Radar radiation source identification method

Abstract

The invention relates to the technical field of communication, and discloses a radar radiation source identification method, which comprises the following steps: acquiring an original radiation source signal, and carrying out signal processing operation on the original radiation source to obtain a clear radiation source signal; performing vector quantization operation according to the clear radiation source signal to obtain a quantized radiation source signal; inputting the quantized radiation source signal into a pre-trained signal screening model to obtain a to-be-identified radiation source signal; according to the to-be-identified radiation source signal, performing a nearest neighbor node extraction operation to obtain a nearest neighbor node set; according to the nearest neighbor node set and a preset target node, performing calculation to obtain a signal distance; and according to the signal distance, carrying out maximum value extraction operation to obtain a maximum signal distance, and determining the radiation source corresponding to the maximum signal distance as a final radiation source. The method can improve the recognition precision of radar radiation source signals in a complex electronic countermeasure environment.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and in particular to a radar radiation source identification method. Background Art

[0002] In the electronic countermeasure environment, the identification of radar emitters is a very important task, which involves analyzing the characteristics of the emitters to obtain key information about the working parameters, mechanism and intention of the emitters. The individual identification of radar emitters can provide important support for electronic reconnaissance, strategic intelligence acquisition, and interference and attack against emitters. Therefore, how to accurately identify radar emitter signals has become a key issue that needs to be solved in electronic countermeasures.

[0003] In the prior art, signal recognition methods based on instantaneous autocorrelation characteristics, time-frequency analysis and other parameters (such as fractal dimension, scale entropy, etc.) have been proposed. These methods have improved the recognition performance to a certain extent, but the instantaneous autocorrelation characteristics are greatly affected by factors such as noise, interference and multipath fading, resulting in a decrease in its resolution and unsatisfactory recognition effect. Although the time-frequency analysis method can improve the resolution of the signal, its recognition effect cannot meet the requirements of high-precision recognition in modern electronic countermeasure environments.

[0004] In summary, the signal recognition method in the prior art has low recognition accuracy for radar radiation source signals in a complex electronic countermeasure environment and is difficult to meet the demand for high-precision recognition. Summary of the invention

[0005] The present invention provides a radar radiation source identification method to improve the identification accuracy of radar radiation source signals in a complex electronic countermeasure environment.

[0006] In a first aspect, in order to solve the above technical problems, the present invention provides a radar radiation source identification method, comprising:

[0007] Acquire an original radiation source signal, and perform a signal processing operation on the original radiation source to obtain a clear radiation source signal;

[0008] Performing a vector quantization operation according to the clear radiation source signal to obtain a quantized radiation source signal;

[0009] Inputting the quantized radiation source signal into a pre-trained signal screening model to obtain a radiation source signal to be identified;

[0010] According to the radiation source signal to be identified, a nearest neighbor node extraction operation is performed to obtain a nearest neighbor node set;

[0011] Calculate the signal distance based on the nearest neighbor node set and the preset target node;

[0012] According to the signal distance, a maximum value extraction operation is performed to obtain a maximum signal distance, and a radiation source corresponding to the maximum signal distance is determined as a final radiation source.

[0013] Preferably, the acquiring of the original radiation source signal and performing signal processing operations on the original radiation source to obtain a clear radiation source signal includes:

[0014] Performing a short-time Fourier transform on the original radiation source signal to obtain a frequency domain radiation source signal;

[0015] Based on Gaussian window function transformation, convolution processing is performed on the frequency domain radiation source signal to obtain a main radiation source signal;

[0016] Based on a preset effective signal threshold, a calculation is performed according to the main radiation source signal to obtain an energy discrimination value;

[0017] The energy discrimination value is compared with a preset energy discrimination threshold, and when the energy discrimination value is greater than the preset energy discrimination threshold, the main radiation source signal corresponding to the energy discrimination value is determined as a valid radiation source signal;

[0018] Performing differential processing on the effective radiation source signal to obtain a pulse difference;

[0019] When the pulse difference is greater than zero, the effective radiation source signal corresponding to the pulse difference is determined as a clear radiation source signal.

[0020] Preferably, the calculation formula of the energy discrimination value is:

[0021] C=M·std(x)+th

[0022] Where M is the mean of the main radiation source signal, std(x) is the standard deviation of the main radiation source signal, th is the effective signal threshold, and C is the energy discrimination value.

[0023] Preferably, performing a vector quantization operation according to the clear radiation source signal to obtain a quantized radiation source signal includes:

[0024] Normalizing the clear radiation source signal to obtain a normalized radiation source signal;

[0025] Performing a feature vector construction operation on the normalized radiation source signal to obtain a signal feature vector;

[0026] Based on a preset quantization weight matrix, a similarity quantization operation is performed on the signal feature vector to obtain vector similarity;

[0027] According to the preset quantization weight matrix and the vector similarity, an operation of updating the weight matrix is ​​performed to obtain an updated weight matrix;

[0028] A quantization mapping operation is performed according to the updated weight matrix and the signal feature vector to obtain a quantized radiation source signal.

[0029] Preferably, the training process of the signal screening model includes:

[0030] Obtain the historical true distance, historical weight matrix and historical bias matrix of the quantized radiation source signal;

[0031] An initial signal screening model is constructed according to the historical weight matrix and the historical bias matrix, and the initial signal screening model is trained based on the historical real distance and the quantized radiation source signal as input data to obtain a trained signal screening model.

