex, and eliminating the scattering phase distortion of the radio frequency signal caused by the building material;

[0016] The adjusted RF signal is jointly decomposed in the time-frequency domain to separate the harmonic components generated by the UAV propeller motor during operation and the residual frequency hopping components remaining during the frequency hopping switching of the communication module, and the transient harmonic feature set associated with the UAV motion state is extracted.

[0017] Optionally, combining the radio frequency signal with a multipath interference suppression parameter in the three-dimensional millimeter wave imaging atlas to generate a spatiotemporally synchronized radio frequency fingerprint coding sequence includes:

[0018] Establishing a mapping relationship between the radio frequency signal propagation path and the millimeter wave imaging spatial position according to the multipath interference suppression parameters in the three-dimensional millimeter wave imaging atlas, and performing spatiotemporal synchronization alignment on the harmonic component and the frequency hopping residual component according to the mapping relationship;

[0019] Based on the spatiotemporal synchronization alignment result, a spatiotemporal synchronized radio frequency fingerprint coding sequence is generated by combining the radio frequency signal with the reflector distribution characteristics in the three-dimensional millimeter wave imaging atlas.

[0020] Optionally, establishing a mapping relationship between the radio frequency signal propagation path and the millimeter wave imaging spatial position according to the multipath interference suppression parameter in the three-dimensional millimeter wave imaging atlas, and performing spatiotemporal synchronization alignment on the harmonic component and the frequency hopping residual component according to the mapping relationship includes:

[0021] Based on the multipath interference suppression parameters in the three-dimensional millimeter wave imaging atlas, the reflector distribution characteristics and the millimeter wave signal attenuation coefficient in the three-dimensional millimeter wave imaging atlas are analyzed to establish a geometric topological mapping relationship between the radio frequency signal propagation path and the millimeter wave imaging space grid;

[0022] Based on the geometric topological mapping relationship and in combination with the time delay and Doppler characteristic information of the radio frequency signal propagation path, a multipath propagation path topology model within the building complex shielding area is constructed, wherein the multipath propagation path topology model associates the reflector position in the millimeter wave imaging space grid with the geometric constraints of the radio frequency signal propagation path;

[0023] Extracting a timestamp sequence from the transient harmonic feature set and a frequency domain hopping interval of the frequency hopping residual component, and performing spatiotemporal reference alignment on the timestamp sequence and the frequency domain hopping interval based on reflector positions and geometric constraints associated with the multipath propagation path topology model;

[0024] The update rate of the millimeter wave imaging space grid is used to dynamically interpolate and compensate the spatiotemporal reference alignment result, generate spatiotemporal constraint parameters synchronized with the UAV motion state and communication frequency hopping mode, and complete the spatiotemporal synchronization alignment of the harmonic component and the frequency hopping residual component.

[0025] Optionally, based on the geometric topological mapping relationship and in combination with the delay and Doppler characteristic information of the radio frequency signal propagation path, a multipath propagation path topology model within the building complex shielding area is constructed, and the multipath propagation path topology model associates the reflector position in the millimeter wave imaging space grid with the geometric constraints of the radio frequency signal propagation path, including:

[0026] Analyze the correlation between the spatial position of the reflector and the millimeter wave signal attenuation coefficient in the geometric topological mapping relationship, and generate initial path propagation parameters by combining the shielding weights of each reflector on the radio frequency signal propagation path in the building complex shielding area;

[0027] Combining the delay and Doppler characteristic information of the radio frequency signal propagation path and the dynamic impact of the UAV's motion state on the Doppler frequency shift, correcting the delay estimation error in the initial path propagation parameters to generate updated path propagation parameters that include path loss gradient and time-frequency joint constraints;

[0028] Based on the updated path propagation parameters, ray tracing modeling is performed on the multipath signal propagation paths within the building complex shielding area to generate a set of path clusters that match the positions of reflectors in the millimeter wave imaging space grid;

[0029] According to the distribution of overlapping areas between the path cluster set and the reflector positions in the millimeter wave imaging space grid, a multipath propagation path topology model is constructed after quantifying the attenuation ratio of the direct path and the multipath reflection path of the RF signal propagation path, and the multipath propagation path topology model is associated with the geometric constraints of the reflector positions in the millimeter wave imaging space grid and the RF signal propagation path.

[0030] Optionally, combining the delay and Doppler characteristic information of the radio frequency signal propagation path, and correcting the delay estimation error in the initial path propagation parameters according to the dynamic impact of the UAV motion state on the Doppler frequency shift, to generate updated path propagation parameters containing path loss gradient and time-frequency joint constraints, includes:

[0031] Extracting the Doppler frequency shift dynamic characteristics associated with the UAV's motion state from the time delay and Doppler characteristic information of the radio frequency signal propagation path, and generating a correlation parameter between the Doppler frequency shift and the motion state by combining the mapping relationship between the propeller rotation frequency and the UAV speed;

[0032] Based on the associated parameters, a dynamic impact model of the UAV's motion acceleration on the multipath propagation delay is established, and the nonlinear relationship between the delay estimation error in the initial path propagation parameters and the UAV's motion acceleration is quantified using the dynamic impact model;

[0033] Based on the dynamic impact model, dynamically weighting the delay estimation values ​​in the initial path propagation parameters is performed, and combining the weighting results with the loss gradient distribution of radio frequency signals by different materials in the building complex shielding area to generate the path propagation parameters after delay compensation;

[0034] The path propagation parameters are matrix-fused with the time-frequency joint constraints of the radio frequency signal. The fusion process eliminates the time-frequency aliasing error caused by the UAV's maneuvering flight and generates updated path propagation parameters containing the path loss gradient and the time-frequency joint constraints.

[0035] Optionally, constructing a millimeter wave and radio frequency coupling model based on the building material reflection parameter set, and identifying concealed drones in an area shielded by urban buildings through cross-validation of the reflection component, the harmonic component, and the radio frequency fingerprint code sequence, includes:

[0036] Based on the multi-band attenuation gradient and the spatiotemporal synchronization constraints in the radio frequency fingerprint coding sequence, a dynamic correlation matrix of the millimeter wave reflection component intensity and the radio frequency harmonic component amplitude is established, and an initial characteristic mapping relationship of the millimeter wave and radio frequency coupling model is generated according to the dynamic correlation matrix;

[0037] According to the spatial distribution density and material attenuation difference of the reflectors in the three-dimensional millimeter wave imaging atlas, the dynamic correlation matrix is ​​subjected to multimodal feature fusion in combination with the initial feature mapping relationship to construct a harmonic component amplitude compensation function under the millimeter wave spatial resolution constraint;

[0038] The harmonic component amplitude compensation function is used to correct the harmonic attenuation distortion in the radio frequency fingerprint code sequence caused by building shielding, and the reflection characteristics of the UAV composite material and the spatiotemporal consistency constraint parameters of the radio frequency harmonic component amplitude are extracted based on the corrected harmonic attenuation distortion;

[0039] Based on the spatiotemporal consistency constraint parameters, a multi-dimensional joint verification is performed on the distribution of reflection components in the three-dimensional millimeter wave imaging map, the frequency hopping residual pattern of the RF harmonic component amplitude, and the RF fingerprint coding sequence. The verification process eliminates the feature confusion between the metal frame inside the building and the composite material of the drone, and completes the spatial positioning and identification of concealed drones in the city.

[0040] Optionally, the multi-dimensional joint verification of the reflection component distribution in the three-dimensional millimeter wave imaging map, the frequency hopping residual pattern of the RF harmonic component amplitude, and the RF fingerprint coding sequence based on the spatiotemporal consistency constraint parameters is performed. The verification process eliminates feature confusion between the metal frame inside the building and the composite material of the UAV, and completes the spatial positioning and identification of the urban concealed UAV, including:

[0041] Based on the spatial distribution correlation between the millimeter wave reflection component and the radio frequency harmonic component in the spatiotemporal consistency constraint parameter, the steady-state reflection characteristics of the metal frame inside the building and the transient scattering characteristics of the composite material of the drone are extracted to generate a joint distribution map of the reflection component and the harmonic component;

[0042] Based on the spatial overlap area between the metal frame inside the building and the composite material of the drone in the joint distribution map, combined with the amplitude frequency hopping residual pattern of the radio frequency harmonic component, a metal reflection suppression function and a composite material enhancement function are constructed to generate a feature confusion elimination parameter set;

[0043] Dynamically weighting the distribution of reflection components in the three-dimensional millimeter-wave imaging atlas using the feature confusion elimination parameter set, and combining the distribution of reflection components after the allocation, the time evolution law of the amplitude frequency hopping residual pattern of the radio frequency harmonic component, and the radio frequency fingerprint coding sequence to generate a joint verification parameter that integrates spatial positioning constraints and harmonic feature identification;

[0044] The joint distribution map is iteratively optimized in multiple dimensions based on the joint verification parameters. The optimization process eliminates feature confusion between the metal frame inside the building and the composite material of the drone, thereby completing the spatial positioning and identification of concealed drones in the city.

