Unmanned aerial vehicle signal intelligent detection method and system for low-altitude unmanned aerial vehicle detection

Abstract

The invention specifically relates to an unmanned aerial vehicle signal intelligent detection method for low-altitude unmanned aerial vehicle detection, and the method comprises the steps: designing an OFDM signal state intelligent sensing mechanism based on phase-space reconstruction, and constructing the phase-space high-dimensional features of an unmanned aerial vehicle signal through introducing a phase-space reconstruction theory; a depth perception model is designed in combination with a VIT neural network, and effective perception of an OFDM signal state is realized; designing an OFDM signal time domain parameter estimation strategy, constructing a parameter estimation target function by using generalized correlation entropy, and performing optimization solution on shape and scale parameters of the target function in combination with a whale optimization algorithm to realize robust estimation of the OFDM signal time domain parameter; according to the method, an unmanned aerial vehicle signal detection scheme based on parameter feature matching is constructed, feature vectors are established by using OFDM signal states and time domain parameters, a matching decision is made according to a target signal parameter library, and accurate detection of unmanned aerial vehicle signals is realized. According to the method provided by the invention, high-precision detection of the image transmission signal of the unmanned aerial vehicle can be realized in a complex interference environment.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) signal detection technology, and specifically to an intelligent UAV signal detection method and system for low-altitude UAV detection. Background Technology

[0002] With the rapid development of flight control and wireless communication technologies, unmanned aerial vehicles (UAVs) have made significant progress in miniaturization, low-altitude flight, and functional diversification. They are widely used in intelligence reconnaissance, material delivery, power line inspection, agricultural plant protection, and film and television production, becoming a key force driving technological innovation in related industries. However, the widespread use of UAVs has also brought many security risks. Unauthorized UAVs frequently intrude into sensitive areas such as airport airspace protection zones and restricted zones. In particular, low-altitude UAVs, with their low flight altitude and high maneuverability, can easily use terrain cover and urban building complexes for covert flight, posing a significant challenge to traditional detection methods and constituting a significant threat to public and national security. Against this backdrop, how to achieve accurate detection of low-altitude UAVs has become a core issue in building an efficient airspace security and supervision system. Currently, UAV detection methods mainly include radar detection, photoelectric detection, acoustic detection, and radio frequency signal detection. Traditional radar, photoelectric, and acoustic methods suffer from limited detection range and low detection accuracy when facing micro-UAVs, complex weather environments, and background noise interference. In contrast, radio frequency (RF) signal detection technology, by capturing and analyzing UAV communication signals, is unaffected by lighting or weather conditions, possessing excellent environmental adaptability and robustness, enabling all-weather, long-range UAV detection. Its detection results can be linked to subsequent jamming and interception systems, forming an integrated response mechanism. RF signal-based detection technology plays a crucial role in counter-UAV systems.

[0003] Despite the significant advantages of radio frequency (RF) signal-based drone detection technology, current drone signal detection still faces numerous challenges. Zhao et al. proposed using Auxiliary Classifier Wasserstein Generative Adversarial Networks (AC-WGANs) to classify drones based on signal waveforms, but their study only investigated signal recognition performance at signal-to-noise ratios (SNR) above 5 dB. ALLAHHAM MS's multi-channel signal recognition method effectively extracts frequency domain features using a Convolutional Neural Network (CNN), but this method neglects to study signal detection performance under different SNR conditions. BASAK S proposed using a Residual Network (ResNet) to extract feature information from the time-frequency amplitude and studied the drone RF signal detection performance under different SNR conditions, but this method struggles to effectively extract features under low SNR conditions, resulting in low detection accuracy. Furthermore, the above time-frequency domain methods only use amplitude information in the time-frequency domain while ignoring phase information, leading to incomplete input data for the model and affecting detection accuracy. Sun et al.'s research suggested that the bandwidth of UAV image transmission signals is 10MHz, while that of Wi-Fi is 20MHz, and bandwidth can be used to distinguish UAVs relatively well. However, with the development of UAV image transmission technology, today's UAV image transmission signals not only have a 10MHz bandwidth but also a 20MHz bandwidth. Therefore, bandwidth characteristics alone are currently insufficient to effectively distinguish UAV image transmission signals from Wi-Fi signals.