[0032] Preferably, the initial signal screening model is constructed according to the historical weight matrix and the historical bias matrix, and the initial signal screening model is trained based on the historical real distance and the quantized radiation source signal as input data to obtain the trained signal screening model, including:

[0033] Initializing the historical weight matrix and the historical bias matrix of the initial signal screening model;

[0034] Based on the gradient descent optimization method, matrix adjustment is performed to obtain an adjustment weight matrix and an adjustment bias matrix;

[0035] Perform matrix updating and training according to the adjustment weight matrix and the adjustment bias matrix;

[0036] When the number of training times is greater than or equal to the preset maximum number of training times, the training is determined to be completed, and a trained signal screening model is obtained.

[0037] Preferably, performing a nearest neighbor node extraction operation according to the radiation source signal to be identified to obtain a nearest neighbor node set includes:

[0038] Calculating based on the radiation source signal to be identified and a preset feature space reference node set to obtain a node distance set;

[0039] A node distance set that is less than a preset node distance threshold value is extracted from the node distance set, and a node set corresponding to the node distance set is determined as a nearest neighbor node set.

[0040] Preferably, the calculating according to the nearest neighbor node set and the preset target node to obtain the signal distance includes:

[0041] The signal distance is calculated by the following formula:

[0042]

[0043] Where D ik is the signal distance; N is the total number of nearest neighbor nodes; d i is the i-th nearest neighbor node; R i is the preset target node; ||d(t) i -R i ||2 is the L2 norm between the i-th nearest neighbor node and the preset target node.

[0044] In a second aspect, the present invention provides a radar emitter identification method and device, comprising:

[0045] A signal processing module, used to obtain an original radiation source signal and perform a signal processing operation on the original radiation source to obtain a clear radiation source signal;

[0046] A signal quantization module, used for performing a vector quantization operation according to the clear radiation source signal to obtain a quantized radiation source signal;

[0047] A training model module, used for inputting the quantized radiation source signal into a pre-trained signal screening model to obtain a radiation source signal to be identified;

[0048] A node distance module is used to perform a nearest neighbor node extraction operation according to the radiation source signal to be identified to obtain a nearest neighbor node set;

[0049] A signal distance module, used to calculate according to the nearest neighbor node set and the preset target node to obtain the signal distance;

[0050] The radiation source determination module is used to perform a maximum value extraction operation according to the signal distance to obtain a maximum signal distance, and determine the radiation source corresponding to the maximum signal distance as a final radiation source.

[0051] In a third aspect, the present invention further provides an electronic device, comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor implements any one of the above-mentioned radar radiation source identification methods when executing the computer program.

[0052] In a fourth aspect, the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute any one of the radar radiation source identification methods described above.

[0053] The invention discloses a radar radiation source identification method. Compared with the prior art, the invention can effectively extract and process clear radiation source signals by performing signal processing and vector quantization on the original radiation source signals, thereby reducing the interference of environmental noise on the signals. This signal processing flow combines short-time Fourier transform, Gaussian window convolution processing and energy discrimination based on effective signal threshold, which not only improves the clarity of the signal, but also can more accurately extract the effective radiation source signal. Through differential processing and pulse difference judgment, the effective radiation source signal is further screened as a clear signal, thereby improving the accuracy of radiation source identification. The method effectively eliminates redundant information in the signal through signal quantization and screening model training, thereby ensuring the accuracy of subsequent identification operations. Secondly, the invention realizes more accurate radiation source identification by introducing a distance-based nearest neighbor node extraction and signal distance calculation method. Node distance calculation and maximum signal distance extraction are performed in the feature space, which can not only locate the target node with the largest difference from the signal to be identified, but also effectively distinguish the differences between different radiation sources. By setting an appropriate neighbor node distance threshold, the node extraction process is further optimized to ensure the robustness of the recognition algorithm in a complex environment. In addition, the signal screening model training based on historical true distance, weight matrix and bias matrix enables this method to gradually adapt to the characteristics of new radiation sources, has strong learning ability, can continuously optimize the recognition effect over time, and improve the overall performance of the system.

[0054] In summary, this method can improve the recognition accuracy of radar emitter signals in complex electronic countermeasure environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 It is a schematic flow chart of a radar radiation source identification method provided by a first embodiment of the present invention;

[0056] Figure 2 It is a schematic diagram of the structure of a radar radiation source identification device provided in the second embodiment of the present invention. DETAILED DESCRIPTION

[0057] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0058] Reference Figure 1 The first embodiment of the present invention provides a radar radiation source identification method, comprising the following steps:

[0059] S11, acquiring an original radiation source signal, and performing a signal processing operation on the original radiation source to obtain a clear radiation source signal;

[0060] S12, performing a vector quantization operation according to the clear radiation source signal to obtain a quantized radiation source signal;

[0061] S13, inputting the quantized radiation source signal into a pre-trained signal screening model to obtain a radiation source signal to be identified;

[0062] S14, performing a nearest neighbor node extraction operation according to the radiation source signal to be identified to obtain a nearest neighbor node set;

[0063] S15, performing calculations based on the nearest neighbor node set and the preset target node to obtain a signal distance;

[0064] S16, performing a maximum value extraction operation according to the signal distance to obtain a maximum signal distance, and determining a radiation source corresponding to the maximum signal distance as a final radiation source.