[0045] In a second aspect, the present application provides a system for identifying hidden urban drones, comprising:

[0046] The acquisition module is used to obtain the geometric shape and material dielectric constant distribution data of the building structures in the urban building complex covered by the drone's activity area, and generate a set of building material reflection parameters including multi-band attenuation gradients and scattering phase offsets through a dynamic reflection attenuation model;

[0047] A screening module is configured to screen a millimeter wave frequency band combination that matches the dynamic reflection attenuation characteristics based on the set of building material reflection parameters, transmit millimeter wave signals to the area shielded by the building complex, separate the reflection components of the metal parts and the UAV composite materials based on the echo signals, and generate a three-dimensional millimeter wave imaging map containing multipath interference suppression parameters;

[0048] an extraction module for synchronously collecting radio frequency signals from the area shielded by the building complex, compensating for path loss using the scattering phase offset, extracting harmonic components of the drone's propeller motor and residual frequency hopping components of the communication module, and combining the radio frequency signals with multipath interference suppression parameters in the three-dimensional millimeter wave imaging atlas to generate a spatiotemporally synchronized radio frequency fingerprint coding sequence;

[0049] The identification module is used to construct a millimeter wave and radio frequency coupling model based on the set of building material reflection parameters, and identify hidden drones in the shielded area of ​​urban buildings through cross-validation of the reflection component, the harmonic component and the radio frequency fingerprint coding sequence.

[0050] In an embodiment of the present application, in an urban building complex covered by a drone activity area, geometric morphology and material dielectric constant distribution data of the building structure are obtained, and a set of building material reflection parameters including multi-band attenuation gradients and scattering phase offsets is generated through a dynamic reflection attenuation model; based on the building material reflection parameter set, a millimeter wave frequency band combination matching the dynamic reflection attenuation characteristics is screened, and a millimeter wave signal is transmitted to the shielded area of ​​the building complex. The reflection components of the metal parts and the drone composite materials are separated according to the echo signal to generate a three-dimensional millimeter wave imaging map including multipath interference suppression parameters; the radio frequency signal of the shielded area of ​​the building complex is synchronously collected, the path loss is compensated by the scattering phase offset, the harmonic component of the drone propeller motor and the frequency hopping residual component of the communication module are extracted, and the multipath interference suppression parameters in the three-dimensional millimeter wave imaging map are combined to generate a spatiotemporally synchronized radio frequency fingerprint coding sequence; a millimeter wave and radio frequency coupling model is constructed based on the building material reflection parameter set, and the concealed drone in the shielded area of ​​the urban building complex is identified by cross-validation of the reflection component and the harmonic component.

[0051] The technical solution of this application has the following beneficial effects:

[0052] By accurately capturing the geometric shape and dielectric constant distribution data of building complexes, the dynamic reflection attenuation model can quantify the millimeter-wave reflection characteristics of different materials (such as attenuation gradient and scattering phase offset), providing a physical basis for subsequent signal processing and ensuring the adaptability of multi-band signals in complex environments. Millimeter-wave frequency band combinations selected based on dynamic reflection attenuation characteristics can penetrate building-blocked areas and suppress multipath interference. By separating the reflection components of metal parts and drone composite materials, a high-resolution three-dimensional millimeter-wave imaging map is generated, significantly improving the detection accuracy of concealed targets. Using scattering phase offset to compensate for path loss, propeller motor harmonic components and communication frequency hopping residual signals are extracted, enhancing the ability to capture drone electromagnetic signatures, especially in low signal-to-noise ratio environments. By cross-validating over-reflection and harmonic components, the coupled model effectively distinguishes drones from background interference (such as metal building structures), achieving high-confidence concealed target recognition. This method integrates the advantages of multi-source data to improve anti-interference performance in complex urban environments.

[0053] Furthermore, based on a set of building material reflection parameters, a millimeter-wave frequency band combination that matches the dynamic reflection attenuation characteristics is selected. Millimeter-wave signals are then transmitted into the area shielded by the building complex. The echo signals are used to separate the reflection components of the metal components from the drone's composite materials, generating a three-dimensional millimeter-wave imaging map that includes multipath interference suppression parameters. This method leverages the strong penetrating properties of millimeter waves and combines them with a multipath interference suppression algorithm to achieve high-resolution three-dimensional imaging of areas shielded by the building complex. This reflection component separation technique effectively distinguishes the drone's composite materials from background metal structures, significantly improving the detection accuracy of concealed targets while reducing the impact of environmental clutter on imaging quality.

[0054] These and other aspects of the present application will become more readily apparent from the description of the following embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0056] Figure 1 A flow chart of a method for identifying hidden urban drones provided by this application is shown;

[0057] Figure 2 The figure shows a structural diagram of an urban covert drone identification system provided by the present application.

[0058] Figure 3 A scene diagram showing a method for identifying hidden urban drones provided by this application is shown. DETAILED DESCRIPTION

[0059] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0060] In some of the processes described in the specification and claims of this application and the above-mentioned figures, multiple operations that appear in a specific order are included, but it should be clearly understood that these operations may not be executed in the order in which they appear in this document or may be executed in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish between different operations, and the serial numbers themselves do not represent any order of execution. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this document are used to distinguish different messages, devices, modules, etc., and do not represent a sequential order, nor do they limit "first" and "second" to being different types.

[0061] Current urban covert drone detection technology faces three core bottlenecks: First, in complex building environments, differences in the dielectric constants of building materials lead to serious electromagnetic wave multipath interference and scattering phase shifts. Single-band millimeter-wave radars are difficult to adapt to dynamic reflection attenuation characteristics, and the reflection components of metal parts and drone composite materials cannot be effectively separated due to frequency band mismatch; second, existing solutions rely on pre-deployed backscatter tags for auxiliary positioning, but tag signals are easily affected by differences in the dielectric properties of multiple materials, path loss compensation accuracy is insufficient, and hardware deployment costs are high and flexibility is poor; third, RF harmonic feature extraction relies on fixed-band filtering and cannot dynamically adapt to the time-varying characteristics of the residual component of the frequency hopping of the drone communication module, resulting in a high RF fingerprint miss rate for concealed targets and limited recognition accuracy.

[0062] To address these shortcomings, the present invention proposes a drone identification method based on the coordinated reflection characteristics of building materials, millimeter-wave penetrating imaging, and radio frequency fingerprinting. The core of this method lies in the construction of a cross-modal dynamic adaptation framework. First, a dynamic reflection attenuation model is used to generate a multi-band attenuation gradient and scattering phase offset set. Millimeter-wave frequency band combinations that match the attenuation characteristics of building materials are selected. Multi-band signals are then transmitted directionally to separate the reflection components of metal and composite materials, generating a three-dimensional millimeter-wave imaging map that incorporates multipath interference suppression parameters. Radio frequency signals are then collected simultaneously, combined with scattering phase offsets to compensate for path loss. Harmonic components of the drone's propellers and residual components from frequency hopping are extracted to generate a spatiotemporally synchronized radio frequency fingerprinting sequence. Finally, the reflection components, harmonic components, and radio frequency fingerprinting sequence are cross-validated using a millimeter-wave-radio frequency coupling model. This approach overcomes the limitations of conventional technologies: multi-band dynamic adaptation effectively suppresses metal interference reflections and eliminates label dependency; scattering phase offset compensation and dynamic extraction of frequency hopping residuals improve radio frequency fingerprinting accuracy; and cross-modal cross-validation enables high-confidence identification of concealed drones in non-line-of-sight (NLOS) scenarios, providing reliable technical support for urban low-altitude security.

[0063] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without making creative efforts are within the scope of protection of this application.

[0064] Figure 1 A flowchart of a method for identifying hidden drones in cities is provided for the embodiment of the present application. Figure 1 As shown, the method includes:

[0065] 101. In the urban building complex covered by the drone activity area, obtain the geometric shape of the building structure and the distribution data of the material dielectric constant, and generate a set of building material reflection parameters including multi-band attenuation gradients and scattering phase offsets through a dynamic reflection attenuation model;

[0066] In the above scheme, geometric form refers to the three-dimensional shape and dimensions of a building structure, including physical topological features such as wall inclination, surface curvature, and building spacing. Material dielectric constant distribution data describes the dielectric response characteristics of building materials to electromagnetic waves at different spatial locations, including both real and imaginary components. The dynamic reflection attenuation model is a mathematical model based on the physics of electromagnetic wave propagation. It calculates the reflection intensity attenuation rate and phase offset of millimeter-waves of different frequency bands in building materials. Its inputs are the dielectric constant distribution and geometric form data. The multi-band attenuation gradient refers to the slope distribution of the reflection intensity attenuation rate of building materials for millimeter-wave bands as a function of frequency, characterizing the material's differentiated attenuation characteristics for multi-band signals. The scattering phase offset describes the phase change caused by differences in the dielectric properties of materials during electromagnetic wave scattering from building surfaces. It is used to quantify the degree of phase distortion in multipath propagation paths. The building material reflection parameter set is a parameter matrix composed of multi-band attenuation gradients and scattering phase offsets, serving as a physical priori knowledge base for millimeter-wave frequency band optimization and RF path compensation.

