A method and system for identifying drone signals in complex electromagnetic environments

Abstract

The present application discloses a method and system for identifying drone signals in a complex electromagnetic environment, which belongs to the field of drones. The present application first adjusts the liquid metal channel morphology of the liquid metal antenna to adapt to the target frequency band, and omnidirectionally collects regional electromagnetic signals. Based on the phase difference and preset features, spatial distribution data containing the drone and the interference direction is generated. The liquid metal droplet deformation is regulated by electrowetting voltage so that the main lobe of the antenna is aligned with the direction of the drone, and the sidelobe suppression area covers the interference source. Subsequently, the target direction signal set is collected, and the multi-dimensional features of time domain fluctuation, frequency domain distribution, and spatial direction are extracted. The multi-dimensional features are input into the joint probability model constructed by the Bayesian network, and the posterior probability information is output. By using Markov chain Monte Carlo adaptive sampling, iterative convergence calculation of classification confidence, and signal category determination based on the threshold, the present application can improve the accuracy of drone signal recognition in a complex electromagnetic environment.

Description

Technical Field

[0001] The present invention relates to the field of drones, and more particularly to a method and system for identifying drone signals in complex electromagnetic environments. Background Art

[0002] Drone signal recognition technology has significant applications in public safety, disaster relief, military reconnaissance, and other fields. With the development of swarming and intelligent drones, the need for signal recognition in complex electromagnetic environments is becoming increasingly urgent. For example, at large-scale events, it is necessary to quickly isolate drone communication links amidst dense Wi-Fi, Bluetooth, and other interference. Efficient and accurate recognition technology is the core foundation for drone countermeasures, collaborative operations, and covert communications, directly impacting emergency response speed and mission success rates.

[0003] Existing technologies for drone signal recognition in complex electromagnetic environments primarily rely on methods that combine antennas with machine learning. For example, microstrip antennas or antenna arrays are used to collect signals, and drone signal recognition is performed using models such as support vector machines and convolutional neural networks.

[0004] However, existing drone signal recognition methods have difficulty effectively separating target signals from interference in complex electromagnetic environments, such as large-scale event sites and disaster relief sites, due to the presence of dense interference sources such as Wi-Fi, Bluetooth, and mobile phone signals. As a result, existing drone signal recognition methods have a high misjudgment rate and low recognition accuracy in complex electromagnetic environments. Summary of the Invention

[0005] This application provides a method, system, device and computer storage medium for drone signal recognition in complex electromagnetic environments, which can improve the accuracy of drone signal recognition in complex electromagnetic environments.

[0006] In a first aspect, the present application provides a method for identifying drone signals in a complex electromagnetic environment, the method comprising:

[0007] By adjusting the liquid metal channel shape of the liquid metal antenna, the target shape parameters of the liquid metal antenna are reconstructed to collect the omnidirectional electromagnetic signal of the characteristic frequency band of the UAV in the target area;

[0008] Generate spatial distribution data including the direction of the drone signal and the direction of the interference signal based on the phase difference of the electromagnetic signal in the omnidirectional electromagnetic signal and the preset signal characteristics of the drone and the interference signal;

[0009] Based on the spatial distribution data, the electrowetting control voltage parameters are adjusted to control the deformation of the liquid metal droplet of the liquid metal antenna so that the main lobe direction of the liquid metal antenna points in the direction of the drone signal, and the sidelobe suppression area of the liquid metal antenna covers the spatial distribution range of the interference signal direction;

[0010] Using a liquid metal antenna with adjusted lobes, the system collects electromagnetic signals from the direction of the drone's signal and extracts multidimensional signal features from each sub-electromagnetic signal. The multidimensional signal features include time domain fluctuation features, frequency domain distribution features, and spatial domain direction features.

[0011] The multi-dimensional signal characteristics of each sub-electromagnetic signal are input into the constructed joint probability distribution model to obtain the posterior probability information of the electromagnetic signal. The joint probability distribution model is constructed based on historical UAV signals and historical interference signals through a Bayesian network.

[0012] Markov chain Monte Carlo is used to adaptively sample the posterior probability information of electromagnetic signals. The signal classification confidence is obtained through iterative convergence, and the identification information of the sub-electromagnetic signal is determined according to the preset confidence threshold.

[0013] In one possible implementation, before extracting the multi-dimensional signal features of each sub-electromagnetic signal from the electromagnetic signal set, the method further includes:

[0014] Perform blind source separation on the electromagnetic signal set, and separate multiple independent signal source electromagnetic signals based on the signal time-frequency sparsity assumption and independent component analysis;

[0015] Extracting multi-dimensional signal features of each sub-electromagnetic signal from the electromagnetic signal set, including:

[0016] A multi-dimensional signal feature of the electromagnetic signal of each independent signal source is extracted from the electromagnetic signals of multiple independent signal sources.

[0017] In one possible implementation, before generating spatial distribution data including the direction of the drone signal and the direction of the interference signal based on the phase difference of the electromagnetic signal in the omnidirectional electromagnetic signal and the preset signal characteristics of the drone and the interference signal, the method further includes:

[0018] Performing signal demodulation and protocol analysis on each electromagnetic signal in the omnidirectional electromagnetic signal to obtain demodulation protocol information including demodulation information and protocol information of each electromagnetic signal;

[0019] Based on the phase difference of the electromagnetic signal in the omnidirectional electromagnetic signal and the preset signal characteristics of the drone and the interference signal, the spatial distribution data including the direction of the drone signal and the direction of the interference signal is generated, including:

[0020] Based on the phase difference and demodulation protocol information of each electromagnetic signal in the omnidirectional electromagnetic signal, as well as the preset signal characteristics of the drone and interference signal, spatial distribution data including the drone signal direction and the interference signal direction is generated.

[0021] In one feasible embodiment, before controlling the deformation of the liquid metal droplet of the liquid metal antenna by adjusting the electrowetting control voltage parameter according to the spatial distribution data so that the main lobe direction of the liquid metal antenna points in the direction of the drone signal and the side lobe direction of the liquid metal antenna points in the direction of the interference signal, the method further includes:

[0022] Based on the preset historical spatial distribution data, the movement direction of the drone signal is predicted through Kalman filtering;

[0023] According to the movement direction of the UAV signal, the deformation of the liquid metal droplet of the liquid metal antenna is controlled by adjusting the electrowetting control voltage parameters so that the main lobe direction of the liquid metal antenna points to the movement direction of the UAV signal.

[0024] In one feasible embodiment, the target morphological parameters of the liquid metal antenna are reconstructed by adjusting the liquid metal channel morphology of the liquid metal antenna to collect omnidirectional electromagnetic signals of the characteristic frequency band of the UAV in the target area, including:

[0025] Based on the center frequency and bandwidth of the characteristic frequency band of the UAV, the target morphological parameters of the liquid metal channel are determined. The target morphological parameters include the target continuous length and target branching angle of the liquid metal filling path of the liquid metal antenna.

[0026] An electrowetting control voltage sequence is generated according to the target continuous length and target branching angle. By gradually loading the single-stage voltage value in the electrowetting control voltage sequence, the liquid metal droplet of the liquid metal antenna is driven to deform along the filling path until the liquid metal distribution state meets the target continuous length and target branching angle.

[0027] Under the condition that the liquid metal distribution state meets the target continuous length and target branching angle, omnidirectional electromagnetic signals in all directions within the target area are collected.

[0028] In one feasible embodiment, based on the spatial distribution data, the liquid metal droplet deformation of the liquid metal antenna is controlled by adjusting the electrowetting control voltage parameter so that the main lobe direction of the liquid metal antenna points in the direction of the drone signal, and the sidelobe suppression area of the liquid metal antenna covers the spatial distribution range of the interference signal direction, including:

[0029] Based on the UAV signal direction and the interference signal direction in the spatial distribution data, the angle deviation between the main lobe direction of the liquid metal antenna and the UAV signal direction, as well as the spatial coverage deviation between the side lobe suppression area of the liquid metal antenna and the interference signal direction are determined;

[0030] generating a control voltage correction value of the electrowetting control voltage parameter according to the angle deviation and the spatial coverage deviation, wherein the control voltage correction value includes a main lobe direction alignment component and a side lobe suppression enhancement component;

[0031] The control voltage correction value is decomposed into multiple voltage gradient values, and the voltage gradient values are applied to the electrode units of the liquid metal antenna in sequence until the angular difference between the main lobe direction and the drone signal direction is less than the preset difference threshold, and the sidelobe suppression area covers the spatial distribution range of the interference signal direction.

[0032] In one feasible embodiment, Markov chain Monte Carlo is used to adaptively sample the posterior probability information based on the electromagnetic signal, obtain the signal classification confidence through iterative convergence, and determine the identification information of the sub-electromagnetic signal according to a preset confidence threshold, including:

[0033] Determining an initial sampling step size and an initial sampling density based on the posterior probability information of the electromagnetic signal, wherein the initial sampling step size and the initial sampling density correspond to the dimensional distribution of the multi-dimensional signal characteristics;

[0034] Performing probability sampling on the dimensional distribution of the multidimensional signal features within a range defined by an initial sampling step size and an initial sampling density to generate a set of candidate sampling points, the set of candidate sampling points including a combination of signal features corresponding to posterior probability information of the electromagnetic signal;

[0035] Adjust the initial sampling step size and initial sampling density according to the distribution discreteness of the candidate sampling point set, and repeat the probability sampling until the distribution discreteness of the candidate sampling point set is lower than the preset convergence threshold;

[0036] The signal classification confidence of each signal category is calculated based on the proportion of the number of each signal category in the candidate sampling point set. When the signal classification confidence with the largest value is greater than the preset confidence threshold, the signal category corresponding to the signal classification confidence is determined as the identification information of the sub-electromagnetic signal.