[0065] In step S11, it is necessary to obtain an original radiation source signal and perform a signal processing operation on the original radiation source to obtain a clear radiation source signal, including:

[0066] In a specific embodiment, the original radiation source signal is captured by a radar receiving antenna, and the signal contains a variety of interference factors, including environmental noise, interference signals from other devices, etc. The original signal is a continuous analog signal and needs to be digitized for further analysis. The signal is sampled and quantized using an analog-to-digital converter (ADC) to obtain a discrete time signal, i.e., the original radiation source signal.

[0067] In a specific embodiment, the original radiation source signal needs to be transformed from the time domain to the time-frequency domain in order to extract the frequency domain characteristics and energy distribution of the signal. This step requires the use of short-time Fourier transform. Short-time Fourier transform (STFT) is a time-frequency analysis tool that divides the signal into short time periods and performs Fourier transform on each period to provide joint distribution information of the signal in the time and frequency dimensions. It can analyze non-stationary signals, extract local frequency features, determine instantaneous frequencies, and generate intuitive time-frequency diagrams for signal visualization. At the same time, STFT can suppress noise and enhance signals by adjusting the window function. It is suitable for segmented processing and feature extraction of complex signals and is a basic tool for dynamic characteristic analysis, target detection, and advanced signal processing.

[0068] Specifically, the original radiation source signal is segmented, and a window function is applied to each segment for local analysis using the short-time Fourier transform formula:

[0069]

[0070] Where x(τ) is the original radiation source signal; g(τ-t) is the window function, such as a Gaussian window; ω is the angular frequency; t is the time factor; S(ω,t) is the frequency domain radiation source signal;

[0071] It should be noted that the window function uses a Gaussian window function, and its expression is:

[0072]

[0073] Where g(t) is the Gaussian window function; σ controls the width of the window function, affecting the time and frequency resolution; t is the time factor;

[0074] Through short-time Fourier transform, a complex time-frequency distribution is obtained, which can be expressed as:

[0075] S(ω,t)=|S(ω,t)|e jφ(ω,t)

[0076] Among them, S(ω,t) is the frequency domain radiator signal; the amplitude spectrum |S(ω,t)| is the modulus of the complex number S(ω,t), which reflects the energy intensity of the signal at a given time and frequency. For example, in radar signals, the amplitude spectrum can show the energy distribution of the radiator signal and help identify the main frequency components. The phase φ(ω,t) is the angle of the complex number S(ω,t), which is used to describe the phase change of the frequency component in the time-frequency structure of the signal: the phase φ(ω,t) contains the instantaneous frequency and phase change information of the signal, which can reflect the dynamic characteristics of the signal. Phase is very important in many scenarios, for example, analyzing the propagation path of the signal and extracting the instantaneous frequency for target detection.

[0077] In a specific embodiment, the frequency domain radiation source signal is subjected to convolution processing based on Gaussian window function transformation, the main purpose of which is to eliminate noise and interference and retain the main radiation source signal.

[0078] Specifically, the short-time Fourier transformed signal is convolved with the Gaussian window function:

[0079] S h (t)=S(ω,t)*g(t)

[0080] In the formula, S h (t) is the convolution result, i.e. the main radiation source signal; S(ω,t) is the frequency domain radiation source signal; g(t) is the Gaussian window function;

[0081] In a specific embodiment, the energy discrimination value is calculated based on the effective signal threshold, the purpose of which is to distinguish effective signals from invalid noises through signal energy characteristics. The energy discrimination value is calculated by the following formula, including:

[0082] C=M·std(x)+th

[0083] Where M is the mean of the main radiation source signal, std(x) is the standard deviation of the main radiation source signal, th is the effective signal threshold, and C is the energy discrimination value.

[0084] It should be noted that the standard deviation std(x) is used to measure the discreteness of the main radiation source signal. The specific calculation formula is:

[0085]

[0086] Among them, x i is the i-th sample value of the main radiation source signal; M is the mean value of the main radiation source signal, and the calculation formula is:

[0087]

[0088] Where N1 is the total number of signal samples.

[0089] It should be noted that the effective signal threshold th is set according to experimental data and is used to distinguish effective signals from noise. Specifically, the standard deviation of multiple groups of radiation source signal samples is calculated, and the standard deviation distribution range of different signals is recorded; a suitable value is selected in the standard deviation distribution range as the threshold th, which is set to the middle value between the maximum standard deviation of the noise signal and the minimum standard deviation of the effective signal. For example, assuming that the standard deviation range of the noise signal obtained by experimental statistics is [0.01, 0.05], and the standard deviation range of the effective signal is [0.06, 0.2], then th = 0.055 can be selected as the threshold for distinguishing effective signals.

[0090] Next, the energy discrimination value is compared with the preset energy discrimination threshold, and when the energy discrimination value is calculated as: greater than the preset energy discrimination threshold, the energy discrimination value is calculated as: the corresponding main radiation source signal is determined as the effective radiation source signal;

[0091] It should be noted that the preset energy discrimination threshold is obtained through signal characteristic analysis. The specific implementation process is to first collect a large number of radiation source signal samples, including effective signals and noise signals, and calculate their energy discrimination values; then perform distribution statistics on the calculated energy discrimination values ​​to observe the energy discrimination value ranges of effective signals and noise signals. Among these distribution data, the middle value between the maximum value of the noise signal energy discrimination value and the minimum value of the effective signal energy discrimination value is selected as the preset energy discrimination threshold.