[0004] Based on the above analysis, the problems and shortcomings of the existing technology are as follows: Current drone signal detection technology has significant shortcomings. With the continuous development of drone image transmission technology, the bandwidth of its transmission signals is insufficient to effectively distinguish between drone image transmission signals and communication signals such as Wi-Fi. The high degree of overlap in bandwidth and modulation methods between the two makes traditional bandwidth-based detection methods face significant challenges in complex electromagnetic environments, making reliable detection difficult. Therefore, it is urgent to explore more distinguishing parameters beyond bandwidth to improve the differentiation between drone signals and Wi-Fi signals.

[0005] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of the present invention, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] This invention provides an intelligent detection method and system for drone signals for low-altitude drone detection, which can better distinguish drone signals from WIFI signals and improve detection accuracy.

[0007] Other features and advantages of the invention will become apparent from the following detailed description, or may be learned in part by practice of the invention.

[0008] According to a first aspect of the present invention, a method for intelligent detection of drone signals for low-altitude drone detection is provided, the method comprising: Acquire UAV image transmission signals, construct high-dimensional phase space features of UAV signals based on phase space reconstruction theory; combine high-dimensional phase space features and corresponding labels into a training set, input the training set into a VIT neural network for training, and use the trained VIT neural network to achieve effective perception of OFDM signal state. Based on the time delay difference of the sensed OFDM signal, a Gaussian probability density function of generalized correlation entropy is constructed. Based on the Gaussian probability density function of generalized correlation entropy, a generalized correlation entropy response function is obtained. Based on the generalized correlation entropy response function, time-domain parameters are obtained. The time-domain parameters include useful symbol length, CP length, and total symbol length, wherein the total symbol length is the sum of useful symbol length and CP length. Based on the OFDM signal state and time-domain parameters, a reconnaissance signal feature vector is established. The reconnaissance signal feature vector is then matched with the target signal parameter database to determine whether it is a UAV signal.

[0009] In some exemplary embodiments, the construction of high-dimensional phase space features of UAV signals based on phase space reconstruction theory specifically includes: The phase space of the UAV image transmission signal is reconstructed to obtain the reconstructed phase space matrix, and the angle between the continuous trajectories between attractors is calculated based on the reconstructed phase space matrix. The feature matrix composed of the mean and variance of the angles between continuous trajectories extracted with different embedding dimensions and different delay times is used as the high-dimensional feature of the phase space.

[0010] In some exemplary embodiments, the step of inputting the training set into the VIT neural network for training specifically includes: Labeled training samples are input into the VIT network, using the cross-entropy function. As the loss function of the network, it is specifically expressed as follows:

[0011] in, For correct labeling, For network prediction results, This represents the probability that an OFDM signal is not present. This indicates the probability of the presence of an OFDM signal. This represents the total number of samples in the training set. The network performance is evaluated based on the set loss function. An iterative optimization strategy is used to adjust the model parameters until the loss value tends to stabilize, thus completing the training process of the VIT network.

[0012] In some exemplary embodiments, the effective perception of OFDM signal state using a trained VIT neural network specifically includes: The test set samples are input into the trained VIT neural network for forward propagation to achieve effective perception of OFDM signal state.

[0013] In some exemplary embodiments, the construction of a Gaussian probability density function of generalized correlation entropy based on the time delay difference of the sensed OFDM signal, the obtaining of a generalized correlation entropy response function based on the Gaussian probability density function of generalized correlation entropy, and the acquisition of time-domain parameters based on the generalized correlation entropy response function are specifically as follows: The Gaussian probability density function of the generalized correlation entropy is:

[0014] in, For shape parameters, For scale or width parameters, For gamma function, To sense the time delay difference of OFDM signals; Define delay The generalized correlation entropy response is:

[0015] By iterating through different delay parameters in sequence Useful symbol length It can be derived from the generalized correlation entropy response function Maximum value estimation point:

[0016] Let the total length of each OFDM symbol be , For each possible CP length, Extract the cyclic prefix segment and the terminal repeat segment for each symbol.

[0017]

[0018] Calculate the generalized correlation entropy responses of both and take the average:

[0019] The estimated CP length is:

[0020] The total length of the symbol is .

[0021] In some exemplary embodiments, the method further includes: optimizing the shape and scale parameters in the Gaussian probability density function of the generalized correlation entropy using the whale optimization algorithm to obtain the optimal parameter pair.