[0067] In this embodiment, LiDAR and oblique photogrammetry are used to scan and densely reconstruct a 3D point cloud of the target building complex, generating sub-meter-level architectural geometry data. For example, a multi-rotor drone equipped with a LiDAR sensor acquires a point cloud of the building facade at a resolution of 0.1 meters, and a Poisson surface reconstruction algorithm is used to generate a 3D mesh model of the building structure.

[0068] Secondly, using a near-field microwave flaw detector and a dielectric resonance probe, non-contact dielectric property measurements of building surface materials are performed. For example, a microwave probe with a frequency tunable from 1 to 100 GHz is deployed on the surface of a concrete wall. The real and imaginary parts of the dielectric constant are inverted by the resonant frequency offset. Combined with a Kalman filter algorithm to eliminate environmental noise interference, a heat map of the spatial distribution of the dielectric constant is generated.

[0069] Finally, the geometric data and dielectric constant distribution are input into the dynamic reflection attenuation model. The propagation path of millimeter waves in the building complex is simulated through time-domain finite difference and algorithm, and the reflection intensity attenuation gradient and scattering phase offset of the multi-band signal are calculated: the reflection intensity attenuation gradient is calculated by the frequency domain attenuation spectrum analysis method, and the reflected signal of each frequency band is short-time Fourier transformed to extract the exponential attenuation curve of the signal energy with the propagation distance, and its logarithmic attenuation slope is calculated as the attenuation gradient; the scattering phase offset is obtained through phase unwrapping technology, and the complex spectrum of the echo signal is discrete Fourier transformed to extract the phase response of the main scattering path. Combined with the surface curvature in the geometric data, the absolute phase offset is solved using the least squares phase gradient integration method, and finally the building material reflection parameter set is output. For example, a 10ns time window is cut for the 24GHz frequency band echo signal, and a time-frequency spectrum matrix is ​​generated by short-time Fourier transform, and the energy attenuation curve is fitted along the propagation distance axis. ,extract The value is used as the attenuation gradient of the frequency band; the discrete Fourier transform is performed on the complex echo signal in the 60GHz frequency band to extract the phase value of the main scattering path , combined with the wall surface curvature radius , through the phase gradient integral formula , solve for the absolute phase offset ,in is the wave number.

[0070] It should be noted that the above dynamic reflex attenuation model is a training model, and its specific training process mainly includes the following steps:

[0071] First, the building's geometric data (such as wall inclination, curvature, and spacing) are acquired through lidar and oblique photography to construct a three-dimensional grid model with sub-meter accuracy. Simultaneously, a near-field microwave flaw detector is used to measure the building's surface dielectric constant (real and imaginary parts). After noise is eliminated through Kalman filtering, a thermal map of the dielectric constant's spatial distribution is generated. Finally, the geometric data is paired with the dielectric distribution data to form a training sample set.

[0072] Secondly, the finite-difference time-domain (FDTD) algorithm is used to simulate the propagation process of millimeter waves in building structures, and the simulation results are physically calculated to generate label data: for the multi-band attenuation gradient, the time domain segment of the reflected signal in each frequency band is intercepted (such as a 10ns time window), and the energy attenuation curve is extracted through the short-time Fourier transform (STFT), and the exponential equation is fitted. , with the attenuation coefficient As the decay gradient label.

[0073] Based on the scattered phase offset, the complex echo signal is subjected to discrete Fourier transform (DFT) to extract the main path phase , combined with the curvature radius in the geometric data , through the phase gradient integral formula ( is the wave number) to solve the absolute phase offset As a label.

[0074] Finally, the attenuation coefficient is calculated by taking the geometry and material dielectric constant distribution data as input. and phase offset To achieve the output target, a deep learning model (such as a neural network based on physical constraints) is constructed. The model is trained with a large number of samples to learn the physical laws of electromagnetic reflection. The model's accuracy is verified using FDTD simulation results until the error between its output and the physical simulation is less than a set threshold. The trained model is then defined as a dynamic reflection attenuation model.

[0075] The dynamic reflection attenuation model has been trained to internalize the interaction between electromagnetic waves and building materials. It can directly predict the reflection physical quantity in any frequency band based on the new geometric-dielectric input data: the model automatically calculates the attenuation coefficient of signal energy with distance through the internal frequency domain attenuation spectrum analysis module. , output multi-band attenuation gradient matrix, combine geometric curvature and dielectric properties, solve the least squares phase integral through embedded phase unwrapping algorithm, and output scattering phase offset Finally, the attenuation gradients and phase offsets of multiple frequency bands (such as 24 / 60 GHz) are integrated into a two-dimensional parameter matrix to form a set of building material reflection parameters, which serves as a priori knowledge base for millimeter wave communication frequency band optimization and multipath phase compensation.

[0076] 102. Based on the set of building material reflection parameters, select a millimeter wave frequency band combination that matches the dynamic reflection attenuation characteristics, transmit millimeter wave signals to the area shielded by the building complex, separate the reflection components of the metal parts and the UAV composite materials based on the echo signals, and generate a three-dimensional millimeter wave imaging map containing multipath interference suppression parameters;

[0077] Optionally, step 102 may specifically include the following steps:

[0078] 1021. Perform a joint time-frequency domain analysis on the echo signal to separate the steady-state reflection component of the metal component and the transient scattering component of the UAV composite material, and construct a multipath interference suppression parameter based on the attenuation gradient distribution in the building material reflection parameter set;

[0079] 1022. Perform parameterized filtering on the transient scattering component using the multipath interference suppression parameters to eliminate aliasing reflections caused by multipath propagation in the building complex shielding area, visualize the spatial distribution characteristics of the multipath interference suppression parameters and the UAV composite material, and generate a three-dimensional millimeter wave imaging map.

[0080] In the above scheme, dynamic reflection attenuation characteristics refer to the temporal or spatial variation of the reflection intensity attenuation rate of millimeter-wave signals of different frequency bands from building materials. A millimeter-wave frequency band combination is a collection of multiple millimeter-wave frequency bands selected based on the attenuation gradient of building materials, used to optimize signal penetration and reflection separation. Time-frequency domain joint analysis simultaneously analyzes the time-domain waveform and frequency-domain spectral characteristics of the echo signal to distinguish the signal components of different reflectors. The steady-state reflection component is a stable, high-intensity reflection signal generated by metal components within the building. The transient scattering component is a time-varying scattering signal generated by the motion or material properties of the drone's composite materials. Multipath interference suppression parameters are a set of parameters that quantify the degree of interference multipath propagation paths have on target signals. A three-dimensional millimeter-wave imaging atlas is a three-dimensional visualization generated based on the spatial distribution of reflection components and the multipath interference suppression parameters, characterizing the location and material characteristics of the target object.

[0081] In the embodiment of the present application, first, through step 1021, based on the attenuation gradient distribution in the building material reflection parameter set, a frequency band optimization algorithm is used to screen out a millimeter wave frequency band combination (such as the 60 GHz and 94 GHz bands) whose penetration capability matches the material's dynamic attenuation characteristics. A multi-band frequency modulated continuous wave signal is directionally transmitted toward the building's shielded area, and the reflection components are separated through joint time-frequency domain analysis: a short-time Fourier transform is performed on the echo signal to generate a time-frequency spectrum matrix, and a wavelet transform is used to extract the steady-state reflection component of the metal component and the transient scattering component of the drone's composite material. For example, for the 60 GHz frequency band echo signal, the fixed frequency shift component of the metal frame is locked using time-frequency ridge tracking technology. Combined with the attenuation gradient distribution, the multipath interference suppression parameters are calculated: the path attenuation coefficient λ = 2.5 dB / m, the phase offset Δφ = 30°, and the suppression parameter matrix is ​​generated.

[0082] Subsequently, in step 1022, a parameterized Wiener filter is applied to the transient scattered component using the multipath interference suppression parameter matrix: a frequency domain filter H(f) = 1 / (1 + λ(f) / SNR(f)) is constructed, where SNR(f) is the signal-to-noise ratio spectrum. This filter suppresses aliased reflections caused by multipath propagation, such as a 6 dB attenuation of the intensity of secondary reflections from a glass curtain wall. The filtered transient scattered component is fused with the multipath interference suppression parameters using the Kriging spatial interpolation algorithm, and a three-dimensional millimeter-wave imaging map is generated based on the spatial distribution characteristics of the UAV's composite materials. For example, the scattered point cloud at coordinates X = 35.2, Y = 78.6, and Z = 12.3 in the UAV's hovering area is density clustered and rendered in pseudo-color, producing a three-dimensional millimeter-wave imaging map with the metal frame highlighted in red and the UAV marked in blue.