[0092] In a specific embodiment, the effective radiation source signal is subjected to differential processing, the purpose of which is to further extract the pulse signal and distinguish the main component and interference in the signal. Specifically, the effective radiation source signal is subjected to differential calculation to obtain the pulse difference;

[0093] When the pulse difference is greater than zero, it indicates that the signal amplitude rises and it is a pulse signal, and the effective radiation source signal corresponding to the pulse difference is determined as a clear radiation source signal.

[0094] In step S12, it is necessary to perform a vector quantization operation according to the clear radiation source signal to obtain a quantized radiation source signal, including:

[0095] In a specific embodiment, the clear radiation source signal is normalized using a normalization formula to obtain a normalized radiation source signal, wherein the normalization formula is:

[0096]

[0097] in, is the normalized radiation source signal; d is the pulse width; ω i is the corresponding pulse frequency; A i is the corresponding pulse amplitude; N c,p is the number of pulse signals;

[0098] It should be noted that the normalization process eliminates the scale differences of the feature components, making the feature vectors have a more balanced representation in the multidimensional space. The normalization process reduces the impact of amplitude, frequency and quantity changes on the feature components, and improves the accuracy of subsequent vector quantization.

[0099] In a specific embodiment, the normalized radiation source signal is subjected to a feature vector construction operation to obtain a signal feature vector, which is defined as:

[0100] d(t)=[a1(t),a2(t),…,a M (t)]

[0101] Among them, d(t) is the signal feature vector; a i (t) is the i-th normalized component; M is the dimension of the signal feature vector;

[0102] It should be noted that the signal feature vector is used to improve the input data for subsequent similarity quantification.

[0103] In a specific embodiment, the preset weight matrix W ij It is the core of the vector quantization operation and is used to store the reference vector. The similarity calculation between the reference vector and the input signal feature vector determines the quantization mapping result.

[0104] First, define the preset weight matrix, the preset weight matrix W ij is a two-dimensional matrix, each row of which represents a reference vector and each column corresponds to a feature dimension:

[0105]

[0106] Among them, W ij is the preset weight matrix; n is the number of reference vectors; m is the dimension of the feature vector;

[0107] Next, a random initialization process is performed to randomly set each element of the matrix to a small value to ensure the diversity of the reference vector, and the preset weight matrix W is obtained. ij .

[0108] Specifically, similarity calculation is used to measure the distance between the input feature vector and each reference vector in the preset weight matrix. The commonly used measurement method is Euclidean distance, which is used to calculate the similarity between the input feature vector d(t) and each row (reference vector) of the preset weight matrix:

[0109]

[0110] Among them, Sim ij is the similarity between the input feature vector and the jth reference vector of the preset weight matrix; d(t) is the signal feature vector; W ij is the preset weight matrix; d k (t) is the kth component of the eigenvector; w ij,k is the kth component of the jth reference vector in the preset weight matrix; m is the dimension of the eigenvector.

[0111] In a specific embodiment, according to the difference between the input feature vector and the current weight matrix, the weight matrix is ​​updated using the learning rate η. Specifically, according to the similarity quantization result, the reference vector closest to the input feature vector is determined, and the weight of the reference vector is gradually adjusted using an update formula to make it closer to the input feature vector, wherein the update formula is:

[0112]

[0113] in, is the weight matrix in the k+1th iteration; is the weight matrix in the kth iteration; d(t) is the signal feature vector; η is the learning rate, which is used to control the update step size.

[0114] In a specific embodiment, an updated weight matrix is ​​obtained by updating the weight operation, and the matrix contains a preset feature space reference vector. In this process, the input feature vector is mapped to the updated weight matrix. Specifically, the input feature vector is first preprocessed to ensure that it has a standardized representation in the feature space. Then, the updated weight matrix is ​​traversed to calculate the similarity between each reference vector and the input feature vector. Common similarity calculation methods include Euclidean distance or cosine similarity. During the similarity calculation process, the difference between the input feature vector and each reference vector is quantified to ensure that the closest reference vector is found.

[0115] It should be noted that during the traversal process, the system will continuously calculate the similarity between each reference vector and the input feature vector, and compare them one by one, and finally select the reference vector that is most similar to the input feature vector. This reference vector is the quantized output vector, which represents the final representation of the quantized radiation source signal. Through this process, the input signal is optimized and updated through the weight matrix to obtain a more accurate quantization result, which can be effectively mapped to the optimal reference point in the feature space, thereby improving the accuracy and stability of radiation source identification. In addition, the operation of updating the weight matrix can be adaptively adjusted through optimization algorithms such as the gradient descent method, so that the process can continue to improve and obtain more accurate quantization results when facing different radiation source signals, further enhancing the self-learning ability of the system.

[0116] In step S13, the quantized radiation source signal needs to be input into a pre-trained signal screening model to obtain a radiation source signal to be identified, including:

[0117] In a specific embodiment, it is necessary to obtain training data, including historical real distances, which are used to describe the average distance between the signal feature vector and the real target; the historical weight matrix W history , used to represent the weight matrix used for screening in previous training; historical bias matrix B history , which is used to represent the historical bias matrix used for screening.