[0037] In a second aspect, the present application provides a drone signal recognition system in a complex electromagnetic environment, the system comprising:

[0038] An acquisition module is used to reconstruct the target morphological parameters of the liquid metal antenna by adjusting the liquid metal channel morphology of the liquid metal antenna, and to collect omnidirectional electromagnetic signals of the characteristic frequency band of the UAV in the target area;

[0039] A generation module is used to generate spatial distribution data including the direction of the drone signal and the direction of the interference signal based on the phase difference of the electromagnetic signal in the omnidirectional electromagnetic signal and the preset signal characteristics of the drone and the interference signal;

[0040] An adjustment module is used to control the deformation of the liquid metal droplet of the liquid metal antenna by adjusting the electrowetting control voltage parameters according to the spatial distribution data, so that the main lobe direction of the liquid metal antenna points in the direction of the drone signal, and the sidelobe suppression area of the liquid metal antenna covers the spatial distribution range of the interference signal direction;

[0041] The acquisition module is also used to use the liquid metal antenna with adjusted lobes to collect the electromagnetic signal set in the direction of the drone signal, and extract the multi-dimensional signal characteristics of each sub-electromagnetic signal from the electromagnetic signal set. The multi-dimensional signal characteristics include time domain fluctuation characteristics, frequency domain distribution characteristics, and spatial domain direction characteristics;

[0042] The input module is used to input the multidimensional signal characteristics of each sub-electromagnetic signal into the constructed joint probability distribution model to obtain the posterior probability information of the electromagnetic signal. The joint probability distribution model is constructed using a Bayesian network based on historical UAV signals and historical interference signals;

[0043] The determination module is used to adaptively sample the posterior probability information based on the electromagnetic signal using Markov chain Monte Carlo, obtain the signal classification confidence through iterative convergence, and determine the identification information of the sub-electromagnetic signal according to a preset confidence threshold.

[0044] In a third aspect, the present application provides an electronic device comprising: a processor, and a memory storing computer program instructions; the processor reads and executes the computer program instructions to implement a method for identifying drone signals in a complex electromagnetic environment as in any one of the embodiments of the first aspect.

[0045] In a fourth aspect, the present application provides a computer-readable storage medium having computer program instructions stored thereon. When the computer program instructions are executed by a processor, a method for identifying drone signals in a complex electromagnetic environment as in any one of the embodiments of the first aspect is implemented.

[0046] The present application implements a method, system, device and computer storage medium for drone signal recognition in a complex electromagnetic environment. Through the dynamic morphological reconstruction capability of the liquid metal antenna, it realizes omnidirectional signal adaptive capture of characteristic frequency bands in a complex electromagnetic environment, overcoming the limitations of traditional fixed antenna structures in interference suppression. Based on the spatial distribution modeling of phase difference and signal characteristics, it can separate the propagation direction of drone signals and dense interference sources from the spatial dimension, thereby enhancing the spatial directivity of the target signal. The electrowetting control technology is combined to dynamically optimize the antenna lobe pattern, so that the main lobe accurately tracks the drone signal while forming an interference shielding area by using sidelobe suppression, effectively improving the signal-to-noise ratio of signal acquisition. By integrating multi-dimensional feature analysis in the time domain, frequency domain and spatial domain, and combining the joint probability model constructed by the Bayesian network, it can distinguish the essential differences of signals from the statistical characteristics level. Finally, the adaptive sampling algorithm is used to dynamically optimize and iterate the posterior probability, solving the problem of blurred classification boundaries of traditional machine learning models in complex electromagnetic coupling environments, thereby significantly reducing the misjudgment rate and improving the accuracy and anti-interference capability of drone signal recognition in complex scenarios.

[0047] Furthermore, by adding signal demodulation and protocol analysis steps before spatial distribution modeling, the electromagnetic signal can be feature deconstructed from the communication protocol level, effectively identifying the differences between drone-specific communication protocols and interference protocols such as Wi-Fi / Bluetooth; combining the modulation type, coding format and protocol interaction characteristics in the demodulation information, the protocol fingerprint matching degree of the drone signal can be enhanced during the airspace positioning process, avoiding the problem of traditional phase difference direction finding being misled by interference signals in similar frequency bands; by fusing the physical layer signal characteristics and the protocol layer semantic information, spatial distribution data with protocol identification capabilities is constructed, so that non-target protocol interference can be excluded based on multi-dimensional features during subsequent antenna beamforming; ultimately, in the scenario of dense heterogeneous signal coexistence, the reliability of drone signal spatial positioning and the ability to resist protocol confusion are significantly improved, providing a dual verification mechanism for accurate identification in complex electromagnetic environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0049] Figure 1 This is a flow chart of a method for identifying drone signals in a complex electromagnetic environment provided by an embodiment of the present application;

[0050] Figure 2 This is a flow chart of a method for adjusting a liquid metal antenna provided by one embodiment of the present application;

[0051] Figure 3 This is a flowchart of a method for determining identification information of a drone signal provided by an embodiment of the present application;

[0052] Figure 4 This is a schematic diagram of the structure of a drone signal recognition system in a complex electromagnetic environment provided by an embodiment of the present application;

[0053] Figure 5 This is a schematic diagram of the hardware structure of an electronic device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0054] The features and exemplary embodiments of various aspects of the present application will be described in detail below. In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain the present application, rather than to limit the present application. For those skilled in the art, the present application can be implemented without the need for some of these specific details. The following description of the embodiments is merely to provide a better understanding of the present application by illustrating the examples of the present application.

[0055] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, the elements defined by the phrase "comprising..." do not exclude the presence of other identical elements in the process, method, article, or device comprising the elements.

[0056] Prior art methods for identifying drone signals in complex electromagnetic environments primarily rely on antennas combined with machine learning. For example, microstrip antennas or array antennas are used to collect signals, and drone signal identification is performed using models such as support vector machines and convolutional neural networks. However, existing drone signal recognition methods struggle to effectively separate target signals from interference in complex electromagnetic environments, such as at large-scale events and disaster relief sites, due to the dense presence of interference sources such as Wi-Fi, Bluetooth, and mobile phone signals. This results in a high misjudgment rate and low recognition accuracy in these environments.

[0057] To address the problems of the prior art, the present invention provides a method, system, device, and computer storage medium for identifying drone signals in a complex electromagnetic environment. The following first introduces the method for identifying drone signals in a complex electromagnetic environment provided by the present invention.

[0058] Figure 1 The following is a flow chart of a method for identifying drone signals in a complex electromagnetic environment provided by an embodiment of the present application. Figure 1 As shown, steps S110 to S160 are included.

[0059] S110: reconstructing target morphological parameters of the liquid metal antenna by adjusting the liquid metal channel morphology of the liquid metal antenna, and collecting omnidirectional electromagnetic signals of characteristic frequency bands of the UAV in the target area.

[0060] A liquid metal antenna is a programmable antenna device based on the dynamic reconfiguration properties of liquid metal fluid. The liquid metal inside it changes its distribution morphology under the influence of an electric field to dynamically adjust its radiation characteristics. Liquid metal channel morphology refers to the spatial arrangement of the liquid metal flow path defined by the antenna's microfluidic structure. The antenna's resonant frequency is adjusted by varying the liquid metal filling length and branching angle within the channel. Target morphology parameters are pre-calculated liquid metal distribution parameters to match the drone's characteristic frequency band. These parameters include the target continuous length of the liquid metal filling path and the target branching angle. The target continuous length refers to the total length of the liquid metal filling path within the microfluidic channel, and the target branching angle refers to the geometric angle at the microchannel bifurcation. The target area refers to the three-dimensional spatial range to be monitored, typically the airspace where drones are likely to operate, such as at large-scale event sites and disaster relief sites. The drone's characteristic frequency band refers to the specific frequency range used by drone remote control, image transmission, and other communication links, such as the 5.8 GHz or 2.4 GHz sub-bands of the ISM band. Omnidirectional electromagnetic signals refer to the original electromagnetic radiation data of all directions and all frequencies in the target area obtained through the omnidirectional receiving mode of the antenna.

[0061] First, the center frequency and bandwidth of the target drone's characteristic frequency band are determined based on the communication protocol. The target morphological parameters, namely the target continuous length and target branching angle, are determined based on the correspondence between the preset center frequency and bandwidth of the drone's characteristic frequency band and the morphological parameters of the liquid metal antenna. The correspondence between the preset center frequency and bandwidth of the drone's characteristic frequency band and the morphological parameters of the liquid metal antenna is established using an electromagnetic simulation model. Subsequently, an electrowetting control voltage sequence is generated, consisting of multiple voltage gradient values, each corresponding to the voltage driving the liquid metal droplet's deformation in a specific region of the electrode array. By gradually applying the voltage gradient values, the electrowetting effect is used to control the movement of the liquid metal droplet along the microchannel filling path, and the liquid metal distribution is monitored in real time until its continuous length and branching angle reach the target parameters. The electrowetting effect refers to the physical phenomenon in which the interfacial tension between the liquid metal and the substrate material is changed by applying a voltage, thereby driving the liquid metal droplet to deform or displace within the microchannel. Finally, in the form of the reconstructed liquid metal antenna, electromagnetic signals in all directions within the target area are collected in omnidirectional scanning mode to form an omnidirectional electromagnetic signal dataset including time domain waveforms, frequency domain spectrum lines and spatial direction information.

[0062] In one embodiment, the liquid metal antenna may be a multi-channel liquid metal antenna array.

[0063] S120: Generate spatial distribution data including the direction of the drone signal and the direction of the interference signal based on the phase difference of the electromagnetic signal in the omnidirectional electromagnetic signal and the preset signal characteristics of the drone and the interference signal.

[0064] The signal characteristics of drones and interference signals refer to a set of differentiated features extracted from historical data between drone communication signals and common interference sources. These include modulation methods, spectrum occupancy characteristics, signal periodicity, and protocol interaction patterns. Common interference sources can include Wi-Fi, Bluetooth, and mobile phone signals. Spatial distribution data is a set of spatial coordinate information representing the direction of the electromagnetic signal's source. It includes the three-dimensional angular coordinates of the drone signal's direction and the azimuth-elevation distribution range of the interference signal's direction. It is used to describe the spatial positional relationships between different signal sources within the target area.

[0065] First, a multi-channel phase difference analysis is performed on the omnidirectional electromagnetic signal. The signal arrival time difference is calculated using the phase difference data received by the array antenna, and the incident direction angle of each electromagnetic signal is estimated by combining the multi-channel phase interferometer algorithm. Subsequently, the signal's time-frequency domain features, including the instantaneous frequency, modulation depth, and spectrum spread characteristics, are extracted and matched with a pre-stored database of drone signal features to screen out candidate drone signals. Simultaneously, the interference type of the remaining signals is determined using a library of interference signal features. The library can include Wi-Fi beacon frame features, Bluetooth frequency hopping patterns, and so on. Finally, the incident direction of the candidate drone signal is fused with the azimuth information of the interference signal to construct a spatial distribution matrix with azimuth and pitch angle as coordinates, marking the main direction of the drone signal and the spatial range of the interference signal-intensive area.