[0022] In some exemplary embodiments, the step of matching the reconnaissance signal feature vector with the target signal parameter library to determine whether it is a UAV signal specifically involves: Log-normalize the feature vector of the reconnaissance signal; The distance between the reconnaissance signal feature vector and the target signal parameter library is calculated based on the corresponding parameter weights. The similarity is calculated based on the distance between the reconnaissance signal feature vector and the target signal parameter library. The similarity is compared with a set threshold. If the similarity is greater than the set threshold, it is a UAV signal.

[0023] According to a second aspect of the present invention, an intelligent detection system for drone signals for low-altitude drone detection is provided, comprising: State perception module: acquires UAV image transmission signals, constructs high-dimensional phase space features of UAV signals based on phase space reconstruction theory; combines high-dimensional phase space features and corresponding labels into a training set, inputs the training set into a VIT neural network for training, and uses the trained VIT neural network to achieve effective perception of OFDM signal state; Parameter estimation module: Constructs a Gaussian probability density function of generalized correlation entropy based on the time delay difference of the sensed OFDM signal, obtains a generalized correlation entropy response function based on the Gaussian probability density function of generalized correlation entropy, and obtains time-domain parameters based on the generalized correlation entropy response function. The time-domain parameters include useful symbol length, CP length and total symbol length, wherein the total symbol length is the sum of useful symbol length and CP length. Matching Decision Module: Based on OFDM signal state and time domain parameters, a reconnaissance signal feature vector is established. The reconnaissance signal feature vector is matched with the target signal parameter library to determine whether it is a UAV signal.

[0024] According to a third aspect of the present invention, a storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the intelligent detection method for low-altitude unmanned aerial vehicle (UAV) signals described in the first aspect.

[0025] According to a fourth aspect of the present invention, an electronic device is provided, comprising: Processor; and Memory for storing the executable instructions of the processor; The processor is configured to implement the intelligent detection method for low-altitude unmanned aerial vehicle (UAV) signals described in the first aspect above by executing the executable instructions.

[0026] The intelligent detection method and system for low-altitude UAV signals provided in this invention designs an OFDM signal state perception based on phase space reconstruction and VIT, overcoming the performance instability of traditional methods based on high-order cumulants in low signal-to-noise ratio environments. It eliminates the need for manually designing thresholds or setting statistical decision rules, significantly improving the perception accuracy of OFDM signals. The adaptive parameter calculation method, which integrates generalized correlation entropy and whale optimization algorithm, overcomes the performance sensitivity issues of traditional methods requiring manual setting. It achieves bivariate dynamic optimization of the shape and scale parameters of the generalized correlation entropy, improving the performance of parameter estimation under low signal-to-noise ratio conditions. This intelligent detection method can be used for efficient detection in any scenario involving UAV signal detection in complex electromagnetic environments, especially in low signal-to-noise ratio and complex environments. The intelligent UAV signal detection method based on multi-feature parameter synergy proposed in this application has extremely high accuracy and strong versatility.

[0027] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit the invention. Attached Figure Description

[0028] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention. It is obvious that the drawings described below are merely some embodiments of the invention, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.

[0029] Figure 1 This is a flowchart of an intelligent detection method for drone signals for low-altitude drone detection proposed in this application, provided in one embodiment of the present application. Figure 2 This is a block diagram of an intelligent detection system for low-altitude drone signals proposed in this application, provided in one embodiment of the present application. In the diagram: 21. State perception module; 22. Parameter estimation module; 23. Matching decision module.

[0030] Figure 3 This is a schematic diagram of the OFDM signal sensing accuracy and false alarm rate provided in one embodiment of this application; Figure 4This is a schematic diagram of the accuracy of OFDM signal time-domain parameter estimation proposed in this application, provided in one embodiment of the present application. Figure 5 This is a schematic diagram illustrating the drone signal detection accuracy of the intelligent drone signal detection method for low-altitude drone detection proposed in this application, provided in one embodiment of the present application. Detailed Implementation

[0031] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the invention will be more comprehensive and complete, and will fully convey the concept of the exemplary embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0032] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0033] The difficulty in overcoming the problems and shortcomings of existing technologies lies in the fact that the structural characteristics of signals are easily destroyed under strong noise interference and non-ideal channel conditions. Achieving stateful perception of OFDM signals and further accurately estimating their key parameters is an extremely challenging process. Therefore, this paper introduces phase space reconstruction theory to construct high-dimensional phase space features of UAV signals; combines this with a VIT neural network to design a deep perception model; utilizes generalized correlation entropy to construct a parameter estimation objective function; and combines the whale optimization algorithm to optimize the shape and scale parameters of the objective function. This leads to the construction of a UAV signal detection scheme based on parameter feature matching, achieving efficient, robust, and effective UAV signal detection.