[0118] First, initialize the weight matrix and bias matrix, build the initial neural network model, and map the input features to the output categories. The model is defined as:

[0119] y=f(W'·x+B)

[0120] Among them, y is the output signal category; x is the input quantized radiation source signal; f is the activation function; W' is the weight matrix of the initial neural network model; B is the bias matrix of the initial neural network model.

[0121] Next, use the mean square error as the loss function:

[0122]

[0123] Among them, y i For the historical real distance; is the model prediction value; N2 is the number of training samples; L is the loss value.

[0124] Next, the gradient descent method is used to update the weight matrix and bias matrix:

[0125]

[0126] Among them, η is the learning rate; k is the number of training iterations.

[0127] In a specific embodiment, the predicted value is repeatedly calculated. Calculate the loss value L and update the weight matrix and bias matrix until the maximum number of training times (set to 200-300 times) is reached, determine that the training is completed, and obtain the trained signal screening model f(W',B).

[0128] In a specific embodiment, the quantized radiation source signal generated in step S12 is input into a trained signal screening model, and the screening result is output through forward propagation to obtain the radiation source signal to be identified output by the screening model.

[0129] In step S14, it is necessary to perform a nearest neighbor node extraction operation according to the radiation source signal to be identified to obtain a nearest neighbor node set, including:

[0130] In a specific embodiment, the acquisition of the preset feature space reference node set is mainly derived from the historical radiation source signal, which is generated by feature extraction, cluster analysis or pre-definition. First, a large number of historical radiation source signals are sampled, and features such as amplitude, frequency, and phase are extracted to form an m-dimensional feature vector set. Then, a clustering algorithm (such as K-means) is used to perform cluster analysis on these feature vectors, calculate the cluster center of each cluster, and extract several center points as reference nodes to ensure that the main distribution of the feature space is covered. If the node coverage needs to be enhanced, random perturbations can be added to the existing set to generate new reference points, and the node distribution can be optimized after removing outliers. Finally, the generated reference node set is stored.

[0131] In a specific embodiment, the preset node threshold is determined based on the statistical characteristics of the signal distribution in the feature space and actual needs. Specifically, first, a large number of historical radiation source signal samples are collected, and the Euclidean distance between each signal and its corresponding real target is calculated to obtain the statistical data of the distance distribution. Then, based on the distribution data, the threshold is set to the maximum distance value that satisfies 90%-95% sample coverage, thereby ensuring that the threshold can contain enough neighboring nodes and effectively exclude abnormal signals and noise interference.

[0132] Specifically, the Euclidean distance is used to calculate the distance between the radiation source signal to be identified and each reference node in the preset feature space reference node set, all calculated distances are stored as a node distance set, the node distance set that is less than the preset node distance threshold in the node distance set is extracted, and the node set corresponding to the node distance set is determined as the nearest neighbor node set.

[0133] In step S15, it is necessary to calculate based on the nearest neighbor node set and the preset target node to obtain the signal distance, including:

[0134] In a specific embodiment, the signal distance is calculated by the following formula:

[0135]

[0136] Where D ik is the signal distance; N is the total number of nearest neighbor nodes; d i is the i-th nearest neighbor node; R i is the preset target node; ||d(t) i -R i ||2 is the L2 norm between the i-th nearest neighbor node and the preset target node.

[0137] It should be noted that the setting of the preset target node can also be achieved by directly defining the range and distribution of the key characteristic values. Specifically, first, based on the characteristic dimensions of the radiation source signal to be identified (such as amplitude, frequency, phase, etc.), the value range of each feature is determined, such as the amplitude in [0.1, 1.0], the frequency in [100, 1000], and the phase in [0, 2π]. Then, each characteristic dimension is segmented, for example, the amplitude is divided into 5 intervals (0.1, 0.3, 0.5, 0.7, 0.9), the frequency is divided into 4 intervals (200, 400, 600, 800), and the phase is divided into 3 intervals (0, π, 2π). The target node is generated by the Cartesian product of these characteristic intervals. For example, one target node is R1 = [0.5, 600, π], and the other is R2 = [0.7, 800, 0]. The advantage of this method is that the granularity and coverage of the feature distribution can be flexibly controlled. If some feature dimensions are more important, the density of their interval division can be increased to generate more targeted target nodes. For example, in radar radiation source signals, if the frequency change is more critical to identification, the frequency dimension can be divided more densely, such as every 50Hz as an interval, while the amplitude and phase can be divided into fewer intervals. The target nodes finally generated can be directly used for signal distance calculation tasks.

[0138] In step S16, it is necessary to perform a maximum value extraction operation according to the signal distance to obtain a maximum signal distance, and determine the radiation source corresponding to the maximum signal distance as the final radiation source, including:

[0139] In a specific embodiment, by extracting the maximum signal distance, the purpose is to ensure that the recognition process focuses on those target nodes that are most different from the signal to be identified. The signal distance reflects the degree of difference between the signal to be identified and the reference target node signal. A larger signal distance (0.8) means that the characteristics of the signal to be identified and the target node are more different, and this difference indicates that the target node is the best match for the signal to be identified. However, the extraction of the maximum signal distance does not mean that we are looking for the most similar target node, but for the most dissimilar node. By finding this most dissimilar node, we can more accurately exclude non-compliant candidate targets and reduce the probability of misidentification. The extraction of the maximum signal distance helps us ensure that the identified radiation source is the most realistic among a group of candidates.