[0083] In practical applications, in urban historical building protection areas dominated by masonry structures and densely packed with metal decorative elements, step 1021 selects the 94 GHz band as the primary transmission frequency band, with a masonry attenuation gradient of α = 1.5 dB / m. Time-frequency ridge tracking is used to separate the steady-state reflection component of the metal decorative elements, with a reflection intensity of 22 dBm and a Doppler shift of ±30 Hz. The transient scattering component of the UAV's fiberglass fuselage is also tracked, with a reflection intensity of 7 dBm and a Doppler shift of ±180 Hz. The resulting suppression parameter matrix is ​​λ = 1.8 dB / m and Δφ = 25°. Step 1022 uses Wiener filtering to eliminate multipath interference from masonry walls, and generates a three-dimensional imaging map through inverse distance weighted interpolation, clearly marking the UAV's location at the top of the bell tower: coordinates X = 102.3, Y = 55.7, and Z = 45.2.

[0084] The overall solution of step 102 above achieves precise separation of metal and composite material reflection components through dynamic frequency band optimization and joint time-frequency analysis, breaking through the material confusion bottleneck of traditional single-band radar; the Wiener filter driven by multipath interference suppression parameters effectively eliminates environmental clutter and improves the signal-to-noise ratio of transient scattering components; the three-dimensional millimeter wave imaging spectrum integrates spatial distribution and material attenuation characteristics to provide covert drones with high-resolution spatial positioning and anti-interference imaging capabilities, significantly enhancing the reliability of target detection in complex urban shielding environments.

[0085] 103. Synchronously collect radio frequency signals from the area shielded by the building complex, compensate for path loss using the scattering phase offset, extract harmonic components of the drone propeller motor and residual frequency hopping components of the communication module, and combine the radio frequency signals with multipath interference suppression parameters in the three-dimensional millimeter wave imaging atlas to generate a spatiotemporally synchronized radio frequency fingerprint coding sequence.

[0086] Optionally, step 103 may specifically include the following steps:

[0087] 1031. Compensate for path loss using the scattering phase offset, construct a dynamic attenuation compensation function for the radio frequency signal propagation path, reconstruct a multipath propagation sequence for the radio frequency signal collected within the building complex shielding area, and eliminate scattering phase distortion of the radio frequency signal caused by building materials.

[0088] 1032. Perform a joint time-frequency decomposition on the adjusted RF signal to separate the harmonic components generated when the drone propeller motor is running and the frequency hopping residual components remaining during the frequency hopping switching process of the communication module, and extract the transient harmonic feature set associated with the drone's motion state.

[0089] 1033. Establish a mapping relationship between the radio frequency signal propagation path and the millimeter wave imaging spatial position according to the multipath interference suppression parameter in the three-dimensional millimeter wave imaging atlas, and perform spatiotemporal synchronization alignment on the harmonic component and the frequency hopping residual component according to the mapping relationship;

[0090] Among them, step 1033 may specifically include the following processes: based on the multipath interference suppression parameters in the three-dimensional millimeter wave imaging map, analyzing the reflector distribution characteristics and the millimeter wave signal attenuation coefficient in the three-dimensional millimeter wave imaging map, and establishing a geometric topological mapping relationship between the radio frequency signal propagation path and the millimeter wave imaging space grid; according to the geometric topological mapping relationship, combined with the time delay and Doppler characteristic information of the radio frequency signal propagation path, constructing a multipath propagation path topology model in the building complex shielding area, and the multipath propagation path topology model is associated with the reflector position in the millimeter wave imaging space grid. and the geometric constraints of the RF signal propagation path; extract the timestamp sequence in the transient harmonic feature set and the frequency domain hopping interval of the frequency hopping residual component, and perform spatiotemporal reference alignment on the timestamp sequence and the frequency domain hopping interval based on the reflector position and geometric constraints associated with the multipath propagation path topology model; use the update rate of the millimeter wave imaging space grid to dynamically interpolate and compensate the spatiotemporal reference alignment result, generate spatiotemporal constraint parameters synchronized with the UAV motion state and communication frequency hopping mode, and complete the spatiotemporal synchronization alignment of the harmonic component and the frequency hopping residual component.

[0091] Among them, the process of constructing a multipath propagation path topology model in the building complex shielding area based on the geometric topology mapping relationship and combining the delay and Doppler characteristic information of the radio frequency signal propagation path, and associating the multipath propagation path topology model with the reflector position in the millimeter wave imaging space grid and the geometric constraints of the radio frequency signal propagation path includes:

[0092] The correlation between the spatial position of the reflector and the millimeter-wave signal attenuation coefficient in the geometric topological mapping relationship is analyzed, and the initial path propagation parameters are generated in combination with the shielding weights of each reflector on the radio frequency signal propagation path in the building complex shielding area; the delay estimation error in the initial path propagation parameters is corrected in combination with the characteristic information of the radio frequency signal propagation path and Doppler according to the dynamic influence of the UAV motion state on the Doppler frequency shift, and updated path propagation parameters containing path loss gradient and time-frequency joint constraints are generated; based on the updated path propagation parameters, ray tracing modeling is performed on the multipath signal propagation path in the building complex shielding area to generate a path cluster set that matches the reflector position in the millimeter-wave imaging space grid; based on the overlapping area distribution of the path cluster set and the reflector position in the millimeter-wave imaging space grid, a multipath propagation path topology model is constructed after quantifying the attenuation ratio of the direct path and the multipath reflection path of the radio frequency signal propagation path, and the multipath propagation path topology model is associated with the reflector position in the millimeter-wave imaging space grid and the geometric constraints of the radio frequency signal propagation path.

[0093] The process of combining the delay and Doppler characteristic information of the radio frequency signal propagation path, correcting the delay estimation error in the initial path propagation parameters according to the dynamic impact of the UAV motion state on the Doppler frequency shift, and generating updated path propagation parameters containing path loss gradient and time-frequency joint constraints includes:

[0094] The Doppler frequency shift dynamic characteristics associated with the UAV's motion state are extracted from the delay and Doppler characteristic information of the radio frequency signal propagation path, and the correlation parameters between the Doppler frequency shift and the motion state are generated by combining the mapping relationship between the propeller rotation frequency and the UAV speed. Based on the correlation parameters, a dynamic impact model of the UAV's motion acceleration on the multipath propagation delay is established, and the nonlinear relationship between the delay estimation error in the initial path propagation parameters and the UAV's motion acceleration is quantified through the dynamic impact model. Based on the dynamic impact model, the delay estimation value in the initial path propagation parameter is dynamically weighted, and the allocation result is combined with the loss gradient distribution of the radio frequency signal due to different materials in the shielding area of ​​the building complex to generate the path propagation parameter after delay compensation. The path propagation parameter is matrix-fused with the radio frequency joint constraint condition of the radio frequency signal. The fusion process eliminates the time-frequency aliasing error caused by the UAV's maneuvering flight, and generates updated path propagation parameters containing the path loss gradient and the time-frequency joint constraint.

[0095] Among them, the training process of the dynamic influence model is:

[0096] By collecting UAV maneuver flight data (including acceleration , propeller speed ), combined with the Doppler frequency shift formula Inversion motion speed , and use the gradient descent method to construct the delay estimation error With acceleration Quadratic function model (For example, when Time fitting ), and then optimize the parameters by the least squares method Minimize the prediction error (objective function ), and finally use the actual data of maneuver flight to verify the accuracy of the model (such as Time error ).

[0097] The dynamic impact model is modeled by the quadratic term Quantify the nonlinear effects of acceleration on multipath delay (e.g. Increased from 2 to hour from 0.45 to 1.65 ns), based on dynamic weight allocation rules hour Correcting initial delay initial , and integrate the building material loss gradient (such as glass ) Generate compensation parameters ), and finally the joint time-frequency constraint is used to eliminate the mixing error (such as frequency shift) by orthogonal decomposition. Automatic correction when 150Hz exceeds the tolerance limit).

[0098] 1034. Based on the spatiotemporal synchronization alignment result, a spatiotemporal synchronization radio frequency fingerprint coding sequence is generated by combining the radio frequency signal with the reflector distribution characteristics in the three-dimensional millimeter wave imaging atlas.