[0066] For example, in the drone signal identification scenario at a large-scale event, the omnidirectional electromagnetic signal dataset contains multiple signals in the 5.8GHz frequency band. Phase difference data is collected using a four-channel receiving array of a liquid metal antenna, and the time delay difference of each signal in the X / Y axis direction is calculated. The azimuth and pitch angles of the signal are obtained using a direction of arrival estimation algorithm. For time-frequency feature analysis, the short-time Fourier transform spectrum and cyclostationary characteristics of the signal are extracted and matched with the orthogonal frequency division multiplexing features in the drone signal feature library to identify two candidate drone signals. At the same time, 12 signals were detected with the fixed time interval characteristics of Wi-Fi beacon frames and were determined to be interference signals. Ultimately, the drone signal direction was annotated as 120 degrees in azimuth and 30 degrees in pitch, and the interference signal direction was clustered into a high-density area with an azimuth of 60-90 degrees and a pitch of 0-20 degrees.

[0067] S130: According to the spatial distribution data, the liquid metal droplet deformation of the liquid metal antenna is controlled by adjusting the electrowetting control voltage parameters so that the main lobe direction of the liquid metal antenna points to the direction of the drone signal, and the side lobe suppression area of the liquid metal antenna covers the spatial distribution range of the interference signal direction.

[0068] Electrowetting control voltage parameters are a set of voltage parameters used to control liquid metal deformation. These parameters include voltage amplitude, application timing, and spatial distribution pattern. By adjusting the voltage combination of the electrode array, the flow path and morphology of the droplets can be precisely controlled. Liquid metal droplets refer to discrete or continuous units of liquid metal fluid within the antenna microchannel. Their deformation behavior directly affects the antenna's radiation pattern characteristics, such as mainlobe directivity and sidelobe suppression.

[0069] First, based on the UAV signal and interference signal directions in the spatial distribution data, the angular deviation between the current liquid metal antenna's main lobe direction and the target direction, as well as the spatial coverage deviation between the sidelobe suppression area and the interference direction, are calculated. The target electrowetting control voltage parameters, i.e., the control voltage correction values containing the mainlobe direction alignment component and the sidelobe suppression enhancement component, are obtained by mapping the angular deviation, spatial coverage deviation, and electrowetting control voltage parameters. The mapping relationship between the angular deviation, spatial coverage deviation, and electrowetting control voltage parameters is established using an electromagnetic field simulation model. The correction values are decomposed into multiple voltage gradient values, which are applied in stages to the corresponding electrodes according to the spatial arrangement order of the electrode array, using the electrowetting effect to drive the gradual deformation of the liquid metal droplet. The mainlobe pointing accuracy and sidelobe suppression effect of the antenna pattern are detected in real time, and the voltage gradient is iteratively adjusted until the mainlobe direction error is less than a preset threshold and the sidelobe suppression range completely covers the interference area.

[0070] S140: Using the liquid metal antenna with adjusted lobes, collect the electromagnetic signal set in the direction of the drone signal, and extract the multi-dimensional signal features of each sub-electromagnetic signal from the electromagnetic signal set. The multi-dimensional signal features include time domain fluctuation features, frequency domain distribution features, and spatial domain direction features.

[0071] An electromagnetic signal set refers to a dataset of electromagnetic signals collected by a liquid metal antenna with a modified beamformation in the direction of the drone's main lobe. It contains the time-domain waveforms, frequency-domain spectra, and spatial positioning information of multiple signals in the target direction. A sub-electromagnetic signal is a single electromagnetic signal entity within the signal set that has been separated and processed, corresponding to an independent signal source or a resolvable component within a mixed signal.

[0072] First, a liquid metal antenna with adjusted lobes is used to collect directional signals in the direction of the drone's main lobe. Beamforming technology enhances signal reception sensitivity in the target direction and suppresses interference from sidelobe sources, forming an electromagnetic signal set with a high signal-to-noise ratio. Subsequently, the signal set undergoes preprocessing, including signal normalization and noise filtering, to eliminate the noise floor introduced by residual interference from the antenna pattern's sidelobes. For each preprocessed electromagnetic signal, a time-domain fluctuation analysis is performed to extract the signal envelope's variance characteristics and zero-crossing rate. A short-time Fourier transform is then performed to obtain frequency-domain energy distribution characteristics, including the main frequency offset and out-of-band attenuation slope. Combined with the antenna array's direction-of-arrival estimation results, the time-varying stability parameters of the signal's azimuth and elevation angles are calculated to form spatial directional characteristics. Finally, the time-domain, frequency-domain, and spatial-domain feature vectors are aligned and integrated according to the signal source dimension to construct a multidimensional signal feature matrix. Multidimensional signal features are a set of cross-domain features extracted from a single-channel signal. The time domain fluctuation features characterize the change pattern of the signal amplitude over time, including the peak fluctuation rate and envelope fluctuation characteristics; the frequency domain distribution features describe the distribution pattern of the signal energy within the frequency band, such as the main lobe bandwidth and the harmonic component intensity; the spatial domain direction features reflect the spatial directionality of the signal source, including the azimuth stability and the pitch angle change rate.

[0073] For example, in a drone signal recognition scenario at a large-scale event, the main lobe of a liquid metal antenna, after lobe adjustment, points toward interference sources at 120 degrees in azimuth and 30 degrees in elevation, with sidelobe suppression covering an azimuth angle of 60-90 degrees. Directional acquisition yields a set of five electromagnetic sub-signals: two drone signals and three residual interference signals. Time domain analysis of each signal extracts the periodic pulse envelope characteristics of the drone signal, while the interference signal exhibits random amplitude fluctuations. Frequency domain analysis reveals that the drone signal has a narrowband spectrum of 5.8 GHz ± 10 MHz, while the interference signal exhibits broadband spectrum characteristics. Spatial domain analysis, through array phase difference calculation, confirms that the drone signal's azimuth angle fluctuation range is less than ± 2 degrees, while the interference signal exhibits azimuth angle jumps of ± 15 degrees. The resulting multidimensional feature matrix shows that the drone signal exhibits a combination of high temporal periodicity, a narrowband spectrum, and stable spatial pointing direction, while the interference signal exhibits random temporal fluctuations, a broadband spectrum, and spatial jumps.

[0074] S150: Input the multidimensional signal features of each sub-electromagnetic signal into the constructed joint probability distribution model to obtain the posterior probability information of the electromagnetic signal. The joint probability distribution model is constructed based on historical UAV signals and historical interference signals through a Bayesian network.

[0075] The joint probability distribution model is a statistical model based on a Bayesian network. It describes the conditional probability relationship between multidimensional signal features and signal categories. Its network nodes represent feature variables and category labels of different dimensions, while edges represent dependencies between features. Signal categories can include drone signals and interference signals. Electromagnetic signal posterior probability information refers to the probability distribution data of a signal belonging to the drone or interference category, given multidimensional signal features. This includes the probability value and confidence interval for each category, and is used to quantify the confidence level of signal classification.

[0076] The multidimensional signal features of each sub-electromagnetic signal are input into the constructed joint probability distribution model. For the multidimensional feature vector of each input sub-electromagnetic signal, its joint probability value under the drone category and the interference category is calculated through the Bayesian network inference algorithm. Combined with the prior probability, Bayesian update is performed, and finally the posterior probability information is output, including the drone signal probability, the interference signal probability and the probability ratio of the two. Among them, the prior probability can be the probability of drone appearance in the target area.

[0077] For example, two candidate drone signals collected at a large-scale event had multi-dimensional features, including a time-domain periodic pulse envelope (Feature A), a 5.8GHz±10MHz narrowband spectrum (Feature B), and an azimuth stability of ±2 degrees (Feature C). After inputting the feature vectors into the joint probability distribution model, the Bayesian network calculated the posterior probability of the signal belonging to the drone category as 0.92 and the probability of the signal belonging to the interference category as 0.08, based on the high conditional probability correlation between drone signal features A and B in historical data. For the other signal, due to random fluctuations in the time domain (Feature D), a wideband spectrum (Feature E), and azimuth jumps (Feature F), the calculated probability of the signal belonging to the drone category was 0.15 and the probability of the signal belonging to the interference category was 0.85.

[0078] In one embodiment, before step S150: inputting the multi-dimensional signal features of each sub-electromagnetic signal into the constructed joint probability distribution model to obtain the posterior probability information of the electromagnetic signal, the method further includes:

[0079] Obtain a training sample set, which includes multiple training samples, each training sample includes drone signal information, interference signal information and corresponding true signal category labels in historical data; for each training sample, perform the following steps respectively: input the drone signal information and interference signal information in each training sample into a preset joint probability distribution model to obtain a predicted signal category label; determine the loss function value of the joint probability distribution model based on the true signal category label and the predicted signal category label; when the loss function value does not meet the training stop condition, adjust the model parameters of the joint probability distribution model to obtain an updated joint probability distribution model, and return to input the drone signal information and interference signal information into the preset joint probability distribution model to obtain the signal category label, until the training stop condition is met, and a trained joint probability distribution model is obtained.

[0080] The joint probability distribution model is constructed using a Bayesian network topology. Feature correlation analysis is used to determine the conditional dependencies between features in each dimension, such as the joint distribution constraints between spatial directional features and frequency domain features. Maximum likelihood estimation is used to learn the conditional probability tables of network nodes, and historical data is used to train the joint probability distribution model parameters.

[0081] S160: Adopting Markov Chain Monte Carlo to adaptively sample the posterior probability information based on the electromagnetic signal, obtaining the signal classification confidence through iterative convergence, and determining the identification information of the sub-electromagnetic signal according to a preset confidence threshold.

[0082] Markov Chain Monte Carlo is a random sampling method based on Markov chains. It iteratively samples in probability space by constructing a state transition probability matrix. It is used to obtain sample sets from complex probability distributions to approximate statistical quantities. The preset confidence threshold is a pre-set classification decision threshold. When the signal classification confidence exceeds this threshold, it is classified as belonging to the corresponding category. For example, the confidence threshold for drone signals is set to 0.8. Below this threshold, it is considered an interference signal. Identification information is the finalized signal category label, including drone signals, interference signals, or unidentified signal types.

[0083] First, the initial sampling step size and sampling density of the multidimensional feature space are determined based on the posterior probability information of the electromagnetic signal. The initial step size is determined by the variance of the posterior probability distribution, and the sampling density is dynamically adjusted based on the number of feature dimensions. Within the initial parameter range, Markov Chain Monte Carlo sampling is performed to generate a set of candidate sampling points, each of which represents a possible combination of signal features. The convergence state is evaluated based on the discreteness of the distribution of the sampling points. If the discreteness exceeds a preset convergence threshold, the sampling step size is reduced and the sampling density is increased. The sampling process is repeated until the distribution of the sampling points stabilizes. Finally, the proportion of samples in each category in the candidate sampling point set is calculated, and the category with the highest proportion is used as the signal classification confidence. If this confidence exceeds a preset confidence threshold, the corresponding category label is output as identification information.