[0140] For example, suppose that in a radiation source identification task of a certain radar system, the signal distances of the signal to be identified from different reference target nodes are calculated to be 0.3, 0.5 and 0.8, respectively, where 0.8 is the maximum value. According to the definition of signal distance, a larger distance means that the target node is more different from the signal to be identified. In actual operation, this differentiated calculation can help the system identify radiation sources that vary greatly with factors such as the environment and signal characteristics. For example, the signal distance of the target node 0.8 is the largest, indicating that there is a large difference between its characteristics and the signal to be identified, and this difference is caused by factors such as environmental noise and signal interference. Therefore, the system selects the node corresponding to the maximum signal distance as the radiation source to be finally identified and completes the radiation source identification task.

[0141] Through this method, the maximum signal distance not only helps the system exclude targets that do not meet the conditions, but also effectively reduces the risk of identification errors or misidentifications in actual applications. For example, in a complex radio wave propagation environment, different radiation sources will present different signal characteristics due to changes in terrain, climate and other factors. By finding the node with the largest signal distance, the system can more accurately identify the true radiation source, ensuring the accuracy and reliability of the identification process.

[0142] In summary, the present invention discloses a radar radiation source identification method. Compared with the prior art, the present invention can effectively extract and process clear radiation source signals by performing signal processing and vector quantization on the original radiation source signal, thereby reducing the interference of environmental noise on the signal. This signal processing process combines short-time Fourier transform, Gaussian window convolution processing and energy discrimination based on effective signal threshold, which not only improves the clarity of the signal, but also can more accurately extract the effective radiation source signal. Through differential processing and pulse difference judgment, the effective radiation source signal is further screened as a clear signal, thereby improving the accuracy of radiation source identification. The method effectively eliminates redundant information in the signal through signal quantization and screening model training, ensuring the accuracy of subsequent identification operations. Secondly, the present invention realizes more accurate radiation source identification by introducing the nearest neighbor node extraction and signal distance calculation method based on distance. Node distance calculation and maximum signal distance extraction are performed in the feature space, which can not only locate the target node with the largest difference from the signal to be identified, but also effectively distinguish the differences between different radiation sources. By setting an appropriate neighbor node distance threshold, the node extraction process is further optimized to ensure the robustness of the recognition algorithm in a complex environment. In addition, the signal screening model training based on historical real distance, weight matrix and bias matrix enables this method to gradually adapt to the characteristics of new radiation sources, has strong learning ability, can continuously optimize the recognition effect over time, and can improve the recognition accuracy of radar radiation source signals in complex electronic countermeasure environments.

[0143] Reference Figure 2 The second embodiment of the present invention provides a radar radiation source identification method and device, including:

[0144] A signal processing module, used to obtain an original radiation source signal and perform a signal processing operation on the original radiation source to obtain a clear radiation source signal;

[0145] A signal quantization module, used for performing a vector quantization operation according to the clear radiation source signal to obtain a quantized radiation source signal;

[0146] A training model module, used for inputting the quantized radiation source signal into a pre-trained signal screening model to obtain a radiation source signal to be identified;

[0147] A node distance module is used to perform a nearest neighbor node extraction operation according to the radiation source signal to be identified to obtain a nearest neighbor node set;

[0148] A signal distance module, used to calculate according to the nearest neighbor node set and the preset target node to obtain the signal distance;

[0149] The radiation source determination module is used to perform a maximum value extraction operation according to the signal distance to obtain a maximum signal distance, and determine the radiation source corresponding to the maximum signal distance as a final radiation source.

[0150] Preferably, the signal processing module is specifically used to obtain an original radiation source signal and perform a signal processing operation on the original radiation source to obtain a clear radiation source signal, including:

[0151] The obtaining of the original radiation source signal and performing signal processing on the original radiation source to obtain a clear radiation source signal includes:

[0152] Performing a short-time Fourier transform on the original radiation source signal to obtain a frequency domain radiation source signal;

[0153] Based on Gaussian window function transformation, convolution processing is performed on the frequency domain radiation source signal to obtain a main radiation source signal;

[0154] Based on a preset effective signal threshold, a calculation is performed according to the main radiation source signal to obtain an energy discrimination value;

[0155] The energy discrimination value is compared with a preset energy discrimination threshold, and when the energy discrimination value is greater than the preset energy discrimination threshold, the main radiation source signal corresponding to the energy discrimination value is determined as a valid radiation source signal;

[0156] Performing differential processing on the effective radiation source signal to obtain a pulse difference;

[0157] When the pulse difference is greater than zero, the effective radiation source signal corresponding to the pulse difference is determined as a clear radiation source signal.

[0158] The calculation formula of the energy discrimination value is:

[0159] C=M·std(x)+th

[0160] Where M is the mean of the main radiation source signal, std(x) is the standard deviation of the main radiation source signal, th is the effective signal threshold, and C is the energy discrimination value.