[0099] In the above scheme, the scattering phase offset is the phase distortion caused by the scattering of RF signals by building materials and is used to compensate for phase distortion in multipath propagation. The dynamic attenuation compensation function is a mathematical function constructed based on the scattering phase offset, which is used to correct for path loss caused by building material attenuation of RF signals. Joint time-frequency decomposition is a technique that simultaneously analyzes the time-domain waveform and frequency-domain spectral characteristics of RF signals to separate different signal components. Harmonic components are periodic electromagnetic radiation signals generated by the operation of the drone's propeller motor, with frequencies that are integer multiples of the motor's fundamental frequency. The frequency hopping residual component is the residual non-continuous frequency band signal during the frequency hopping switching of the drone's communication module, representing the characteristics of the frequency hopping pattern. Spatiotemporal synchronization alignment is the process of spatially and temporally correlating the timing characteristics of the RF signal with the spatial position of the millimeter-wave image. The RF fingerprint coding sequence is a spatiotemporal correlation feature sequence that integrates the harmonic components, the frequency hopping residual component, and the millimeter-wave spatial constraints.

[0100] In an embodiment of the present application, first, through step 1031, a broadband RF sensor array is deployed in the area shielded by the building complex to collect the original waveform of the multipath propagated RF signal in real time. Based on the scattering phase offset generated in step 101, a dynamic attenuation compensation function H(f) = e^{jΔφ(f)}·A(f)^{-1} is constructed, where Δφ is the phase offset and A(f) is the frequency-dependent attenuation coefficient. Phase rotation and amplitude compensation are performed on the original signal to eliminate signal distortion caused by building materials. For example, for a 5.8 GHz frequency band signal, the propagation delay difference between the direct path and the multipath reflection path is reconstructed through a reverse ray tracing algorithm, and combined with phase gradient alignment technology, a compensated RF signal sequence is output.

[0101] Subsequently, step 1032 performs a joint time-frequency decomposition of the compensated RF signal sequence. Adaptive wavelet packet transform is used to separate the periodic harmonic components generated by the propeller motor and the discontinuous pulse components left over from the frequency hopping in the communication module on the time-frequency plane. Specifically, the instantaneous frequency curve of the harmonic components is extracted using time-frequency ridge tracking technology. The Viterbi state transition algorithm is then used to identify the frequency band switching interval of the residual frequency hopping components, such as a 50ms hop interval from 5.8 GHz to 2.4 GHz. This generates a transient harmonic feature set containing harmonic amplitudes and frequency hopping timing.

[0102] At the same time, step 1033 establishes a geometric mapping relationship between the RF signal propagation path and the millimeter wave spatial grid based on the multipath interference suppression parameters in the three-dimensional millimeter wave imaging atlas, such as a path attenuation coefficient <3 dB / m: the incident angle of the RF signal is analyzed based on the direction of arrival estimation algorithm, and the time delay information of the harmonic component is mapped to the three-dimensional space in combination with the millimeter wave imaging grid coordinates. The dynamic time warping algorithm is used to align the timing characteristics of the frequency hopping residual component with the millimeter wave imaging refresh period (e.g., 10 Hz) to generate spatiotemporal synchronization alignment parameters.

[0103] Finally, in step 1034, the harmonic components and the frequency-hopping residual components are fused based on the spatiotemporal alignment parameters. A spatial density clustering algorithm is used to spatially filter reflector distribution areas (e.g., areas with dense metal frames) in millimeter-wave imaging. The frequency-hopping pattern is then modeled using a hidden Markov model to generate a spatiotemporally synchronized RF fingerprint sequence that incorporates the X, Y, and Z spatial coordinates, harmonic amplitudes, and frequency-hopping intervals. For example, the 5.8 GHz frequency-hopping residual pulse is associated with the reflected component of the hovering area in millimeter-wave imaging to generate a unique RF fingerprint sequence: "F_5.8G_50ms_35.2_78.6_12.3."

[0104] In actual applications, distributed RF sensors and millimeter-wave radar arrays are deployed in shielded areas of urban commercial complexes (glass curtain walls account for 60% and steel structure corridors are dense). In step 1031, dynamic attenuation compensation is performed on the 5.8 GHz RF signal to eliminate the 120° phase offset and 15 dB path loss caused by the steel structure corridor, and the direct path signal is reconstructed. In step 1032, the 1.2 kHz harmonic component of the propeller motor (amplitude -30 dBm) and the 5.8 GHz / 2.4 GHz dual-band pulses (interval 50 ms) remaining from the frequency hopping are extracted. In step 1033, millimeter wave imaging is used to determine that the drone is hovering in an area 3 meters behind the glass curtain wall (coordinates X=35.2, Y=78.6, Z=12.3), and the harmonic delay (8 ns) is mapped to these coordinates. In step 1034, the RF fingerprint code "F_5.8G_50ms_35.2_78.6_12.3" is generated, which integrates the spatial coordinates, frequency hopping interval, and harmonic amplitude as the target's unique RF fingerprint code sequence.

[0105] The overall solution of step 103 above improves the quality of RF signals through dynamic attenuation compensation and multipath reconstruction, overcoming the harmonic distortion caused by interference from building materials; the time-frequency joint decomposition accurately separates the power and communication characteristics of the UAV, solving the problem of dynamic capture of the residual components of frequency hopping; the time-space synchronization alignment mechanism maps the RF timing characteristics to millimeter-wave spatial coordinates to enhance target relevance; the final generated RF fingerprint coding sequence integrates multi-dimensional features in space, time and frequency domains, significantly improving the feature recognition and anti-aliasing capabilities of concealed UAVs in complex shielding environments, and providing highly robust identification support for urban low-altitude security.

[0106] 104. Construct a millimeter wave and radio frequency coupling model based on the building material reflection parameter set, and identify hidden drones in the shielded area of ​​urban buildings through cross-validation of the reflection component, the harmonic component, and the radio frequency fingerprint coding sequence.

[0107] Optionally, step 104 may specifically include the following steps:

[0108] 1041. Based on the multi-band attenuation gradient and the spatiotemporal synchronization constraints in the radio frequency fingerprint coding sequence, establish a dynamic correlation matrix of the millimeter wave reflection component intensity and the radio frequency harmonic component amplitude, and generate an initial characteristic mapping relationship of the millimeter wave and radio frequency coupling model according to the dynamic correlation matrix;

[0109] 1042. Perform multimodal feature fusion on the dynamic correlation matrix based on the spatial distribution density and material attenuation difference of the reflectors in the three-dimensional millimeter wave imaging atlas and the initial feature mapping relationship to construct a harmonic component amplitude compensation function under the millimeter wave spatial resolution constraint.

[0110] 1043. Correcting harmonic attenuation distortion in the radio frequency fingerprint code sequence due to building shielding using the harmonic component amplitude compensation function, and extracting the spatiotemporal consistency constraint parameters of the reflection characteristics of the UAV composite material and the radio frequency harmonic component amplitude based on the corrected harmonic attenuation distortion;

[0111] 1044. Based on the spatiotemporal consistency constraint parameters, a multi-dimensional joint verification is performed on the distribution of reflection components in the three-dimensional millimeter wave imaging spectrum, the frequency hopping residual pattern of the RF harmonic component amplitude, and the RF fingerprint coding sequence. The verification process eliminates feature confusion between the metal frame inside the building and the composite material of the UAV, thereby completing the spatial positioning and identification of the urban concealed UAV.

[0112] Among them, step 1044 may specifically include the following processes: based on the spatial distribution correlation between the millimeter wave reflection component and the radio frequency harmonic component in the spatiotemporal consistency constraint parameter, extracting the steady-state reflection characteristics of the metal frame inside the building and the transient scattering characteristics of the composite material of the drone, and generating a joint distribution map of the reflection component and the harmonic component; based on the spatial overlap area of ​​the metal frame inside the building and the composite material of the drone in the joint distribution map, combined with the amplitude frequency hopping residual pattern of the radio frequency harmonic component, constructing a metal reflection suppression function and a composite material enhancement function to generate a feature confusion elimination parameter set; using the feature confusion elimination parameter set to dynamically assign weights to the distribution of the reflection component in the three-dimensional millimeter wave imaging map, combining the assigned reflection component distribution, the time evolution law of the amplitude frequency hopping residual pattern of the radio frequency harmonic component, and the radio frequency fingerprint code sequence, to generate joint verification parameters that integrate spatial positioning constraints and harmonic feature identification; based on the joint verification parameters, performing multi-dimensional iterative optimization on the joint distribution map, the optimization process eliminates feature confusion between the metal frame inside the building and the composite material of the drone, and completes the spatial positioning and identification of urban concealed drones.

