Radar signal sorting method based on convex clustering

Abstract

The invention relates to a radar signal sorting method based on convex clustering, and the method comprises the following steps: obtaining pulse signals from different radar signal sources detected by a receiver, carrying out the standardization processing of the parameters of the detected pulse signals of different radars, constructing a convex optimization function corresponding to the signal sorting, and carrying out the sorting of the signals. Performing optimization solution on the convex optimization objective function by using an alternating minimization method, iterating the steps until the variable quantity of the convex objective function is smaller than a preset threshold value, obtaining a global optimal solution of the convex optimization function, comparing Euclidean distances among vectors of the optimal solution, and when the distances are smaller than the preset threshold value, obtaining a global optimal solution of the convex optimization function; and classifying the pulse signals represented by the similar vectors into the same class cluster to obtain a final signal sorting result, so that the method has the advantages of high precision and good stability and noise robustness.

Description

Technical Field

[0001] The invention belongs to the technical field of electronic countermeasure radar signal processing, and in particular relates to a radar signal sorting method based on convex clustering. Background Art

[0002] Signal sorting is one of the key technologies in radar signal processing. Its main task is to separate and identify radar signals from different radiation sources in a complex electromagnetic environment. With the development of modern electronic warfare, the radar signal environment is becoming increasingly complex and the signal modulation parameters are changeable, which poses a severe challenge to radar signal sorting technology. Radar signal sorting requires separating each radar pulse train from the complex and interlaced pulse stream, and then analyzing the signal parameters of each pulse train, identifying the radiation source type, and judging the threat level. As a result, signal sorting technology not only affects the accuracy and efficiency of radar signal detection and processing, but is also the core step in electronic battlefield situation awareness, which has important military significance and practical value.

[0003] Currently, multi-parameter based clustering methods are widely used in radar signal sorting tasks. As an unsupervised classification method, clustering algorithm discovers the potential rules and patterns in the data by mining the correlation between input data. It is suitable for solving radar signal sorting problems that lack prior knowledge. It mainly uses parameters such as carrier frequency, pulse width and arrival angle in the signal pulse descriptor to cluster and perform correlation analysis on radar signals to ensure the accuracy of radar signal sorting.

[0004] The existing signal sorting algorithms are mainly improved on several mainstream methods such as the K-means clustering algorithm and the fuzzy C-means clustering algorithm. First of all, the K-means clustering algorithm has been widely used in radar signal sorting due to its simple algorithm, fast convergence speed and no need for prior knowledge. However, the traditional K-means algorithm requires manual preset of the number of clusters and is sensitive to the cluster centers, which affects the final signal sorting accuracy. At the same time, these existing clustering algorithms and their improvements are based on the optimization of non-convex objective functions, which are very easy to fall into local optimal solutions and are sensitive to the order of sample input, making it difficult to ensure the stability of solutions and performance. Therefore, a radar signal sorting method based on convex clustering is urgently needed to solve the problems of low accuracy and unstable results of the radar signal sorting methods existing in the above-mentioned clustering algorithms. Summary of the invention

[0005] In view of this, the present invention proposes a radar signal sorting method based on convex clustering, which is applied to the field of electronic countermeasure radar signal processing technology. It can solve the technical problems of existing clustering algorithms, such as low accuracy of radar signal sorting methods, reliance on manually preset number of clusters, sensitivity to signal input order and unstable results.

[0006] In order to achieve the above technical objectives, the specific technical solutions adopted by the present invention are:

[0007] A radar signal sorting method based on convex clustering comprises the following steps:

[0008] S1. Obtain pulse signals from different radar signal sources detected by the receiver;

[0009] S2. Standardize the parameters of pulse signals received from different radars;

[0010] S3, constructing a convex optimization function corresponding to signal sorting;

[0011] S4, using the alternating minimization method to optimize and solve the convex optimization objective function;

[0012] S5, iterating the above step S4 until the change of the convex objective function is less than a preset threshold, obtaining the global optimal solution of the convex optimization function;

[0013] S6. Compare the Euclidean distances between the vectors of the optimal solution. When the distance is less than a preset threshold, classify the pulse signals represented by similar vectors into the same cluster to obtain the final signal sorting result.

[0014] Furthermore, in step S2, the detected pulse signal parameters are distributed in the same [0,1] interval and participate in the radar signal sorting task with the same dimension. The pulse signal parameters include carrier frequency, pulse width and arrival angle. The pulse signal parameter matrix is:

[0015]

[0016] Among them, cf i ,pw i ,doa i ,i=1,...,N are the frequency, pulse width and arrival direction of the i-th pulse respectively, and N is the total number of pulse signals.

[0017] Furthermore, in step S2, the mean and standard deviation of the k-th dimension data of the pulse signal parameter matrix are:

[0018]

[0019]

[0020] Among them, 1≤k≤3 is the dimension of the pulse signal parameter matrix; the standardized value of the sample pulse vector is obtained:

[0021]

[0022] The extreme value normalization formula is used to make the normalized value of the pulse signal matrix distributed in the interval [0,1]. The formula is as follows:

[0023]

[0024] Among them, pdw′ kmin and pdw′ kmax They are pdw′ ik The minimum and maximum values ​​in .