[0161] Preferably, the signal quantization module is specifically used to perform a vector quantization operation according to the clear radiation source signal to obtain a quantized radiation source signal, including:

[0162] The step of performing a vector quantization operation according to the clear radiation source signal to obtain a quantized radiation source signal includes:

[0163] Normalizing the clear radiation source signal to obtain a normalized radiation source signal;

[0164] Performing a feature vector construction operation on the normalized radiation source signal to obtain a signal feature vector;

[0165] Based on a preset quantization weight matrix, a similarity quantization operation is performed on the signal feature vector to obtain vector similarity;

[0166] According to the preset quantization weight matrix and the vector similarity, an operation of updating the weight matrix is ​​performed to obtain an updated weight matrix;

[0167] A quantization mapping operation is performed according to the updated weight matrix and the signal feature vector to obtain a quantized radiation source signal.

[0168] Preferably, the training model module is specifically used to input the quantized radiation source signal into a pre-trained signal screening model to obtain the radiation source signal to be identified, including:

[0169] The training process of the signal screening model includes:

[0170] Obtain the historical true distance, historical weight matrix and historical bias matrix of the quantized radiation source signal;

[0171] An initial signal screening model is constructed according to the historical weight matrix and the historical bias matrix, and the initial signal screening model is trained based on the historical real distance and the quantized radiation source signal as input data to obtain a trained signal screening model.

[0172] The step of constructing an initial signal screening model according to the historical weight matrix and the historical bias matrix, and training the initial signal screening model based on the historical real distance and the quantized radiation source signal as input data to obtain a trained signal screening model includes:

[0173] Initializing the historical weight matrix and the historical bias matrix of the initial signal screening model;

[0174] Based on the gradient descent optimization method, matrix adjustment is performed to obtain an adjustment weight matrix and an adjustment bias matrix;

[0175] Perform matrix updating and training according to the adjustment weight matrix and the adjustment bias matrix;

[0176] When the number of training times is greater than or equal to the preset maximum number of training times, the training is determined to be completed, and a trained signal screening model is obtained.

[0177] Preferably, the node distance module is specifically used to perform a nearest neighbor node extraction operation according to the radiation source signal to be identified to obtain a nearest neighbor node set, including:

[0178] The step of performing a nearest neighbor node extraction operation according to the radiation source signal to be identified to obtain a nearest neighbor node set includes:

[0179] Calculating based on the radiation source signal to be identified and a preset feature space reference node set to obtain a node distance set;

[0180] A node distance set that is less than a preset node distance threshold value is extracted from the node distance set, and a node set corresponding to the node distance set is determined as a nearest neighbor node set.

[0181] Preferably, the signal distance module is specifically used to calculate according to the nearest neighbor node set and the preset target node to obtain the signal distance, including:

[0182] The calculating according to the nearest neighbor node set and the preset target node to obtain the signal distance includes:

[0183] The signal distance is calculated by the following formula:

[0184]

[0185] Where D ik is the signal distance; N is the total number of nearest neighbor nodes; d i is the i-th nearest neighbor node; R i is the preset target node; ||d(t) i -R i ||2 is the L2 norm between the i-th nearest neighbor node and the preset target node.

[0186] Preferably, the radiation source determination module is specifically configured to perform a maximum value extraction operation according to the signal distance to obtain a maximum signal distance, and determine the radiation source corresponding to the maximum signal distance as the final radiation source.

[0187] It should be noted that a radar radiation source identification device provided in an embodiment of the present invention is used to execute all process steps of a radar radiation source identification method in the above embodiment, and the working principles and beneficial effects of the two correspond one to one, so they will not be repeated here.

[0188] An embodiment of the present invention further provides an electronic device. The electronic device includes: a processor, a memory, and a computer program stored in the memory and executable on the processor, such as a radiation source determination program. When the processor executes the computer program, the steps in the above-mentioned radar radiation source identification method embodiments are implemented, such as Figure 1 Alternatively, when the processor executes the computer program, the functions of the modules / units in the above-mentioned device embodiments are implemented, such as the training model module.

[0189] Exemplarily, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to implement the present invention. The one or more modules / units may be a series of computer program instruction segments capable of implementing specific functions, which are used to describe the execution process of the computer program in the electronic device.

[0190] The electronic device may be a computing device such as a desktop computer, a notebook, a PDA, and a smart tablet. The electronic device may include, but is not limited to, a processor and a memory. Those skilled in the art will appreciate that the above components are merely examples of electronic devices and do not constitute a limitation on the electronic device. The electronic device may include more or fewer components than the above components, or may combine certain components, or different components. For example, the electronic device may also include input and output devices, network access devices, buses, etc.

[0191] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the electronic device, and uses various interfaces and lines to connect various parts of the entire electronic device.

[0192] The memory can be used to store the computer program and / or module, and the processor realizes various functions of the electronic device by running or executing the computer program and / or module stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0193] Wherein, if the module / unit integrated in the electronic device is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.

[0194] It should be noted that the device embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the accompanying drawings of the device embodiments provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines. A person of ordinary skill in the art may understand and implement it without paying any creative effort.

[0195] The specific embodiments described above further describe the purpose, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. It is particularly pointed out that for those skilled in the art, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of protection of the present invention.