[0113] In the above scheme, the dynamic correlation matrix is ​​a nonlinear mapping relationship matrix that characterizes the intensity of the millimeter-wave reflection component and the amplitude of the RF harmonic component, and is used to quantify the coupling characteristics of the two in the spatiotemporal dimension. The harmonic component amplitude compensation function is a function constructed based on the millimeter-wave spatial resolution constraint, which is used to correct the attenuation distortion of the RF harmonic amplitude caused by building shielding. The spatiotemporal consistency constraint parameter is a parameter that describes the correlation between the reflection characteristics of the UAV composite material and the RF harmonic component in the spatiotemporal dimension, and is used to eliminate feature confusion. Multi-dimensional joint verification is a multimodal feature cross-validation mechanism based on the reflection component distribution, frequency hopping residual pattern, and RF fingerprint coding sequence.

[0114] In this embodiment, first, in step 1041, a kernel regression algorithm is used to establish a nonlinear initial feature mapping relationship between the millimeter wave reflection component intensity and the RF harmonic component amplitude based on the multi-band attenuation gradients in the building material reflection parameter set (e.g., the attenuation gradient α = 2.3 dB / m in the 24 GHz band) and the spatiotemporal synchronization constraints in the RF fingerprint code sequence (e.g., coordinates X = 35.2, Y = 78.6, Z = 12.3 and a timestamp T = 100 ms). Specifically, using the millimeter wave reflection intensity as input and the harmonic amplitude as output, a dynamic correlation matrix is ​​calculated using the Gaussian kernel function K(S_i, H_j) = exp(-||S_i - H_j||² / 2σ²), where σ is the bandwidth parameter, S_i is the reflection intensity sample, and H_j is the harmonic amplitude sample. For example, for the drone hovering area (reflection intensity 12 dBm), the dynamic correlation matrix output matches the harmonic amplitude of -28 dBm, generating an initial feature mapping relationship table.

[0115] Subsequently, in step 1042, based on the spatial distribution density and material attenuation differences of reflectors in the millimeter-wave imaging spectrum (e.g., metal attenuation gradient α = 15 dB / m, concrete α = 3 dB / m), a tensor decomposition technique is used to decompose the dynamic correlation matrix into spatial distribution factors, material attenuation factors, and time series factors. Combined with the millimeter-wave spatial resolution constraint, a harmonic component amplitude compensation function H'(f) is constructed. For example, for a concrete-shaded area with a spatial resolution of 0.5 meters, the compensation function H'(f) = H(f)·e^{βΔα} is calculated, where β = 0.8 is the material attenuation difference coefficient and Δα = 1.2 is the spatial resolution attenuation correction. After compensation, the harmonic amplitude is increased by 3 dB.

[0116] At the same time, in step 1043, a compensation function is used to correct the harmonic attenuation distortion caused by building shadowing in the RF fingerprint code sequence. The corrected harmonic amplitude is noise-reduced using an adaptive Kalman filter, and its spatiotemporal correlation with the UAV composite material reflection characteristics (e.g., transient scattering intensity of 8 dB) is extracted. For example, within a time window of T = 100 ms, the sliding window correlation coefficient ρ between the harmonic amplitude and the reflection intensity is calculated to be greater than 0.8, and a spatiotemporal consistency constraint parameter table is generated.

[0117] Finally, in step 1044, based on the spatiotemporal consistency constraint parameter table, a multi-dimensional joint verification is performed on the reflection component distribution, frequency hopping residual pattern, and RF fingerprint coding sequence in the millimeter wave imaging spectrum: the evidence theory is used to fuse the reflection component confidence level of 0.9, the harmonic amplitude confidence level of 0.85, and the frequency hopping pattern confidence level of 0.8. The fuzzy clustering algorithm is used to separate the characteristic confusion areas of the metal frame with a cluster center reflection intensity greater than 18 dBm and the UAV composite material with a cluster center harmonic amplitude greater than -30 dBm. Finally, the spatial coordinates and identification label of the UAV are output.

[0118] It should be noted that the coupling model is a training model, and its training process includes the following steps:

[0119] First, based on the multi-band attenuation gradient in the building material reflection parameter set (such as the attenuation gradient α=2.3dB / m in the 24GHz band) and the spatiotemporal synchronization constraints of the RF fingerprint coding sequence (such as coordinates X=35.2, Y=78.6, Z=12.3 and timestamp T=100ms), the kernel regression algorithm is used to establish the nonlinear dynamic correlation matrix of the millimeter wave reflection component intensity and the RF harmonic component amplitude. Specifically, the millimeter wave reflection intensity is used as input and the harmonic amplitude is used as output. The Gaussian kernel function is used to calculate the dynamic correlation matrix of the millimeter wave reflection component intensity and the RF harmonic component amplitude. ) calculates the mapping relationship (where is the bandwidth parameter, is the reflection intensity sample, is a harmonic amplitude sample), generating an initial feature mapping relationship table (for example, the reflection intensity of 12dBm in the hovering area matches the harmonic amplitude of -28dBm).

[0120] Then, according to the spatial distribution density of reflectors and material attenuation differences (such as metal attenuation gradient) in the three-dimensional millimeter wave imaging map, , concrete ), using tensor decomposition technology to decompose the dynamic correlation matrix into spatial distribution factors, material attenuation factors and time series factors, combined with the millimeter wave spatial resolution constraint (such as 0.5 meter resolution), to construct the harmonic component amplitude compensation function (in is the material attenuation difference coefficient, is the spatial resolution attenuation correction, and the harmonic amplitude is increased by 3 dB after compensation).

[0121] Finally, the corrected harmonic amplitude, reflection component distribution and RF fingerprint coding sequence are used as training data, and the coupling law of millimeter wave and RF signals is learned through a deep learning model (such as a multimodal fusion network). After iterative optimization, the model output and the measured error converge, and the trained model is defined as a coupling model.

[0122] The coupling model is based on a set of building material reflection parameters. It maps the nonlinear relationship between millimeter wave reflection intensity and RF harmonic amplitude through a dynamic correlation matrix (for example, a reflection intensity of 12 dBm corresponds to a harmonic amplitude of -28 dBm), and uses a harmonic component amplitude compensation function to correct the harmonic attenuation distortion caused by building shielding (for example, a 3 dB compensation for the concrete shielding area). Based on the corrected harmonic amplitude, the transient scattering characteristics of the UAV composite material (such as an intensity of 8 dB) and the spatiotemporal consistency constraint parameters of the harmonic amplitude (such as a sliding window correlation coefficient ρ>0.8) are extracted. Combined with the metal reflection suppression function (suppressing steady-state metal frames with reflection intensity>18 dBm) and the composite material enhancement function (enhancing harmonic amplitude>-30 dBm transient frequency hopping residual pattern) to eliminate feature confusion; finally, it fuses the reflection component distribution confidence (0.9), harmonic amplitude confidence (0.85) and the frequency hopping pattern confidence (0.8) of the RF fingerprint coding sequence, and uses a fuzzy clustering algorithm to separate the metal and drone feature areas. The output is the drone's spatial coordinates (such as X=35.2, Y=78.6, Z=12.3) and identification label, achieving covert target recognition with a positioning error of less than 0.5 meters.

[0123] In practical applications, in shielded areas of urban transportation hubs, such as dense steel structure roofs and concrete columns, step 1041 generates a dynamic correlation matrix based on the 60 GHz frequency band attenuation gradient α = 1.8 dB / m and the spatiotemporal constraints X = 22.5, Y = 45.8, Z = 8.2, T = 150 ms in the RF fingerprint coding sequence, mapping the reflection intensity of 10 dBm to the harmonic amplitude of -26 dBm. Step 1042 optimizes the matrix through tensor decomposition, constructs a compensation function to correct the -6 dB harmonic attenuation caused by the concrete columns, and outputs the corrected amplitude of -20 dBm. Step 1043 extracts the spatiotemporal consistency parameter ρ = 0.85 between the harmonic amplitude and the reflection intensity. Step 1044 integrates multimodal confidence to eliminate the reflection interference of the steel structure roof, locates the drone at coordinates X = 22.5, Y = 45.8, Z = 8.2, and marks it as a "high-risk target."

[0124] The overall solution of step 104 above realizes deep coupling of millimeter wave and radio frequency data through dynamic correlation matrix construction and multimodal tensor decomposition, breaking through the perception limitations of a single mode; the harmonic component amplitude compensation function accurately corrects the signal attenuation distortion caused by building shielding and enhances feature consistency; the spatiotemporal consistency constraint parameter integrates the spatiotemporal correlation of reflection and harmonics to solve the target confusion problem under the interference of complex materials; the multi-dimensional joint verification mechanism integrates spatial, temporal and frequency domain features through evidence theory and fuzzy clustering, and ultimately realizes high-confidence positioning and identification of covert drones, providing anti-interference and highly robust technical support for urban low-altitude security.

[0125] The following is a complete embodiment for steps 101 to 104. Figure 3 As shown:

[0126] Suppose that in a city's core business district, where glass curtain walls account for 60% and steel-structured corridors are dense, there are covert drone activities that use building shielding to conduct illegal filming. The reflection characteristics of the metal frames in the area are highly similar to those of the drone's composite materials, and multipath interference is severe.