[0034] The significance of solving existing technical problems and defects is that drone signal detection is an important prerequisite for realizing an anti-drone system. It can not only ensure the safety and airspace order of important areas, but also provide accurate technical support for subsequent countermeasures such as interference and interception.

[0035] To address the shortcomings and deficiencies of existing technologies, this example embodiment provides an intelligent UAV signal detection method for low-altitude UAV detection. It mainly involves an intelligent sensing mechanism for OFDM signal state based on phase space reconstruction, designing an OFDM signal time-domain parameter estimation strategy, and constructing a UAV signal detection scheme based on parameter feature matching to achieve UAV signal detection. (Reference) Figure 1 As shown, the specific steps may include: Step S11: Design an intelligent sensing mechanism for OFDM signal state based on phase space reconstruction. By introducing phase space reconstruction theory, construct high-dimensional phase space features of UAV signals; and combine Vision Transformer (VIT) neural network to design a deep sensing model to achieve effective sensing of OFDM signal state. Step S12: Design a time-domain parameter estimation strategy for OFDM signals. Construct a parameter estimation objective function by utilizing generalized correlation entropy, and combine it with the whale optimization algorithm to optimize the shape and scale parameters of the objective function, thereby achieving robust estimation of OFDM signal time-domain parameters. Step S13: Construct a UAV signal detection scheme based on parameter feature matching. By using OFDM signal state and time domain parameters to establish feature vectors, and making matching decisions based on the target signal parameter library, accurate detection of UAV signals can be achieved.

[0036] The steps in this exemplary embodiment will now be described in more detail with reference to the accompanying drawings and embodiments.

[0037] In step S11, the construction of high-dimensional phase space features of the UAV signal, combined with the design of a deep perception model using a VIT neural network, to achieve effective perception of the OFDM signal state includes: First, let the image transmission signal of the UAV be... , Let be the signal length. Then the reconstructed phase space can be expressed as:

[0038] Where, it is assumed that the embedding dimension is The time delay parameter is X is the reconstructed phase space matrix, which is formed by... It consists of 3 reconstructed vectors, each vector being equivalent to 1 A point in a dimensional phase space.

[0039]

[0040] Where, vector An attractor for reconstructing phase space.

[0041] Then, the geometric features of the phase space are reconstructed by extracting the characterization signals to obtain the angle between the continuous trajectories of the attractors:

[0042] in,

[0043] Subsequently, the feature matrix composed of the mean and variance of the angles between continuous trajectories extracted for different embedding dimensions and different latency times was used as the training sample set. The dataset is in the form of... The data pairs consist of, among which Represents the geometric characteristic matrix. The correct labels are assigned: label "0" represents other signal types, and label "1" represents UAV image transmission signals. The dataset is divided into training and test sets in an 8:2 ratio. A pre-built VIT network is used to train the training samples to obtain the training model, including: inputting labeled training samples into the VIT network and applying the cross-entropy function. As the loss function of the network, it is specifically expressed as follows:

[0044] in, For correct labeling, For network prediction results, This represents the probability that an OFDM signal is not present. This indicates the probability of the presence of an OFDM signal. This represents the total number of samples in the training set. Based on the defined loss function, the network performance is evaluated. An iterative optimization strategy is used to adjust the model parameters until the loss value stabilizes, completing the training process of the VIT network. The trained model then performs forward propagation on the test set samples to achieve effective perception of the OFDM signal state.

[0045] In step S12, the step of constructing a parameter estimation objective function based on generalized correlation entropy and optimizing the shape and scale parameters of the objective function using the whale optimization algorithm includes: First, the Gaussian probability density function of the generalized correlation entropy is constructed as follows:

[0046] in, For shape parameters, For scale or width parameters, This is a gamma function.