Claims

1. A radar radiation source identification method, characterized in that include: Acquire an original radiation source signal, and perform a signal processing operation on the original radiation source to obtain a clear radiation source signal; Performing a vector quantization operation according to the clear radiation source signal to obtain a quantized radiation source signal; Inputting the quantized radiation source signal into a pre-trained signal screening model to obtain a radiation source signal to be identified; According to the radiation source signal to be identified, a nearest neighbor node extraction operation is performed to obtain a nearest neighbor node set; Calculate the signal distance based on the nearest neighbor node set and the preset target node; According to the signal distance, a maximum value extraction operation is performed to obtain a maximum signal distance, and a radiation source corresponding to the maximum signal distance is determined as a final radiation source.

2. The radar radiation source identification method according to claim 1, characterized in that: The obtaining of the original radiation source signal and performing signal processing on the original radiation source to obtain a clear radiation source signal includes: Performing a short-time Fourier transform on the original radiation source signal to obtain a frequency domain radiation source signal; Based on Gaussian window function transformation, convolution processing is performed on the frequency domain radiation source signal to obtain a main radiation source signal; Based on a preset effective signal threshold, a calculation is performed according to the main radiation source signal to obtain an energy discrimination value; The energy discrimination value is compared with a preset energy discrimination threshold, and when the energy discrimination value is greater than the preset energy discrimination threshold, the main radiation source signal corresponding to the energy discrimination value is determined as a valid radiation source signal; Performing differential processing on the effective radiation source signal to obtain a pulse difference; When the pulse difference is greater than zero, the effective radiation source signal corresponding to the pulse difference is determined as a clear radiation source signal.

3. The radar radiation source identification method according to claim 2, characterized in that: The calculation formula of the energy discrimination value is: C=M·std(x)+th Where M is the mean of the main radiation source signal, std(x) is the standard deviation of the main radiation source signal, th is the effective signal threshold, and C is the energy discrimination value.

4. The radar radiation source identification method according to claim 1, characterized in that: The step of performing a vector quantization operation according to the clear radiation source signal to obtain a quantized radiation source signal includes: Normalizing the clear radiation source signal to obtain a normalized radiation source signal; Performing a feature vector construction operation on the normalized radiation source signal to obtain a signal feature vector; Based on a preset quantization weight matrix, a similarity quantization operation is performed on the signal feature vector to obtain vector similarity; According to the preset quantization weight matrix and the vector similarity, an operation of updating the weight matrix is ​​performed to obtain an updated weight matrix; A quantization mapping operation is performed according to the updated weight matrix and the signal feature vector to obtain a quantized radiation source signal.

5. The radar radiation source identification method according to claim 1, characterized in that: The training process of the signal screening model includes: Obtain the historical true distance, historical weight matrix and historical bias matrix of the quantized radiation source signal; An initial signal screening model is constructed according to the historical weight matrix and the historical bias matrix, and the initial signal screening model is trained based on the historical real distance and the quantized radiation source signal as input data to obtain a trained signal screening model.

6. The radar radiation source identification method according to claim 5, characterized in that: The step of constructing an initial signal screening model according to the historical weight matrix and the historical bias matrix, and training the initial signal screening model based on the historical real distance and the quantized radiation source signal as input data to obtain a trained signal screening model includes: Initializing the historical weight matrix and the historical bias matrix of the initial signal screening model; Based on the gradient descent optimization method, matrix adjustment is performed to obtain an adjustment weight matrix and an adjustment bias matrix; Perform matrix updating and training according to the adjustment weight matrix and the adjustment bias matrix; When the number of training times is greater than or equal to the preset maximum number of training times, the training is determined to be completed, and a trained signal screening model is obtained.

7. The radar radiation source identification method according to claim 1, characterized in that: The step of performing a nearest neighbor node extraction operation according to the radiation source signal to be identified to obtain a nearest neighbor node set includes: Calculating based on the radiation source signal to be identified and a preset feature space reference node set to obtain a node distance set; A node distance set that is less than a preset node distance threshold value is extracted from the node distance set, and a node set corresponding to the node distance set is determined as a nearest neighbor node set.

8. The radar radiation source identification method according to claim 1, characterized in that: The calculating according to the nearest neighbor node set and the preset target node to obtain the signal distance includes: The signal distance is calculated by the following formula: Where D ik is the signal distance; N is the total number of nearest neighbor nodes; d i is the i-th nearest neighbor node; R i is the preset target node; ||d(t) i -R i ||2 is the L2 norm between the i-th nearest neighbor node and the preset target node.

9. A radar radiation source identification method and device, characterized in that: include: A signal processing module, used to obtain an original radiation source signal and perform a signal processing operation on the original radiation source to obtain a clear radiation source signal; A signal quantization module, used for performing a vector quantization operation according to the clear radiation source signal to obtain a quantized radiation source signal; A training model module, used for inputting the quantized radiation source signal into a pre-trained signal screening model to obtain a radiation source signal to be identified; A node distance module is used to perform a nearest neighbor node extraction operation according to the radiation source signal to be identified to obtain a nearest neighbor node set; A signal distance module, used to calculate according to the nearest neighbor node set and the preset target node to obtain the signal distance; The radiation source determination module is used to perform a maximum value extraction operation according to the signal distance to obtain a maximum signal distance, and determine the radiation source corresponding to the maximum signal distance as a final radiation source.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the radar radiation source identification method according to any one of claims 1 to 8.

Images