[0127] In step 101, a drone-mounted 0.1-meter-resolution lidar is used to scan the building complex and generate building geometry data, including building spacing and surface curvature. A 1-100 GHz tunable near-field microwave flaw detector is used to measure the dielectric constant distribution of the glass curtain wall (ε=5.2-j0.7) and the steel structure (ε=1e6-j1e7). Finite-difference time-domain simulation is used to generate multi-band attenuation gradients, such as a 94 GHz glass attenuation gradient of α=1.8 dB / m and a scattering phase offset of Δφ=25°, and output a set of building material reflection parameters.

[0128] Step 102 selects the 94 GHz frequency band with an attenuation gradient of α=1.8 dB / m as the main transmission frequency band, supplemented by the 60 GHz frequency band with a metal attenuation gradient of α=15 dB / m; transmits a frequency-modulated continuous wave signal, and separates the steady-state reflection component of the steel structure and the transient scattering component of the UAV's carbon fiber fuselage through time-frequency ridge tracking. The reflection intensities are 20 dBm and 8 dBm, respectively; constructs multipath interference suppression parameters with a path attenuation coefficient of λ=1.5 dB / m, and generates a three-dimensional millimeter-wave imaging map, marking the UAV hovering behind the glass curtain wall at coordinates X=102.3, Y=55.7, and Z=45.2.

[0129] Step 103 synchronously collects 5.8 GHz RF signals and uses a scattering phase offset of Δφ = 25° to compensate for path loss. For example, a glass curtain wall causes a -10 dB attenuation. Adaptive wavelet packet transform is used to extract the propeller motor harmonic components, with a fundamental frequency of 1.2 kHz and an amplitude of -28 dBm. The frequency hopping residual components are extracted as 5.8 GHz → 2.4 GHz with an interval of 50 ms. Combined with the multipath interference suppression parameters used in millimeter-wave imaging, a spatiotemporal synchronized RF fingerprint code "F_5.8G_50ms_102.3_55.7_45.2" is generated.

[0130] Step 104 constructs a millimeter-wave and radio frequency coupling model, dynamically correlating reflection intensity and harmonic amplitude. DS theory is used to fuse the reflection component, harmonic component, and radio frequency fingerprint coding confidence levels of 0.9 / 0.85 / 0.8. Fuzzy clustering is used to eliminate steel structure interference, such as reflection intensity > 18 dBm. Finally, the drone is identified and located at coordinates X = 102.3, Y = 55.7, and Z = 45.2.

[0131] This application optimizes the millimeter wave frequency band based on the material attenuation gradient, breaking through the limitation of insufficient signal penetration ability of a single frequency band and significantly improving the target separation accuracy in complex building environments; through joint time-frequency analysis and RF fingerprint coding, it effectively suppresses metal structure reflection and multipath propagation interference, and enhances the transient feature capture capability of concealed targets; and integrates the coupling model of millimeter wave spatial resolution and RF timing characteristics to solve the feature confusion problem in non-line-of-sight scenarios and achieve high-precision positioning and robust recognition of drones.

[0132] Figure 2 The present invention provides a structural diagram of a hidden urban drone identification system. Figure 2 As shown, the system includes:

[0133] Acquisition module 21 is used to obtain the geometric shape and material dielectric constant distribution data of the building structures in the urban building complex covered by the drone activity area, and generate a set of building material reflection parameters including multi-band attenuation gradients and scattering phase offsets through a dynamic reflection attenuation model;

[0134] A screening module 22 is configured to screen a millimeter wave frequency band combination that matches the dynamic reflection attenuation characteristics based on the set of building material reflection parameters, transmit millimeter wave signals to the area shielded by the building complex, separate the reflection components of the metal parts and the UAV composite materials based on the echo signals, and generate a three-dimensional millimeter wave imaging map containing multipath interference suppression parameters;

[0135] An extraction module 23 is configured to synchronously collect radio frequency signals from the area shielded by the building complex, compensate for path loss using the scattering phase offset, extract harmonic components of the drone's propeller motor and residual frequency hopping components of the communication module, and combine the radio frequency signals with multipath interference suppression parameters in the three-dimensional millimeter wave imaging spectrum to generate a spatiotemporally synchronized radio frequency fingerprint coding sequence.

[0136] The identification module 24 is used to construct a millimeter wave and radio frequency coupling model based on the building material reflection parameter set, and identify hidden drones in the shielded area of ​​urban buildings through cross-validation of the reflection component, the harmonic component and the radio frequency fingerprint code sequence.

[0137] Figure 2 The urban concealed drone identification system can be implemented Figure 1 The implementation principles and technical effects of the urban covert drone identification method described in the illustrated embodiment are not further elaborated. The specific manner in which each module and unit performs operations in the urban covert drone identification system in the above embodiment has been described in detail in the relevant embodiments of the method and will not be elaborated on here.

[0138] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for identifying hidden drones in cities, characterized in that: include: In urban building clusters covered by drone activity, the geometric morphology and dielectric constant distribution data of building structures are obtained, and a set of building material reflection parameters including multi-band attenuation gradients and scattering phase offsets is generated using a dynamic reflection attenuation model. Based on the set of building material reflection parameters, a millimeter wave frequency band combination that matches the dynamic reflection attenuation characteristics is selected, and a millimeter wave signal is transmitted to the area shielded by the building complex. The reflection components of the metal parts and the UAV composite material are separated according to the echo signal, and a three-dimensional millimeter wave imaging map containing multipath interference suppression parameters is generated; Synchronously collecting radio frequency signals from the area shielded by the building complex, compensating for path loss using the scattering phase offset, extracting harmonic components of the drone's propeller motor and residual frequency hopping components of the communication module, and combining the radio frequency signals with multipath interference suppression parameters in the three-dimensional millimeter wave imaging atlas to generate a spatiotemporally synchronized radio frequency fingerprint coding sequence; A millimeter wave and radio frequency coupling model is constructed based on the building material reflection parameter set, and hidden drones in the shielded area of ​​urban buildings are identified through cross-validation of the reflection component, the harmonic component and the radio frequency fingerprint coding sequence.

2. The method according to claim 1, characterized in that The method of separating the reflection components of the metal parts and the UAV composite materials according to the echo signals and generating a three-dimensional millimeter wave imaging map containing multipath interference suppression parameters includes: Performing a joint time-frequency domain analysis on the echo signal to separate the steady-state reflection component of the metal parts from the transient scattering component of the UAV composite material, and constructing multipath interference suppression parameters based on the attenuation gradient distribution in the building material reflection parameter set; The transient scattering component is parameterized and filtered using the multipath interference suppression parameters to eliminate aliasing reflections caused by multipath propagation in the building complex shielding area. The spatial distribution characteristics of the multipath interference suppression parameters and the UAV composite material are visualized to generate a three-dimensional millimeter wave imaging map.

3. The method according to claim 1, characterized in that The synchronous collection of radio frequency signals in the area shielded by the building complex, compensating for path loss by using the scattering phase offset, and extracting harmonic components of the drone propeller motor and the frequency hopping residual components of the communication module include: Compensating for path loss by using the scattering phase offset, constructing a dynamic attenuation compensation function for the radio frequency signal propagation path, reconstructing a multipath propagation sequence for the radio frequency signal collected in the area shielded by the building complex, and eliminating the scattering phase distortion of the radio frequency signal caused by the building material; The adjusted RF signal is jointly decomposed in the time-frequency domain to separate the harmonic components generated by the UAV propeller motor during operation and the residual frequency hopping components remaining during the frequency hopping switching of the communication module, and the transient harmonic feature set associated with the UAV motion state is extracted.

4. The method according to claim 3, characterized in that Combining the radio frequency signal with the multipath interference suppression parameter in the three-dimensional millimeter wave imaging atlas to generate a spatiotemporally synchronized radio frequency fingerprint coding sequence includes: Establishing a mapping relationship between the radio frequency signal propagation path and the millimeter wave imaging spatial position according to the multipath interference suppression parameters in the three-dimensional millimeter wave imaging atlas, and performing spatiotemporal synchronization alignment on the harmonic component and the frequency hopping residual component according to the mapping relationship; Based on the spatiotemporal synchronization alignment result, a spatiotemporal synchronized radio frequency fingerprint coding sequence is generated by combining the radio frequency signal with the reflector distribution characteristics in the three-dimensional millimeter wave imaging atlas.