[0047] Let the sensed OFDM signal be . The time delay difference is defined as:

[0048] in, For delay amount, , Let be the signal length, then the Gaussian probability density function of the generalized correlation entropy is:

[0049] Next, define the delay. The generalized correlation entropy response is:

[0050] By iterating through different delay parameters in sequence Useful symbol length It can be derived from the generalized correlation entropy response function Maximum value estimation point:

[0051] Estimate Then, let the total length of each OFDM symbol be... , For each possible CP length, Extract the cyclic prefix segment and the terminal repeat segment for each symbol. The cyclic prefix segment and the terminal repeat segment are represented as follows:

[0052]

[0053] Calculate the generalized correlation entropy responses of both and take the average:

[0054] The final estimated CP length is:

[0055] The total length of the symbol is .

[0056] Subsequently, the Whale Optimization Algorithm (WOA) is introduced to optimize the shape and scale parameters in the Gaussian probability density function of the generalized correlation entropy to obtain the optimal parameter pair. WOA achieves efficient search of the parameter space by simulating the hunting behavior of humpback whales, including spiral bubble attacks and prey encirclement mechanisms. The initialization of individual whale positions is performed; the position of each individual in the current... , Fitness value under parameters; iteratively perform position updates and boundary constraints; the final output maximizes the peak value of the generalized correlation entropy response. , The parameter pairs enable the construction of the optimal Gaussian probability density function, thereby enhancing the ability to estimate the time-domain parameter characteristics of OFDM signals. WOA simulates humpback whale hunting behavior, setting up search agent individuals as... The fitness function is defined as follows:

[0057] This means minimizing the maximum response value and improving the prominence of latency peaks. WOA updates its position in each iteration; the prey encirclement and bubble attack phases are defined as follows:

[0058]

[0059] in, and As a random factor, and For the helical parameters.

[0060] In step S13, the step of establishing a feature vector using OFDM signal state and time-domain parameters, and making matching decisions based on the target signal parameter library to achieve accurate detection of UAV signals includes: First, logarithmic normalization is performed on the feature vectors. To allow parameters of different orders of magnitude to be compared on the same scale, logarithmic normalization ensures they take values ​​within the same range, helping to eliminate the influence of dimensions and numerical ranges during the matching process. Assume the normalized dataset is... The normalization formula is

[0061] in, A constant that satisfies λ ≥ 1, ensuring its value is positive when taking the logarithm. (Constant in the logarithm) This is to ensure that the logarithm of small data points is positive. Because the natural logarithm function is an increasing function, it guarantees that the logarithm will be positive. Large small data points become positive after taking their logarithm.

[0062] Subsequently, a metric is calculated based on the distance between the estimated parameters of the reconnaissance signal and the target signal parameter database. Weights are then adaptively generated based on the dispersion of the samples. It is assumed that the target signal parameter database contains a total of [number missing]. There are signal types, One feature parameter, Let be the eigenparameter matrix normalized from the target signal parameter library, then the ... Entropy calculation of each feature parameter:

[0063] in, .

[0064] No. The weight coefficients of each feature parameter are:

[0065] The distance between the estimated parameters of the reconnaissance signal and the target signal parameter library is:

[0066] .

[0067] in, The distance between the estimated parameters of the reconnaissance signal and the target signal parameter library. For the reconnaissance signal feature vector, The first element in the reconnaissance signal feature vector One feature parameter; For the target signal database feature vector, The first element in the feature vector of the target signal One feature parameter, For the first The maximum value of each feature parameter; For the first The weight of each feature parameter.

[0068] After that, Represents the parameters for estimating reconnaissance signals and Similarity between target parameter libraries:

[0069] The larger the value, the greater the similarity; conversely, the smaller the value, the smaller the similarity. If the similarity exceeds the set threshold, it is identified as a drone signal.

[0070] It should be noted that, as another aspect, this application also provides a drone system for detecting low-altitude drones, comprising: The state perception module constructs high-dimensional phase space features of UAV signals by introducing phase space reconstruction theory, and designs a deep perception model by combining VIT neural network to achieve effective perception of OFDM signal state.

[0071] The parameter estimation module constructs a parameter estimation objective function using generalized correlation entropy and combines it with the whale optimization algorithm to optimize the shape and scale parameters of the objective function.