[0084] This embodiment uses the dynamic morphology reconstruction capability of the liquid metal antenna to achieve omnidirectional signal adaptive capture in characteristic frequency bands in complex electromagnetic environments, overcoming the limitations of traditional fixed antenna structures in interference suppression. Based on the spatial distribution modeling of phase difference and signal characteristics, it can separate the propagation direction of drone signals and dense interference sources from the spatial dimension, enhancing the spatial directivity of the target signal. The electrowetting control technology is combined to dynamically optimize the antenna lobe pattern, so that the main lobe accurately tracks the drone signal while using sidelobe suppression to form an interference shielding area, effectively improving the signal-to-noise ratio of signal acquisition. By integrating multi-dimensional feature analysis in the time domain, frequency domain and spatial domain, and combining it with a joint probability model constructed by a Bayesian network, it can distinguish the essential differences in signals from the statistical characteristics level. Finally, the adaptive sampling algorithm is used to dynamically optimize and iterate the posterior probability, solving the problem of blurred classification boundaries of traditional machine learning models in complex electromagnetic coupling environments, thereby significantly reducing the misjudgment rate and improving the accuracy and anti-interference capability of drone signal recognition in complex scenarios.

[0085] In one feasible embodiment, before extracting the multi-dimensional signal features of each sub-electromagnetic signal from the electromagnetic signal set in step S140, the method further includes:

[0086] Blind source separation is performed on the electromagnetic signal set, and based on the signal time-frequency sparsity assumption and independent component analysis, multiple independent signal source electromagnetic signals are separated.

[0087] The signal time-frequency sparsity assumption states that the time-frequency energy distributions of different signal sources in the joint time-frequency domain exhibit non-overlapping or low-overlapping characteristics. This assumption assumes that at any time and frequency point, only one signal source occupies the dominant energy, while the contributions of other signal sources are negligible. In complex electromagnetic environments, drone signals and interfering signals exhibit locally sparse short-term spectra due to differences in modulation methods and protocols. For example, the periodic burst spectrum of drone image transmission signals and the continuous broadband spectrum of Wi-Fi signals exhibit alternating occupancy characteristics on the time-frequency plane. Independent component analysis is a blind source separation method whose core assumption is that the source signals in a mixed signal are statistically independent. By finding a linear transformation matrix that maximizes the independence of the output signals, the original independent signal sources can be recovered.

[0088] First, the electromagnetic signal set collected after lobe adjustment is transformed in time-frequency mode to generate a time-frequency energy distribution matrix. Based on the assumption of signal time-frequency sparsity, the short-time Fourier transform is used to calculate the spectral energy in each time window, and a time-frequency domain mixed signal model is constructed. The time-frequency energy matrix is decomposed using a non-negative matrix decomposition algorithm to obtain the basis vectors and activation coefficient matrices representing different signal sources, and the signal components with significant time-frequency sparsity are preliminarily separated. Subsequently, independent component analysis is applied to unmix the time-domain mixed signal, and the separation matrix is optimized using a fixed-point iterative algorithm to maximize the non-Gaussianity of the output signal and further separate statistically independent signal sources. Ultimately, multiple independent signal source electromagnetic signals are output, each of which corresponds to a single physical emission source or a group of signals with the same modulation characteristics.

[0089] Extracting the multidimensional signal features of each sub-electromagnetic signal from the electromagnetic signal set includes: extracting the multidimensional signal features of each independent signal source electromagnetic signal from multiple independent signal source electromagnetic signals.

[0090] For each separated electromagnetic signal from an independent signal source, feature extraction is performed in the time, frequency, and spatial domains. Time-domain analysis uses the Hilbert transform to extract the signal envelope, calculating its variance and zero-crossing rate as fluctuation characteristics. Frequency-domain analysis uses power spectral density estimation to determine the mainlobe bandwidth and harmonic component proportions. Spatial-domain analysis combines the direction-of-arrival (DOA) estimation results of the liquid metal antenna array to calculate the mean and standard deviation of the signal's azimuth and elevation angles. The three types of features are aligned according to the signal source dimensions to construct a multidimensional signal feature vector, which serves as input for the subsequent classification model.

[0091] For example, in a drone signal recognition scenario at a large-scale event, a liquid metal antenna with adjusted lobe patterns collected a mixed electromagnetic signal set, including drone image transmission signals, Wi-Fi hotspot signals, and Bluetooth device signals. First, a short-time Fourier transform (SFT) was performed on the mixed signal to generate a time-frequency matrix. Periodic high-energy pulses (drone signals) and continuous broadband energy regions (Wi-Fi signals) were observed in the 5.8 GHz band. Non-negative matrix factorization was used to isolate three sets of basis vectors, corresponding to the typical time-frequency patterns of the three signal sources. Independent component analysis was then applied to demix the time domain signals, and the kurtosis maximization criterion was used to separate two independent signals: one with high-kurtosis pulse modulation characteristics (drone signals) and the other with low-kurtosis continuous characteristics (Wi-Fi signals). The separated drone signal features were extracted, including a time domain envelope variance of 0.12, a frequency domain main lobe bandwidth of 12 MHz, and a spatial azimuth standard deviation of 1.5 degrees. The interference signal had a time domain variance of 0.35, a bandwidth of 40 MHz, and an azimuth standard deviation of 8 degrees. This process ensures that the signal features processed by the subsequent joint probability distribution model originate from physically separated independent signal sources, thereby improving classification accuracy.

[0092] In one feasible embodiment, before step S120: generating spatial distribution data including the direction of the drone signal and the direction of the interference signal based on the phase difference of the electromagnetic signal in the omnidirectional electromagnetic signal and the preset signal characteristics of the drone and the interference signal, the method further includes:

[0093] Signal demodulation and protocol analysis are performed on each electromagnetic signal in the omnidirectional electromagnetic signal to obtain demodulation protocol information including demodulation information and protocol information of each electromagnetic signal.

[0094] Signal demodulation is the process of converting a received modulated signal into a baseband signal. Its core is extracting the original information waveform from a high-frequency carrier. Specifically, this involves identifying the signal's modulation method, symbol rate, and carrier frequency, and recovering the baseband data stream through coherent or incoherent demodulation algorithms. Modulation methods can include FSK, QPSK, and others. Protocol parsing involves performing protocol-layer analysis on the demodulated baseband signal to extract communication protocol features, including frame structure, synchronization header format, checksum mechanism, and interaction timing. Demodulation protocol information is the combined output of signal demodulation and protocol parsing, and includes the demodulated symbol sequence, modulation type identifier, protocol type identifier, and key protocol fields. Key protocol fields can include device addresses and frame types. Demodulation protocol information is used to correlate signal physical layer characteristics with higher-level protocol behavior. For example, it can distinguish drone control signals from Wi-Fi data frames by protocol type.

[0095] First, adaptive demodulation is performed on each electromagnetic signal in the omnidirectional electromagnetic signal. Based on the signal's spectral characteristics and instantaneous amplitude distribution, the modulation scheme is identified using a maximum likelihood estimation algorithm, and the symbol clock is recovered using phase-locked loop synchronization technology. Demodulation parameters are dynamically configured for different modulation types, such as using a frequency discriminator for FSK signals and quadrature coherent demodulation for QPSK signals. After demodulation, a baseband symbol sequence and modulation parameters, such as symbol rate and frequency offset, are output. Protocol parsing is then performed on the baseband symbol sequence, and pattern matching is performed using a protocol signature library. The protocol signature library stores frame structure templates for common communication protocols, such as Wi-Fi 802.11n and drone image transmission protocols. The synchronization header pattern in the baseband sequence is detected using a sliding window to determine the protocol type and extract the protocol field in the payload. Finally, the demodulation parameters and protocol field are combined to generate demodulation protocol information, including the modulation scheme, symbol rate, protocol type, and key frame structure parameters.

[0096] Step S120: Generate spatial distribution data including the direction of the drone signal and the direction of the interference signal based on the phase difference of the electromagnetic signal in the omnidirectional electromagnetic signal and the signal characteristics of the preset drone and interference signal, including: generate spatial distribution data including the direction of the drone signal and the direction of the interference signal based on the phase difference and demodulation protocol information of each electromagnetic signal in the omnidirectional electromagnetic signal and the signal characteristics of the preset drone and interference signal.

[0097] First, based on the phase difference of the received signals from a multi-channel liquid metal antenna array, a multiple signal classification algorithm is used to calculate the direction of arrival (DOA) of each electromagnetic signal, obtaining preliminary azimuth and elevation angle estimates. Subsequently, the demodulated protocol information is matched against a pre-defined library of drone signal features. If the frame structure characteristics of the drone-specific protocol are detected and the modulation parameters match historical drone signal patterns, the signal is marked as a candidate drone signal. Otherwise, the signal is identified as an interference signal based on a library of interference signal features, such as Wi-Fi beacon frame interval characteristics. For candidate drone signals, the direction estimate is weighted and corrected by calculating the similarity between their signal features (such as time-domain pulse period and frequency-domain harmonic distribution) and the drone feature library to reduce the azimuth error caused by multipath effects. For interference signals, the directions of similar interference sources are clustered based on the protocol analysis results, and their spatial distribution density is calculated. Finally, the corrected direction information and class labels of all signals are integrated to generate spatial distribution data, including the three-dimensional directional coordinates of the drone signal and the directional clustering areas of the interference signals.

[0098] For example, in a drone signal identification scenario at a large-scale event, the omnidirectional electromagnetic signal consists of multiple 5.8 GHz band signals. Each signal is first demodulated, and one is identified as QPSK modulated with a symbol rate of 2 Msps. Protocol analysis detects that its frame structure conforms to the drone image transmission protocol, with a synchronization header of 0xAA55 and a drone serial number field, marking it as a candidate drone signal. Another signal is demodulated as OFDM modulated, and protocol analysis matches it with Wi-Fi beacon frame features, identifying it as an interference signal. Phase difference calculation determines the initial direction of the drone signal to be 125 degrees in azimuth and 28 degrees in elevation, but with fluctuations of ±5 degrees. Combining the protocol matching results with the time-domain pulse characteristics of historical drone signals (10ms period), the corrected direction is 120 degrees in azimuth and 30 degrees in elevation, reducing the fluctuation range to ±2 degrees. Clustering of interference signal directions reveals a dense distribution in the 60-90 degree azimuth and 0-20 degree elevation regions. Finally, in the generated spatial distribution data, the drone signal direction accurately points to the corrected coordinates, and the interference signal is marked as a high-density area, providing a spatial suppression basis for subsequent beamforming.