5. The method according to claim 3, characterized in that The step of establishing a mapping relationship between the radio frequency signal propagation path and the millimeter wave imaging spatial position according to the multipath interference suppression parameters in the three-dimensional millimeter wave imaging atlas, and performing spatiotemporal synchronization alignment on the harmonic component and the frequency hopping residual component according to the mapping relationship, includes: Based on the multipath interference suppression parameters in the three-dimensional millimeter wave imaging atlas, the reflector distribution characteristics and the millimeter wave signal attenuation coefficient in the three-dimensional millimeter wave imaging atlas are analyzed to establish a geometric topological mapping relationship between the radio frequency signal propagation path and the millimeter wave imaging space grid; Based on the geometric topological mapping relationship and in combination with the time delay and Doppler characteristic information of the radio frequency signal propagation path, a multipath propagation path topology model within the building complex shielding area is constructed, wherein the multipath propagation path topology model associates the reflector position in the millimeter wave imaging space grid with the geometric constraints of the radio frequency signal propagation path; Extracting a timestamp sequence from the transient harmonic feature set and a frequency domain hopping interval of the frequency hopping residual component, and performing spatiotemporal reference alignment on the timestamp sequence and the frequency domain hopping interval based on reflector positions and geometric constraints associated with the multipath propagation path topology model; The update rate of the millimeter wave imaging space grid is used to dynamically interpolate and compensate the spatiotemporal reference alignment result, generate spatiotemporal constraint parameters synchronized with the UAV motion state and communication frequency hopping mode, and complete the spatiotemporal synchronization alignment of the harmonic component and the frequency hopping residual component.

6. The method according to claim 5, characterized in that The method further comprises: constructing a multipath propagation path topology model within the building complex shielding area based on the geometric topology mapping relationship and combining the delay and Doppler characteristic information of the radio frequency signal propagation path. The multipath propagation path topology model associates the reflector position in the millimeter wave imaging space grid with the geometric constraints of the radio frequency signal propagation path, including: Analyze the correlation between the spatial position of the reflector and the millimeter wave signal attenuation coefficient in the geometric topological mapping relationship, and generate initial path propagation parameters by combining the shielding weights of each reflector on the radio frequency signal propagation path in the building complex shielding area; Combining the delay and Doppler characteristic information of the radio frequency signal propagation path and the dynamic impact of the UAV's motion state on the Doppler frequency shift, correcting the delay estimation error in the initial path propagation parameters to generate updated path propagation parameters that include path loss gradient and time-frequency joint constraints; Based on the updated path propagation parameters, ray tracing modeling is performed on the multipath signal propagation paths within the building complex shielding area to generate a set of path clusters that match the positions of reflectors in the millimeter wave imaging space grid; According to the distribution of overlapping areas between the path cluster set and the reflector positions in the millimeter wave imaging space grid, a multipath propagation path topology model is constructed after quantifying the attenuation ratio of the direct path and the multipath reflection path of the RF signal propagation path, and the multipath propagation path topology model is associated with the geometric constraints of the reflector positions in the millimeter wave imaging space grid and the RF signal propagation path.

7. The method according to claim 6, characterized in that The method combines the delay and Doppler characteristic information of the radio frequency signal propagation path, corrects the delay estimation error in the initial path propagation parameters according to the dynamic impact of the UAV motion state on the Doppler frequency shift, and generates updated path propagation parameters containing path loss gradient and time-frequency joint constraints, including: Extracting the Doppler frequency shift dynamic characteristics associated with the UAV's motion state from the time delay and Doppler characteristic information of the radio frequency signal propagation path, and generating a correlation parameter between the Doppler frequency shift and the motion state by combining the mapping relationship between the propeller rotation frequency and the UAV speed; Based on the associated parameters, a dynamic impact model of the UAV's motion acceleration on the multipath propagation delay is established, and the nonlinear relationship between the delay estimation error in the initial path propagation parameters and the UAV's motion acceleration is quantified using the dynamic impact model; Based on the dynamic impact model, dynamically weighting the delay estimation values ​​in the initial path propagation parameters is performed, and combining the weighting results with the loss gradient distribution of radio frequency signals by different materials in the building complex shielding area to generate the path propagation parameters after delay compensation; The path propagation parameters are matrix-fused with the time-frequency joint constraints of the radio frequency signal. The fusion process eliminates the time-frequency aliasing error caused by the UAV's maneuvering flight and generates updated path propagation parameters containing the path loss gradient and the time-frequency joint constraints.

8. The method according to claim 1, characterized in that The method includes: constructing a millimeter wave and radio frequency coupling model based on the building material reflection parameter set, and identifying concealed drones in an area shielded by urban buildings through cross-validation of the reflection component, the harmonic component, and the radio frequency fingerprint code sequence, including: Based on the multi-band attenuation gradient and the spatiotemporal synchronization constraints in the radio frequency fingerprint coding sequence, a dynamic correlation matrix of the millimeter wave reflection component intensity and the radio frequency harmonic component amplitude is established, and an initial characteristic mapping relationship of the millimeter wave and radio frequency coupling model is generated according to the dynamic correlation matrix; According to the spatial distribution density and material attenuation difference of the reflectors in the three-dimensional millimeter wave imaging atlas, the dynamic correlation matrix is ​​subjected to multimodal feature fusion in combination with the initial feature mapping relationship to construct a harmonic component amplitude compensation function under the millimeter wave spatial resolution constraint; The harmonic component amplitude compensation function is used to correct the harmonic attenuation distortion in the radio frequency fingerprint code sequence caused by building shielding, and the reflection characteristics of the UAV composite material and the spatiotemporal consistency constraint parameters of the radio frequency harmonic component amplitude are extracted based on the corrected harmonic attenuation distortion; Based on the spatiotemporal consistency constraint parameters, a multi-dimensional joint verification is performed on the distribution of reflection components in the three-dimensional millimeter wave imaging map, the frequency hopping residual pattern of the RF harmonic component amplitude, and the RF fingerprint coding sequence. The verification process eliminates the feature confusion between the metal frame inside the building and the composite material of the drone, and completes the spatial positioning and identification of concealed drones in the city.

9. The method according to claim 8, characterized in that The multi-dimensional joint verification of the reflection component distribution in the three-dimensional millimeter wave imaging spectrum, the frequency hopping residual pattern of the RF harmonic component amplitude, and the RF fingerprint coding sequence based on the spatiotemporal consistency constraint parameters is performed. The verification process eliminates the feature confusion between the metal frame inside the building and the composite material of the drone, and completes the spatial positioning and identification of the urban concealed drone, including: Based on the spatial distribution correlation between the millimeter wave reflection component and the radio frequency harmonic component in the spatiotemporal consistency constraint parameter, the steady-state reflection characteristics of the metal frame inside the building and the transient scattering characteristics of the composite material of the drone are extracted to generate a joint distribution map of the reflection component and the harmonic component; Based on the spatial overlap area between the metal frame inside the building and the composite material of the drone in the joint distribution map, combined with the amplitude frequency hopping residual pattern of the radio frequency harmonic component, a metal reflection suppression function and a composite material enhancement function are constructed to generate a feature confusion elimination parameter set; Dynamically weighting the distribution of reflection components in the three-dimensional millimeter-wave imaging atlas using the feature confusion elimination parameter set, and combining the distribution of reflection components after the allocation, the time evolution law of the amplitude frequency hopping residual pattern of the radio frequency harmonic component, and the radio frequency fingerprint coding sequence to generate a joint verification parameter that integrates spatial positioning constraints and harmonic feature identification; The joint distribution map is iteratively optimized in multiple dimensions based on the joint verification parameters. The optimization process eliminates feature confusion between the metal frame inside the building and the composite material of the drone, thereby completing the spatial positioning and identification of concealed drones in the city.

10. An urban hidden drone identification system, characterized in that: include: The acquisition module is used to obtain the geometric shape and material dielectric constant distribution data of the building structures in the urban building complex covered by the drone's activity area, and generate a set of building material reflection parameters including multi-band attenuation gradients and scattering phase offsets through a dynamic reflection attenuation model; A screening module is configured to screen a millimeter wave frequency band combination that matches the dynamic reflection attenuation characteristics based on the set of building material reflection parameters, transmit millimeter wave signals to the area shielded by the building complex, separate the reflection components of the metal parts and the UAV composite materials based on the echo signals, and generate a three-dimensional millimeter wave imaging map containing multipath interference suppression parameters; an extraction module for synchronously collecting radio frequency signals from the area shielded by the building complex, compensating for path loss using the scattering phase offset, extracting harmonic components of the drone's propeller motor and residual frequency hopping components of the communication module, and combining the radio frequency signals with multipath interference suppression parameters in the three-dimensional millimeter wave imaging atlas to generate a spatiotemporally synchronized radio frequency fingerprint coding sequence; The identification module is used to construct a millimeter wave and radio frequency coupling model based on the set of building material reflection parameters, and identify hidden drones in the shielded area of ​​urban buildings through cross-validation of the reflection component, the harmonic component and the radio frequency fingerprint coding sequence.

Images