[0072] The matching decision module constructs a UAV signal detection scheme based on parameter feature matching. It establishes feature vectors by utilizing OFDM signal state and time-domain parameters, and makes matching decisions based on the target signal parameter library.

[0073] The technical effects of this application will be described in detail below with reference to simulation.

[0074] The experiment was conducted using a computer with an AMD Ryzen 9 6900HX CPU, an NVIDIA GeForce RTX 3060 GPU, and 32GB of RAM. The computing environment was Windows 10, and the software environment was Python 3.7. The VIT network was constructed and trained using the Tensorflow 2.5.0 deep learning framework. During training, the learning rate was set to 0.001, the batch size to 32, the number of epochs to 50, the stochastic gradient descent optimizer was used, and the parameter matching detection threshold was set to 0.9.

[0075] In the simulation experiment, various signals were selected as the main research objects, including Bluetooth GFSK signals, WIFI QPSK, 16QAM, and 64QAM modulated signals, UAV control signals (such as frequency hopping spread spectrum signals), and UAV image transmission signals. The signal-to-noise ratio (SNR) of the training set ranged from -5dB to 5dB, and the SNR of the test set ranged from -10dB to 10dB, with values ​​taken in 1dB increments. The channel model was a probabilistic line-of-sight model, and the noise type was Gaussian white noise. The sample set was divided into a training set and a test set. The total number of UAV image transmission signal samples in the training set was 11 × 200 = 2200, and GFSK, 16QAM, 64QAM, QPSK, and UAV control signals were generated 5 × 11 × 40 = 2200, for a total of 4400 training samples. The test set contained 880 samples.

[0076] Figure 3 This diagram illustrates the sensing accuracy and false alarm rate of OFDM signals under different signal-to-noise ratios (SNRs). A higher SNR results in higher sensing accuracy and a lower false alarm rate. When the SNR is greater than -7dB, the accuracy is above 90% and the false alarm rate is below 12%, demonstrating good performance even at low SNRs.

[0077] Figure 4 The estimation accuracy of OFDM signal time-domain parameters under different signal-to-noise ratios is shown. When the useful symbol length of OFDM is -6dB or higher, the estimation accuracy of the useful symbol length based on generalized correlation entropy and whale optimization algorithm reaches 100%. For the estimation of OFDM cyclic prefix length, the estimation accuracy reaches 100% when the signal-to-noise ratio is -5dB or higher.

[0078] Figure 5The diagram illustrates the detection rates of image transmission signals from three drones. At an SNR of -3dB, all three drones achieved a detection rate of 100%. At an SNR of -11dB, the DJI 4A drone's detection rate was 0%, while at an SNR of -4dB, it achieved a 100% detection rate. Experimental results demonstrate that the algorithm can effectively detect drone image transmission signals even at low signal-to-noise ratios.

[0079] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented, in whole or in part, as a computer program product, the computer program product includes one or more computer instructions. When the computer program instructions are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., a solid-state drive (SSD)).

[0080] Other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention herein. This application is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The specification and embodiments are to be considered exemplary only, and the true scope and spirit of the invention are indicated by the claims.

[0081] It should be understood that the present invention is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is defined only by the appended claims.

Claims

1. A method for intelligent detection of drone signals for low-altitude drone detection, characterized in that, The method includes: Acquire UAV image transmission signals, construct high-dimensional phase space features of UAV signals based on phase space reconstruction theory; combine high-dimensional phase space features and corresponding labels into a training set, input the training set into a VIT neural network for training, and use the trained VIT neural network to achieve effective perception of OFDM signal state. Based on the time delay difference of the sensed OFDM signal, a Gaussian probability density function of generalized correlation entropy is constructed. Based on the Gaussian probability density function of generalized correlation entropy, a generalized correlation entropy response function is obtained. Based on the generalized correlation entropy response function, time-domain parameters are obtained. The time-domain parameters include useful symbol length, CP length, and total symbol length, wherein the total symbol length is the sum of useful symbol length and CP length. Based on the OFDM signal state and time-domain parameters, a reconnaissance signal feature vector is established. The reconnaissance signal feature vector is then matched with the target signal parameter database to determine whether it is a UAV signal.