[0099] By adding signal demodulation and protocol analysis steps before spatial distribution modeling, this implementation can deconstruct the characteristics of electromagnetic signals from the communication protocol level, effectively identifying the differential characteristics of drone-specific communication protocols and interference protocols such as Wi-Fi / Bluetooth; combining the modulation type, coding format and protocol interaction characteristics in the demodulation information, the protocol fingerprint matching degree of the drone signal can be enhanced during the airspace positioning process, avoiding the problem of traditional phase difference direction finding being misled by interference signals in similar frequency bands; by fusing the physical layer signal characteristics and protocol layer semantic information, spatial distribution data with protocol identification capabilities is constructed, so that non-target protocol interference can be excluded based on multi-dimensional characteristics during subsequent antenna beamforming; finally, in the scenario of dense heterogeneous signal coexistence, the reliability of drone signal spatial positioning and the ability to resist protocol confusion are significantly improved, providing a dual verification mechanism for accurate identification in complex electromagnetic environments.

[0100] In one feasible embodiment, before step S130: controlling the deformation of the liquid metal droplet of the liquid metal antenna by adjusting the electrowetting control voltage parameter according to the spatial distribution data so that the main lobe direction of the liquid metal antenna points in the direction of the drone signal and the side lobe direction of the liquid metal antenna points in the direction of the interference signal, the method further includes:

[0101] Based on preset historical spatial distribution data, the Kalman filter predicts the movement direction of drone signals. Historical spatial distribution data is a collection of drone signal direction information recorded over a period of time. It includes parameters such as timestamps, three-dimensional azimuth and pitch angle coordinates, and signal strength. It is used to describe the movement trajectory and dynamic changes of drone signals in the airspace.

[0102] First, construct the state vector and observation vector for the Kalman filter. The state vector contains the dynamic parameters of the drone's signal direction, such as azimuth, pitch angle, and their first-order derivatives, such as angular velocity. The observation vector consists of the drone's directional coordinates from the current spatially distributed data. Initialize the Kalman filter parameters, including the state transfer matrix, process noise covariance, and observation noise covariance. The state transfer matrix describes the linear relationship between azimuth and pitch angle over time, the process noise covariance characterizes the uncertainty of the motion model, and the observation noise covariance reflects the directional measurement error.

[0103] The Kalman filter algorithm uses a historical spatially distributed data sequence as input in chronological order. At each time step, the prediction phase predicts the current drone direction based on the previous state estimate and the state transition matrix. The update phase calculates the Kalman gain and modifies the prediction value based on the current measured direction, outputting the optimal estimated direction. After iteratively processing all historical data, the latest state estimate is used to extrapolate the drone's signal movement direction at future moments, including the rate of change of azimuth and pitch angles, to form a predicted direction.

[0104] According to the movement direction of the UAV signal, the deformation of the liquid metal droplet of the liquid metal antenna is controlled by adjusting the electrowetting control voltage parameters so that the main lobe direction of the liquid metal antenna points to the movement direction of the UAV signal.

[0105] Based on the predicted movement direction of the drone signal, the angular deviation between the current main lobe of the liquid metal antenna and the predicted direction is calculated. The angular deviation consists of an azimuth deviation component and a pitch deviation component, corresponding to the differences in horizontal and vertical directions, respectively. By mapping the angular deviation with the electrowetting control voltage parameters, the target electrowetting control voltage parameters are obtained, namely, voltage correction values containing azimuth and pitch adjustment components. The mapping between the angular deviation and the electrowetting control voltage parameters is established using an electromagnetic field simulation model. The voltage correction values are decomposed into multiple voltage gradients and applied in stages according to the spatial distribution of the electrode array. For example, azimuth adjustment corresponds to a voltage gradient sequence for the horizontal electrode pair, while pitch adjustment corresponds to a voltage gradient sequence for the vertical electrode pair. The electrowetting effect is used to drive the liquid metal droplet to deform along the microchannel, and the changes in the main lobe direction are monitored in real time. Voltage adjustment is stopped when the deviation between the main lobe pointing and the predicted direction falls below a preset threshold, completing real-time tracking of the antenna beam.

[0106] For example, in a drone signal recognition scenario at a large-scale event, historical spatial distribution data shows that the drone signal's azimuth angle increased linearly from 100 degrees to 120 degrees over the past five seconds, while the pitch angle remained stable at 30 degrees. A Kalman filter was used to construct a state vector containing the azimuth angle, pitch angle, and angular velocity. The state transition matrix was set as a uniform motion model, and the process noise covariance was determined based on historical angular velocity fluctuations. After inputting 10 consecutive historical directional data points, the Kalman filter predicted that the azimuth angle would increase by 5 degrees to 125 degrees at the next moment, while the pitch angle remained at 30 degrees. Based on the prediction results, the azimuth deviation between the current main lobe pointing direction and the predicted direction was calculated to be 5 degrees. By mapping the angular deviation to the electrowetting control voltage parameters, a voltage correction sequence was derived for the horizontal electrode pair. Gradient voltages were applied three times, for example, 3V, 5V, and 7V, to drive the liquid metal droplet to the right. The main lobe azimuth angle change was monitored in real time. Voltage application was stopped when it reached 125 degrees. After adjustment, the antenna's main lobe beam continued to track the drone's direction of movement.

[0107] In one feasible embodiment, step S110: reconstructing target morphological parameters of the liquid metal antenna by adjusting the liquid metal channel morphology of the liquid metal antenna to collect omnidirectional electromagnetic signals of a characteristic frequency band of a drone within a target area, includes:

[0108] Based on the center frequency and bandwidth of the characteristic frequency band of the UAV, the target morphological parameters of the liquid metal channel are determined. The target morphological parameters include the target continuous length and target branching angle of the liquid metal filling path of the liquid metal antenna.

[0109] The center frequency of the drone's characteristic frequency band refers to the midpoint frequency of the frequency band used by the target drone's communication link, for example, the center frequency of the 5.8GHz band is 5.8GHz; the bandwidth refers to the effective frequency range of the frequency band, for example, ±10MHz.

[0110] First, the center frequency and bandwidth parameters of the characteristic frequency band are determined based on the target drone's communication protocol. Then, based on these center frequency and bandwidth parameters, the corresponding relationship between the center frequency and bandwidth of the drone's characteristic frequency band and the morphological parameters of the liquid metal antenna established using the electromagnetic simulation model is used to calculate the liquid metal filling path length and branching angle required to meet the frequency band coverage. For example, for the 5.8GHz±10MHz frequency band, the target continuous length is 12mm and the target branching angle is 60 degrees based on the center frequency and bandwidth parameters. This process ensures that the antenna morphological parameters are precisely matched to the target frequency band, providing the hardware foundation for subsequent signal acquisition.

[0111] An electrowetting control voltage sequence is generated according to the target continuous length and target branching angle. By gradually loading the single-stage voltage value in the electrowetting control voltage sequence, the liquid metal droplets of the liquid metal antenna are driven to deform along the filling path until the liquid metal distribution state meets the target continuous length and target branching angle.

[0112] The electrowetting control voltage sequence consists of multiple voltage gradients, each corresponding to the driving voltage at a specific electrode region. This sequence is used to control the flow path and deformation amplitude of the liquid metal droplet in stages. The liquid metal distribution state refers to the filling pattern of the liquid metal in the microchannel.

[0113] First, the target continuous length and branch angle are converted into the voltage configuration parameters of the electrode array through the correspondence between the continuous length, branch angle and the voltage configuration parameters of the electrode array, wherein the correspondence between the continuous length, branch angle and the voltage configuration parameters of the electrode array is constructed based on the historical morphological parameters and the voltage configuration parameters. Afterwards, based on the nonlinear relationship between voltage and wettability of liquid metal in the electrowetting effect, the voltage gradient sequence for staged loading is determined. For example, when the target continuous length needs to be increased by 5mm, it is decomposed into three voltage gradients, such as 3V, 5V, and 7V, which are applied to the horizontal electrode pairs in sequence to drive the droplets to gradually extend. At the same time, in response to the branch angle adjustment, differentiated voltages are applied to the bifurcated electrodes to control the droplet diversion. It is also possible to monitor the distribution state of the liquid metal and stop the voltage loading when it is detected that the error between the continuous length and the branch angle is less than the preset threshold.

[0114] Under the condition that the liquid metal distribution state meets the target continuous length and target branching angle, omnidirectional electromagnetic signals in all directions within the target area are collected.

[0115] Once the liquid metal distribution meets the target parameters, the antenna activates its omnidirectional scanning mode. A multi-channel receiving array simultaneously collects signals from all directions and digitizes them using a high-speed analog-to-digital converter. During acquisition, the antenna maintains dynamic impedance matching to minimize reflection losses and ensure high-fidelity reception across the entire frequency band. This ultimately generates an omnidirectional electromagnetic signal dataset containing time-domain waveforms, spectral energy, and phase difference information.

[0116] For example, in the drone signal recognition scenario at a large-scale event, the drone's characteristic frequency band is 5.8GHz, the center frequency is 5.8GHz, and the bandwidth is 20MHz. First, based on the correspondence between the center frequency and bandwidth of the drone's characteristic frequency band and the morphological parameters of the liquid metal antenna, the target continuous length of the liquid metal antenna is determined to be 12.9mm, and the branching angle is 30 degrees. Then, based on the nonlinear relationship between voltage and liquid metal wettability in the electrowetting effect, an electrowetting control voltage sequence is generated. The horizontal electrode pairs are sequentially loaded with 3V, 5V, and 7V voltages, and the bifurcated electrodes are applied with a 2V voltage to form a 30-degree branch. Real-time impedance detection shows that the liquid metal extends to 12.8mm and the branching angle is 29.5 degrees, which meets the error threshold. Subsequently, the antenna switches to omnidirectional mode and collects electromagnetic signals in all directions of the 5.8GHz±10MHz frequency band through a four-channel array, forming an omnidirectional electromagnetic signal data set including time domain waveforms, frequency domain spectrum lines, and spatial direction information.

[0117] Figure 2 FIG. 1 is a flow chart of a method for adjusting a liquid metal antenna according to an embodiment of the present application. Figure 1 As shown, steps S210 to S230 are included.

[0118] In one feasible embodiment, step S130: controlling the deformation of the liquid metal droplet of the liquid metal antenna by adjusting the electrowetting control voltage parameter according to the spatial distribution data, so that the main lobe direction of the liquid metal antenna points in the direction of the drone signal, and the side lobe suppression area of the liquid metal antenna covers the spatial distribution range of the interference signal direction, including:

[0119] S210: Based on the drone signal direction and the interference signal direction in the spatial distribution data, determine the angular deviation between the main lobe direction of the liquid metal antenna and the drone signal direction, and the spatial coverage deviation between the side lobe suppression area of the liquid metal antenna and the interference signal direction.

[0120] Angular deviation refers to the difference in azimuth and elevation between the current mainlobe radiation direction of the liquid metal antenna and the direction of the target drone signal. The mainlobe direction is determined by the maximum radiation direction of the antenna pattern, while the target direction is derived from the three-dimensional coordinates of the drone signal annotated in the spatial distribution data. Spatial coverage deviation refers to the spatial gap where the antenna's sidelobe suppression area fails to fully cover the directional distribution range of the interference signal. This is manifested as insufficient overlap in azimuth and elevation between the interference direction clustering area and the boundary of the sidelobe suppression zone.

[0121] First, the three-dimensional directional coordinates of the drone signal, including azimuth and elevation angles, are extracted from the spatial distribution data. The current pattern parameters of the liquid metal antenna are then used to determine the direction of maximum mainlobe radiation. The azimuth and elevation deviation components of the two in the horizontal and vertical planes are calculated to determine the angular deviation. Simultaneously, the boundary coordinates of the interference signal directional clustering area, such as the azimuth and elevation ranges, are extracted. The azimuth and elevation coverage of the sidelobe suppression zone are compared with the boundaries of the interference zone to calculate the angular span of the uncovered area, i.e., the spatial coverage deviation.

[0122] S220: Generate a control voltage correction value of the electrowetting control voltage parameter according to the angle deviation and the spatial coverage deviation, where the control voltage correction value includes a main lobe direction alignment component and a side lobe suppression enhancement component.

[0123] The mainlobe alignment component is the set of voltage parameters required to adjust the mainlobe direction of the liquid metal antenna. By changing the electrode voltage to drive the liquid metal's deformation, the mainlobe points in the direction of the target drone. The sidelobe suppression enhancement component is the set of voltage parameters that optimizes the coverage of the sidelobe suppression area. By adjusting the liquid metal's distribution, the sidelobe energy is compressed, expanding the spatial coverage of the suppression area.

[0124] Based on the angular deviation and spatial coverage deviation, the mapping relationship between the preset angular deviation, spatial coverage deviation, and electrowetting control voltage parameters is queried to obtain the corresponding electrowetting control voltage parameters, namely the voltage correction values for the corresponding electrode areas of the liquid metal antenna. A mapping relationship table of angular deviation, spatial coverage deviation, and electrowetting control voltage parameters is established through electromagnetic simulation, recording the voltage correction values of each electrode pair under different angular deviations and spatial coverage deviations. The voltage correction values include the voltage parameters required to adjust the main lobe direction of the liquid metal antenna, namely the main lobe direction alignment component; the voltage correction values also include the voltage parameters for optimizing the coverage range of the sidelobe suppression area, namely the sidelobe suppression enhancement component.

[0125] S230: Decompose the control voltage correction value into multiple voltage gradient values, and apply the voltage gradient values to the electrode units of the liquid metal antenna in sequence until the angular difference between the main lobe direction and the drone signal direction is less than a preset difference threshold, and the side lobe suppression area covers the spatial distribution range of the interference signal direction.

[0126] The voltage parameters of the main lobe alignment component and the side lobe suppression component are decomposed into multiple sub-steps according to the spatial distribution of the electrodes. For example, the horizontal electrode pair is loaded with 3V, 5V, and 7V voltages in three steps according to the azimuth deviation, and the vertical electrode pair is loaded with 2V and 4V voltages in two steps according to the pitch deviation. The voltage gradient is applied to the corresponding electrode unit in stages, and the distribution morphology of the liquid metal is detected. The liquid metal filling path length and branch angle are used to determine the main lobe pointing and side lobe suppression area coverage of the current antenna pattern. If the main lobe angle deviation is still higher than the threshold or the ratio of the side lobe suppression area coverage to the interference direction distribution range is less than the preset ratio threshold, the next gradient voltage is loaded. Iterative adjustment is performed until the main lobe direction error is less than the preset threshold, and the ratio of the side lobe suppression area coverage to the interference direction distribution range is greater than or equal to the preset ratio threshold.

[0127] For example, in a drone signal recognition scenario at a large-scale event, spatial distribution data shows that the target drone signal is located at 120 degrees in azimuth and 30 degrees in elevation, while interference signals are densely distributed in the 60-90 degree azimuth region. The liquid metal antenna's initial mainlobe orientation is 115 degrees in azimuth and 28 degrees in elevation, with sidelobe suppression covering only 70-85 degrees in azimuth. First, based on the angular deviation (5 degrees in azimuth and 2 degrees in elevation) and the spatial coverage deviation (60-70 degrees in azimuth and 85-90 degrees are not covered), a control voltage correction value is generated: the mainlobe alignment component corresponds to a total increment of 8V for the horizontal electrode pair and 4V for the vertical electrode pair, while the sidelobe enhancement component corresponds to a 6V increment for the outer horizontal electrodes. This correction value is decomposed into a three-step loading process (3V, 3V, 2V) for the horizontal electrodes, a two-step loading process (2V, 2V) for the vertical electrodes, and a three-step loading process (2V, 2V, 2V) for the outer electrodes, and then applied to the electrode units in stages. After each loading step, the liquid metal shape is detected by an impedance sensor and the directional pattern is inferred. After three iterations, the main lobe pointing is adjusted to 120.3 degrees in azimuth and 30.1 degrees in elevation, that is, the deviation is <0.5 degrees. The sidelobe suppression area is extended to 60-90 degrees in azimuth, completely covering the interference direction, and ultimately achieving high-precision beamforming and interference shielding.

[0128] Figure 3 The following is a flow chart showing a method for determining identification information of a drone signal provided by an embodiment of the present application. Figure 1 As shown, steps S310 to S340 are included.

[0129] In one feasible embodiment, step S160: using Markov Chain Monte Carlo to adaptively sample the posterior probability information based on the electromagnetic signal, obtaining the signal classification confidence through iterative convergence, and determining the identification information of the sub-electromagnetic signal according to a preset confidence threshold, includes:

[0130] S310: Determine an initial sampling step length and an initial sampling density based on the posterior probability information of the electromagnetic signal, where the initial sampling step length and the initial sampling density correspond to the dimensional distribution of the multi-dimensional signal feature.

[0131] The initial sampling step size refers to the maximum jump distance for each state transition in the Markov Chain Monte Carlo algorithm. In a multidimensional feature space, it manifests as the maximum allowable range of eigenvalues for each dimension. The initial sampling density refers to the number of candidate points sampled per unit volume of the feature space, reflecting the degree of probability density coverage of the feature dimensional distribution. The dimensional distribution of multidimensional signal features refers to the joint probability distribution of time-domain fluctuation characteristics, frequency-domain distribution characteristics, and spatial-domain directional characteristics in multidimensional space, representing the clustering areas of different signal categories in the feature space.

[0132] Based on the posterior probability information of the electromagnetic signal output by the Bayesian network, the covariance matrix of the multidimensional signal features is analyzed to determine the variance range of each feature dimension. The initial sampling step size is set to the square root of the variance of each dimension to ensure that the step size is proportional to the feature distribution range. The initial sampling density is set using an exponential decay rule based on the number of feature dimensions, with the density decreasing with increasing dimensions. For example, the initial density in a three-dimensional feature space is set to 10 sampling points per unit cube, while in a four-dimensional space, it is reduced to 5. The step size direction is adjusted through eigenvalue decomposition of the covariance matrix to align the sampling direction with the principal component, thereby improving initial sampling efficiency.

[0133] S320: within the range defined by the initial sampling step size and the initial sampling density, perform probability sampling on the dimensional distribution of the multidimensional signal feature to generate a set of candidate sampling points, where the set of candidate sampling points includes a combination of signal features corresponding to the posterior probability information of the electromagnetic signal.

[0134] The candidate sampling point set is a set of feature space points generated by the Markov Chain Monte Carlo algorithm. Each point represents a possible signal feature combination, corresponding to a probability distribution sample of a drone or interference category. A signal feature combination refers to a specific combination of time, frequency, and spatial domain features, such as a time domain variance of 0.1, a frequency band bandwidth of 15 MHz, and an azimuth fluctuation of ±2 degrees.

[0135] The Metropolis-Hastings algorithm is used to construct a Markov chain. The current posterior probability peak is used as the initial state, and candidate points are generated at the initial step size. For each candidate point, its acceptance probability is calculated: if the posterior probability is higher than the current point, it is accepted; otherwise, it is randomly accepted based on the probability ratio. In the three-dimensional feature space, a Gaussian distribution proposal function is applied to the time, frequency, and spatial dimensions to generate candidate points. For example, time domain features are sampled from the current value using a normal distribution with a step size of 0.05, and frequency domain features are sampled with a step size of 2 MHz. After iteratively generating 1000 candidate points, duplicate samples are removed to form a set of candidate sampling points.

[0136] S330: adjusting the initial sampling step size and the initial sampling density according to the distribution discreteness of the candidate sampling point set, and repeatedly performing probability sampling until the distribution discreteness of the candidate sampling point set is lower than a preset convergence threshold.

[0137] Distribution dispersion measures the degree of dispersion of a set of candidate sampling points in feature space, calculated by weighting and quantifying the variance of samples in each dimension. The preset convergence threshold is the upper limit of the dispersion at which the sampling results are stable. For example, in three-dimensional space, the dispersion threshold is set to 0.01.

[0138] The covariance matrix of the candidate sampling point set is calculated, and the weighted sum of the variances of each dimension is used to obtain the dispersion index. If the dispersion exceeds the threshold, the sampling step size is reduced to 0.8 times the original value, and the sampling density is increased by 1.5 times. For example, the initial step size of 0.1 is adjusted to 0.08, and the density is increased from 10 points / unit to 15 points / unit. An adaptive Markov chain Monte Carlo algorithm is used to dynamically update the proposal function parameters. The sampling-evaluation-adjustment cycle is repeated until the rate of change in the dispersion is less than 5% for three consecutive iterations, which is considered convergence.

[0139] S340: Calculate the signal classification confidence of each signal category based on the proportion of the number of each signal category in the candidate sampling point set. When the signal classification confidence with the largest value is greater than a preset confidence threshold, determine that the signal category corresponding to the signal classification confidence is identification information of the sub-electromagnetic signal.

[0140] Signal classification confidence refers to the normalized value of the proportion of sampling points in a certain category, reflecting the probability that the signal belongs to that category. For example, if the drone category accounts for 80%, the confidence is 0.8.

[0141] Count the proportion of each class label in the candidate sampling point set. For the drone class, calculate the ratio of the number of sampling points to the total number as the confidence score. If the highest confidence score exceeds the threshold, the signal is classified as that class. For example, if 820 of 1000 sampling points are drones and the confidence score is 0.82, which exceeds the 0.75 threshold, the signal is classified as a drone.