2. The method according to claim 1, characterized in that, The high-dimensional phase space features of the UAV signal constructed based on phase space reconstruction theory are as follows: The phase space of the UAV image transmission signal is reconstructed to obtain the reconstructed phase space matrix, and the angle between the continuous trajectories between attractors is calculated based on the reconstructed phase space matrix. The feature matrix composed of the mean and variance of the angles between continuous trajectories extracted with different embedding dimensions and different delay times is used as the high-dimensional feature of the phase space.

3. The method according to claim 2, characterized in that, The process of inputting the training set into the VIT neural network for training specifically involves: Labeled training samples are input into the VIT network, using the cross-entropy function. As the loss function of the network, it is specifically expressed as follows: in, For correct labeling, For network prediction results, This represents the probability that an OFDM signal is not present. This indicates the probability of the presence of an OFDM signal. This represents the total number of samples in the training set. The network performance is evaluated based on the set loss function. An iterative optimization strategy is used to adjust the model parameters until the loss value tends to stabilize, thus completing the training process of the VIT network.

4. The method according to claim 3, characterized in that, The effective perception of OFDM signal state using a trained VIT neural network is specifically as follows: The test set samples are input into the trained VIT neural network for forward propagation to achieve effective perception of OFDM signal state.

5. The method according to claim 1, characterized in that, The generalized correlation entropy Gaussian probability density function is constructed based on the time delay difference of the sensed OFDM signal. The generalized correlation entropy response function is obtained based on this Gaussian probability density function. The time-domain parameters are then obtained based on the generalized correlation entropy response function. Specifically: The Gaussian probability density function of the generalized correlation entropy is: in, For shape parameters, For scale or width parameters, For gamma function, To sense the time delay difference of OFDM signals; Define delay The generalized correlation entropy response is: By iterating through different delay parameters in sequence Useful symbol length It can be derived from the generalized correlation entropy response function Maximum value estimation point: Let the total length of each OFDM symbol be , For each possible CP length, Extract the cyclic prefix segment and the terminal repeat segment for each symbol. Calculate the generalized correlation entropy responses of both and take the average: The estimated CP length is: The total length of the symbol is .

6. The method according to claim 5, characterized in that, The method further includes: optimizing the shape and scale parameters in the Gaussian probability density function of the generalized correlation entropy using the whale optimization algorithm to obtain the optimal parameter pair.

7. The method according to claim 1, characterized in that, The step of matching the reconnaissance signal feature vector with the target signal parameter database to determine whether it is a UAV signal specifically involves: Log-normalize the feature vector of the reconnaissance signal; The distance between the reconnaissance signal feature vector and the target signal parameter library is calculated based on the corresponding parameter weights. The similarity is calculated based on the distance between the reconnaissance signal feature vector and the target signal parameter library. The similarity is compared with a set threshold. If the similarity is greater than the set threshold, it is a UAV signal.

8. A system for implementing the intelligent detection method for drone signals for low-altitude drone detection as described in any one of claims 1-7, characterized in that, include State perception module: acquires UAV image transmission signals and constructs high-dimensional phase space features of UAV signals based on phase space reconstruction theory; The high-dimensional features of the phase space and their corresponding labels are combined to form a training set. The training set is then input into the VIT neural network for training. The trained VIT neural network is then used to achieve effective perception of the state of OFDM signals. Parameter estimation module: Constructs a Gaussian probability density function of generalized correlation entropy based on the time delay difference of the sensed OFDM signal, obtains a generalized correlation entropy response function based on the Gaussian probability density function of generalized correlation entropy, and obtains time-domain parameters based on the generalized correlation entropy response function. The time-domain parameters include useful symbol length, CP length and total symbol length, wherein the total symbol length is the sum of useful symbol length and CP length. Matching Decision Module: Based on OFDM signal state and time domain parameters, a reconnaissance signal feature vector is established. The reconnaissance signal feature vector is matched with the target signal parameter library to determine whether it is a UAV signal.

9. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the intelligent detection method for drone signals for low-altitude drone detection as described in any one of claims 1 to 7.

10. An electronic device, characterized in that, include: processor; as well as Memory for storing the executable instructions of the processor; The processor is configured to execute the intelligent detection method for drone signals for low-altitude drone detection according to any one of claims 1 to 7 by executing the executable instructions.

Images