[0142] For example, in a drone signal identification scenario at a large-scale event, a sub-electromagnetic signal was extracted with multi-dimensional features: a time domain variance of 0.15, a frequency bandwidth of 18 MHz, and an azimuth fluctuation of ±2 degrees. Based on the posterior probability information output by a Bayesian network, the initial sampling step size was set to 0.06 in the time domain, 1.2 MHz in the frequency domain, and 0.6 degrees in the spatial domain, with a sampling density of 10 points per unit space. After generating an initial set of candidate sampling points using a Markov Chain Monte Carlo algorithm, the distribution dispersion was detected to be 0.18, exceeding the preset threshold. The step size was dynamically adjusted to 0.04 in the time domain, 0.8 MHz in the frequency domain, and 0.4 degrees in the spatial domain, and the sampling density was increased to 15 points per unit space. After three iterative optimizations, the dispersion was reduced to 0.009, and the drone category accounted for 85% of the candidate point set. Because the confidence level of 0.85 exceeded the preset threshold of 0.8, the signal was determined to be a drone signal.

[0143] Based on the same concept, the embodiment of the present application provides a drone signal recognition system in a complex electromagnetic environment. Figure 4 The drone signal recognition system in a complex electromagnetic environment provided by the embodiment of the present application is described in detail.

[0144] Figure 4 This is a structural block diagram of a drone signal recognition system in a complex electromagnetic environment shown in an embodiment of the present application.

[0145] like Figure 4 As shown, the drone signal recognition system in this complex electromagnetic environment may include:

[0146] The acquisition module 410 is configured to reconstruct the target morphological parameters of the liquid metal antenna by adjusting the morphology of the liquid metal channel of the liquid metal antenna, and to acquire omnidirectional electromagnetic signals of the characteristic frequency band of the UAV within the target area;

[0147] A generating module 420 is configured to generate spatial distribution data including the direction of the UAV signal and the direction of the interference signal based on the phase difference of the electromagnetic signal in the omnidirectional electromagnetic signal and the preset signal characteristics of the UAV and the interference signal;

[0148] Adjustment module 430 is configured to control the deformation of the liquid metal droplet of the liquid metal antenna by adjusting the electrowetting control voltage parameter based on the spatial distribution data, so that the main lobe direction of the liquid metal antenna points in the direction of the drone signal, and the sidelobe suppression area of the liquid metal antenna covers the spatial distribution range of the interference signal direction;

[0149] The acquisition module 410 is further configured to use the liquid metal antenna after lobe adjustment to collect an electromagnetic signal set in the direction of the drone signal, and extract multidimensional signal features of each sub-electromagnetic signal from the electromagnetic signal set, wherein the multidimensional signal features include time domain fluctuation features, frequency domain distribution features, and spatial domain direction features;

[0150] Input module 440 is used to input the multidimensional signal characteristics of each sub-electromagnetic signal into the constructed joint probability distribution model to obtain the posterior probability information of the electromagnetic signal. The joint probability distribution model is constructed using a Bayesian network based on historical UAV signals and historical interference signals;

[0151] The determination module 450 is used to perform adaptive sampling based on the posterior probability information of the electromagnetic signal using Markov Chain Monte Carlo, obtain the signal classification confidence through iterative convergence, and determine the identification information of the sub-electromagnetic signal according to a preset confidence threshold.

[0152] In one embodiment, the acquisition module 410 is also used to perform blind source separation processing on the electromagnetic signal set before extracting the multidimensional signal features of each sub-electromagnetic signal from the electromagnetic signal set, and separate multiple independent signal source electromagnetic signals based on the signal time-frequency sparsity assumption and independent component analysis; and extract the multidimensional signal features of each independent signal source electromagnetic signal from the multiple independent signal source electromagnetic signals.

[0153] In one embodiment, the generation module 420 is also used to perform signal demodulation and protocol analysis on each electromagnetic signal in the omnidirectional electromagnetic signal before generating spatial distribution data including the drone signal direction and the interference signal direction based on the phase difference of the electromagnetic signal in the omnidirectional electromagnetic signal and the preset signal characteristics of the drone and the interference signal, so as to obtain demodulation protocol information including demodulation information and protocol information of each electromagnetic signal; and generate spatial distribution data including the drone signal direction and the interference signal direction based on the phase difference and demodulation protocol information of each electromagnetic signal in the omnidirectional electromagnetic signal and the preset signal characteristics of the drone and the interference signal.

[0154] In one embodiment, the adjustment module 430 is also used to control the deformation of the liquid metal droplet of the liquid metal antenna by adjusting the electrowetting control voltage parameters according to the spatial distribution data so that the main lobe direction of the liquid metal antenna points to the direction of the drone signal and the side lobe direction of the liquid metal antenna points to the direction of the interference signal. According to the preset historical spatial distribution data, the moving direction of the drone signal is obtained by Kalman filtering prediction; according to the moving direction of the drone signal, the liquid metal droplet of the liquid metal antenna is controlled by adjusting the electrowetting control voltage parameters so that the main lobe direction of the liquid metal antenna points to the moving direction of the drone signal.

[0155] In one embodiment, the acquisition module 410 is specifically used to determine the target morphological parameters of the liquid metal channel based on the center frequency and bandwidth of the characteristic frequency band of the drone, the target morphological parameters including the target continuous length and target branch angle of the liquid metal filling path of the liquid metal antenna; generate an electrowetting control voltage sequence according to the target continuous length and target branch angle, and drive the liquid metal droplets of the liquid metal antenna to deform along the filling path by gradually loading the single-stage voltage value in the electrowetting control voltage sequence until the liquid metal distribution state meets the target continuous length and target branch angle; under the condition that the liquid metal distribution state meets the target continuous length and target branch angle, collect omnidirectional electromagnetic signals in all directions within the target area.

[0156] In one embodiment, the adjustment module 430 is specifically used to determine the angular deviation between the main lobe direction of the liquid metal antenna and the drone signal direction, as well as the spatial coverage deviation between the side lobe suppression area of the liquid metal antenna and the interference signal direction based on the drone signal direction and the interference signal direction in the spatial distribution data; generate a control voltage correction value of the electrowetting control voltage parameter according to the angular deviation and the spatial coverage deviation, and the control voltage correction value includes a main lobe direction alignment component and a side lobe suppression enhancement component; decompose the control voltage correction value into multiple voltage gradient values, and apply the voltage gradient values to the electrode units of the liquid metal antenna in sequence until the angular difference between the main lobe direction and the drone signal direction is less than a preset difference threshold, and the side lobe suppression area covers the spatial distribution range of the interference signal direction.

[0157] In one embodiment, the determination module 450 is specifically configured to determine an initial sampling step size and an initial sampling density based on the posterior probability information of the electromagnetic signal, where the initial sampling step size and the initial sampling density correspond to the dimensional distribution of the multidimensional signal feature; within a range defined by the initial sampling step size and the initial sampling density, perform probability sampling on the dimensional distribution of the multidimensional signal feature to generate a set of candidate sampling points, where the set of candidate sampling points includes a combination of signal features corresponding to the posterior probability information of the electromagnetic signal; adjust the initial sampling step size and the initial sampling density based on the distribution discreteness of the set of candidate sampling points, and repeatedly perform probability sampling until the distribution discreteness of the set of candidate sampling points is lower than a preset convergence threshold; calculate the signal classification confidence of each signal category based on the proportion of the number of each signal category in the set of candidate sampling points, and when the signal classification confidence with the largest value is greater than a preset confidence threshold, determine that the signal category corresponding to the signal classification confidence is the identification information of the sub-electromagnetic signal.

[0158] Figure 4 Each module in the system shown has the function of implementing Figures 1 to 3 The functions of each step in the embodiment can achieve the corresponding technical effects, which will not be described in detail here for the sake of brevity.

[0159] Figure 5A schematic diagram of the hardware structure of an electronic device provided in one embodiment of the present application is shown.

[0160] The electronic device may include a processor 510 and a memory 520 storing computer program instructions.

[0161] Specifically, the processor 510 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application.

[0162] The memory 520 may include a large-capacity memory for data or instructions. By way of example and not limitation, the memory 520 may include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive, or a combination of two or more of these. Where appropriate, the memory 520 may include removable or non-removable (or fixed) media. Where appropriate, the memory 520 may be inside or outside the integrated gateway disaster recovery device. In a specific embodiment, the memory 520 is a non-volatile solid-state memory.

[0163] The memory may include read-only memory (ROM), random access memory (RAM), magnetic disk storage media devices, optical storage media devices, flash memory devices, electrical, optical, or other physical / tangible memory storage devices. Thus, typically, the memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the method according to the first aspect of the present disclosure.

[0164] The processor 510 reads and executes computer program instructions stored in the memory 520 to implement any of the drone signal recognition methods in a complex electromagnetic environment in the above embodiments.

[0165] In one example, the electronic device may further include a communication interface 530 and a bus 540. Figure 5 As shown, the processor 510 , the memory 520 , and the communication interface 530 are connected via a bus 540 and communicate with each other.

[0166] The communication interface 530 is mainly used to implement communication between various modules, devices, units and / or equipment in the embodiments of the present application.

[0167] Bus 540 includes hardware, software, or both, and couples the components of the online data traffic metering device to each other. By way of example, and not limitation, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industrial Standard Architecture (EISA) bus, a Front Side Bus (FSB), a HyperTransport (HT) interconnect, an Industrial Standard Architecture (ISA) bus, an InfiniBand interconnect, a Low Pin Count (LPC) bus, a memory bus, a Micro Channel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local Area Network (VLB) bus, or other suitable buses, or a combination of two or more of these. Where appropriate, bus 540 may include one or more buses. Although the embodiments of the present application describe and illustrate specific buses, the present application contemplates any suitable bus or interconnect.

[0168] The electronic device can execute the drone signal recognition method in a complex electromagnetic environment in the embodiment of the present application, thereby realizing the combination of Figures 1 to 3 Described is a method for UAV signal recognition in complex electromagnetic environments.

[0169] In addition, in conjunction with the above-described methods for identifying drone signals in complex electromagnetic environments, embodiments of the present application may provide a computer-readable storage medium for implementation. The computer-readable storage medium stores computer program instructions; when executed by a processor, the computer program instructions implement any of the above-described methods for identifying drone signals in complex electromagnetic environments.

[0170] It should be understood that the present application is not limited to the specific configurations and processes described above and illustrated in the figures. For the sake of brevity, a detailed description of known methods is omitted here. In the above embodiments, several specific steps are described and illustrated as examples. However, the method process of the present application is not limited to the specific steps described and illustrated. Those skilled in the art can make various changes, modifications, and additions, or change the order of the steps after understanding the spirit of the present application.

[0171] The functional blocks shown in the block diagrams described above can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they may be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, and the like. When implemented in software, the elements of this application are programs or code segments used to perform the required tasks. Programs or code segments may be stored in a machine-readable medium or transmitted via a data signal carried in a carrier wave over a transmission medium or communication link. "Machine-readable medium" may include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROMs, flash memory, erasable ROMs (EROMs), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, and the like. Code segments may be downloaded via a computer network such as the Internet or an intranet.

[0172] It should also be noted that the exemplary embodiments mentioned in this application describe some methods or systems based on a series of steps or devices. However, this application is not limited to the order of the above steps. In other words, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0173] Aspects of the present application have been described above with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present application. It should be understood that each block in the flowcharts and / or block diagrams, as well as combinations of blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device to produce a machine such that execution of these instructions by the processor of the computer or other programmable data processing device enables the implementation of the functions / actions specified in one or more blocks in the flowcharts and / or block diagrams. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field programmable logic circuit. It should also be understood that each block in the block diagrams and / or flowcharts, as well as combinations of blocks in the block diagrams and / or flowcharts, can also be implemented by dedicated hardware that performs the specified functions or actions, or by a combination of dedicated hardware and computer instructions.

[0174] The above description is only a specific embodiment of the present application. Those skilled in the art will clearly understand that for the convenience and brevity of description, the specific working processes of the systems, modules and units described above can refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here. It should be understood that the scope of protection of the present application is not limited thereto. Any person skilled in the art can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present application, and these modifications or replacements should be included in the scope of protection of the present application.

Claims

1. A method for identifying drone signals in a complex electromagnetic environment, characterized in that: include: Reconstructing the target morphological parameters of the liquid metal antenna by adjusting the liquid metal channel morphology of the liquid metal antenna to collect omnidirectional electromagnetic signals of the characteristic frequency band of the UAV in the target area; Generate spatial distribution data including the direction of the drone signal and the direction of the interference signal based on the phase difference of the electromagnetic signal in the omnidirectional electromagnetic signal and the preset signal characteristics of the drone and the interference signal; According to the spatial distribution data, the liquid metal droplet of the liquid metal antenna is controlled to deform by adjusting the electrowetting control voltage parameter so that the main lobe direction of the liquid metal antenna points in the direction of the drone signal, and the side lobe suppression area of the liquid metal antenna covers the spatial distribution range of the interference signal direction; Using the liquid metal antenna with adjusted lobes, collecting an electromagnetic signal set in the signal direction of the drone, and extracting multidimensional signal features of each sub-electromagnetic signal from the electromagnetic signal set, the multidimensional signal features including time domain fluctuation features, frequency domain distribution features, and spatial domain direction features; Inputting the multidimensional signal features of each sub-electromagnetic signal into a constructed joint probability distribution model to obtain posterior probability information of the electromagnetic signal, wherein the joint probability distribution model is constructed based on historical UAV signals and historical interference signals through a Bayesian network; Markov chain Monte Carlo is used to adaptively sample the posterior probability information based on the electromagnetic signal, and the signal classification confidence is obtained through iterative convergence, and the identification information of the sub-electromagnetic signal is determined according to a preset confidence threshold.

2. The method according to claim 1, characterized in that Before extracting the multi-dimensional signal feature of each sub-electromagnetic signal from the electromagnetic signal set, the method further includes: Performing blind source separation on the electromagnetic signal set to separate electromagnetic signals from multiple independent signal sources based on the signal time-frequency sparsity assumption and independent component analysis; The step of extracting the multi-dimensional signal features of each sub-electromagnetic signal from the electromagnetic signal set includes: A multi-dimensional signal feature of each independent signal source electromagnetic signal is extracted from the multiple independent signal source electromagnetic signals.

3. The method according to claim 2, characterized in that Before generating spatial distribution data including the direction of the drone signal and the direction of the interference signal based on the phase difference of the electromagnetic signal in the omnidirectional electromagnetic signal and the preset signal characteristics of the drone and the interference signal, the method further includes: Performing signal demodulation and protocol analysis on each electromagnetic signal in the omnidirectional electromagnetic signal to obtain demodulation protocol information including demodulation information and protocol information of each electromagnetic signal; The generating of spatial distribution data including the direction of the UAV signal and the direction of the interference signal based on the phase difference of the electromagnetic signal in the omnidirectional electromagnetic signal and the preset signal characteristics of the UAV and the interference signal includes: According to the phase difference of each electromagnetic signal in the omnidirectional electromagnetic signal and the demodulation protocol information, as well as the preset signal characteristics of the drone and the interference signal, spatial distribution data including the drone signal direction and the interference signal direction is generated.

4. The method according to claim 3, characterized in that Before controlling the deformation of the liquid metal droplet of the liquid metal antenna by adjusting the electrowetting control voltage parameter according to the spatial distribution data so that the main lobe direction of the liquid metal antenna points in the direction of the drone signal and the side lobe direction of the liquid metal antenna points in the direction of the interference signal, the method further includes: Based on the preset historical spatial distribution data, the movement direction of the drone signal is predicted through Kalman filtering; According to the moving direction of the drone signal, the deformation of the liquid metal droplet of the liquid metal antenna is controlled by adjusting the electrowetting control voltage parameter so that the main lobe direction of the liquid metal antenna points to the moving direction of the drone signal.

5. The method according to claim 1, wherein The method of reconstructing the target morphological parameters of the liquid metal antenna by adjusting the liquid metal channel morphology of the liquid metal antenna and collecting the omnidirectional electromagnetic signal of the characteristic frequency band of the UAV in the target area includes: Determining target morphological parameters of the liquid metal channel based on the center frequency and bandwidth of the characteristic frequency band of the UAV, wherein the target morphological parameters include a target continuous length and a target branching angle of the liquid metal filling path of the liquid metal antenna; generating an electrowetting control voltage sequence according to the target continuous length and the target branching angle, and driving the liquid metal droplets of the liquid metal antenna to deform along a filling path by gradually loading single-stage voltage values in the electrowetting control voltage sequence until the liquid metal distribution state satisfies the target continuous length and the target branching angle; Under the condition that the liquid metal distribution state satisfies the target continuous length and the target branching angle, the omnidirectional electromagnetic signals in all directions within the target area are collected.

6. The method according to claim 1, characterized in that According to the spatial distribution data, the liquid metal droplet deformation of the liquid metal antenna is controlled by adjusting the electrowetting control voltage parameter so that the main lobe direction of the liquid metal antenna points to the direction of the drone signal, and the sidelobe suppression area of the liquid metal antenna covers the spatial distribution range of the interference signal direction, including: Based on the drone signal direction and the interference signal direction in the spatial distribution data, determining an angular deviation between the main lobe direction of the liquid metal antenna and the drone signal direction, and a spatial coverage deviation between the side lobe suppression area of the liquid metal antenna and the interference signal direction; generating a control voltage correction value of an electrowetting control voltage parameter according to the angle deviation and the spatial coverage deviation, wherein the control voltage correction value includes a main lobe direction alignment component and a side lobe suppression enhancement component; The control voltage correction value is decomposed into multiple voltage gradient values, and the voltage gradient values are applied to the electrode units of the liquid metal antenna in sequence until the angular difference between the main lobe direction and the drone signal direction is less than a preset difference threshold, and the sidelobe suppression area covers the spatial distribution range of the interference signal direction.

7. The method according to claim 1, characterized in that The method of using Markov Chain Monte Carlo to adaptively sample the posterior probability information of the electromagnetic signal, obtaining the signal classification confidence through iterative convergence, and determining the identification information of the sub-electromagnetic signal according to a preset confidence threshold comprises: Determining an initial sampling step length and an initial sampling density based on the posterior probability information of the electromagnetic signal, wherein the initial sampling step length and the initial sampling density correspond to the dimensional distribution of the multidimensional signal feature; performing probability sampling on the dimensional distribution of the multidimensional signal feature within a range defined by the initial sampling step size and the initial sampling density to generate a set of candidate sampling points, the set of candidate sampling points including a combination of signal features corresponding to the posterior probability information of the electromagnetic signal; Adjusting the initial sampling step size and the initial sampling density according to the distribution discreteness of the candidate sampling point set, and repeatedly performing probability sampling until the distribution discreteness of the candidate sampling point set is lower than a preset convergence threshold; The signal classification confidence of each signal category is calculated based on the proportion of the number of each signal category in the candidate sampling point set. When the signal classification confidence with the largest value is greater than the preset confidence threshold, the signal category corresponding to the signal classification confidence is determined to be the identification information of the sub-electromagnetic signal.

8. A drone signal recognition system in a complex electromagnetic environment, characterized by: The system comprises: an acquisition module, configured to reconstruct target morphological parameters of the liquid metal antenna by adjusting the morphology of the liquid metal channel of the liquid metal antenna, and to acquire omnidirectional electromagnetic signals of characteristic frequency bands of the UAV within the target area; A generating module, configured to generate spatial distribution data including the direction of the drone signal and the direction of the interference signal based on the phase difference of the electromagnetic signal in the omnidirectional electromagnetic signal and the preset signal characteristics of the drone and the interference signal; an adjustment module, configured to control the deformation of the liquid metal droplet of the liquid metal antenna by adjusting the electrowetting control voltage parameter according to the spatial distribution data, so that the main lobe direction of the liquid metal antenna points in the direction of the UAV signal, and the sidelobe suppression area of the liquid metal antenna covers the spatial distribution range of the interference signal direction; The acquisition module is further configured to use the liquid metal antenna after lobe adjustment to collect an electromagnetic signal set in the signal direction of the drone, and extract multidimensional signal features of each sub-electromagnetic signal from the electromagnetic signal set, wherein the multidimensional signal features include time domain fluctuation features, frequency domain distribution features, and spatial domain direction features; An input module, configured to input the multidimensional signal characteristics of each of the electromagnetic sub-signals into a constructed joint probability distribution model to obtain posterior probability information of the electromagnetic signal, wherein the joint probability distribution model is constructed using a Bayesian network based on historical UAV signals and historical interference signals; The determination module is used to adaptively sample the posterior probability information based on the electromagnetic signal using Markov chain Monte Carlo, obtain the signal classification confidence through iterative convergence, and determine the identification information of the sub-electromagnetic signal according to a preset confidence threshold.

9. An electronic device, characterized in that: The device includes: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, it implements the method for identifying drone signals in a complex electromagnetic environment as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer program instructions, and when the computer program instructions are executed by the processor, the method for identifying drone signals in a complex electromagnetic environment according to any one of claims 1 to 7 is implemented.

Images