[0025] Furthermore, in step S3, the convex optimization function includes a loss function and a regularization term, and the optimization objective function is:

[0026]

[0027] Among them, the first term is the loss function, and the second term is the regularization term and generally takes the l1 norm. i is the normalized pulse signal parameter, u i It is pdw i The cluster center, weight parameter ω ij Non-negative and ω ij =ω ji , γ is a non-negative tuning constant.

[0028] Furthermore, in step S4, the convex clustering problem is redefined as the following equivalent constraint problem:

[0029]

[0030]

[0031] According to the constraint solving method, the augmented Lagrangian expression is as follows:

[0032]

[0033] Among them, l is the constructed centroid pair l = (l1, l2) and l1 <l2,ε={l=(l1,l2):ω l >0} is an edge set defined on non-zero weights, and a new variable is introduced to define the difference between the centroids.

[0034] Furthermore, in step S4, the alternating minimization method is used to solve the minimum value of the convex optimization function, and the convex optimization objective function is optimized and solved. The m-times iteration process is as follows:

[0035]

[0036] in, is z to the closed convex set C l The mapping of Cl ={λ l :||λ l || + ≤γω l}, yes The dual norm of , υ is the iteration step size,

[0037] By adopting the above technical solution, the present invention can also bring the following beneficial effects:

[0038] 1. The present invention mentions a radar signal sorting method based on convex clustering. Compared with the traditional clustering method based on non-convex objective function optimization, the present invention can obtain the global optimal solution of signal sorting by fusing cluster centers, more accurately identify and separate different radar signal sources, thereby improving the accuracy of signal sorting, and can also make the radar signal sorting effect not affected by the order of signal input, with higher stability. Even when the signal characteristics are complex and changeable and the noise interference is large, it can still maintain a stable clustering effect and has high noise robustness. It can also make full use of the multi-dimensional characteristics of radar signals for cluster analysis, and has the advantages of comprehensively considering information in multiple dimensions, comprehensively characterizing radar signal characteristics, improving the accuracy of signal sorting and providing effective support for electronic countermeasure systems.

[0039] 2. The present invention mentions a radar signal sorting method based on convex clustering, which can efficiently and accurately separate different signal sources from a complex electromagnetic environment, provide strong support for military decision-making and intelligence analysis, and is applied to electronic reconnaissance tasks that require accurate identification of target radars. It is a key step in extracting valuable intelligence, and can also be used in electronic confrontation scenarios under low signal-to-noise ratio conditions. It has good algorithm stability and noise robustness, and has great room for promotion. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0041] Figure 1 A flowchart of a radar signal sorting method based on convex clustering is mentioned in a specific embodiment of the present invention;

[0042] Figure 2 It is a radar pulse signal parameter information data diagram in a specific implementation manner of the present invention;

[0043] Figure 3It is a data diagram of signal sorting accuracy information in a specific implementation manner of the present invention;

[0044] Figure 4 This is a stability comparison diagram of the specific implementation of the present invention with K-means;

[0045] Figure 5 This is a comparison chart of the sorting accuracy of the specific implementation of the present invention and K-means under different signal-to-noise ratios. DETAILED DESCRIPTION

[0046] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0047] The following describes the embodiments of the present invention through specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work belong to the scope of protection of the present invention.

[0048] It should be noted that various aspects of the embodiments within the scope of the appended claims are described below. It should be apparent that the aspects described herein can be embodied in a wide variety of forms, and any specific structure and / or function described herein is merely illustrative. Based on the present invention, it should be understood by those skilled in the art that an aspect described herein can be implemented independently of any other aspect, and two or more of these aspects can be combined in various ways. For example, any number of aspects described herein can be used to implement the device and / or practice the method. In addition, other structures and / or functionalities other than one or more of the aspects described herein can be used to implement this device and / or practice this method.

[0049] It should also be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present invention. The drawings only show components related to the present invention rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.

[0050] Additionally, in the following description, specific details are provided to facilitate a thorough understanding of the examples. However, it will be understood by those skilled in the art that the aspects described may be practiced without these specific details.

[0051] In one embodiment of the present invention, Figure 1 As shown, the present invention mentions a radar signal sorting method based on convex clustering, comprising the following steps:

[0052] In step S1, the pulse signals from various signal sources detected by the receiver are obtained;

[0053] In step S2, the detected pulse signal parameters are standardized to highlight the essential characteristics of the data, wherein the pulse signal parameters include carrier frequency, pulse width and arrival angle;

[0054] Specifically, for the N pulse signals received by the receiver, the corresponding pulse signal parameter matrix is ​​constructed:

[0055]

[0056] Among them, cf i ,pw i ,doa i ,i=1,...,N are the frequency, pulse width and arrival direction of the ith pulse respectively, and N is the total number of pulse signals. The mean and standard deviation of the k-th dimension of the pulse signal parameter matrix are:

[0057]

[0058]

[0059] Among them, 1≤k≤3 is the dimension of the pulse signal parameter matrix. From this, the standardized value of the sample pulse vector can be obtained:

[0060]

[0061] The extreme value normalization formula is used to make the normalized values ​​of the pulse signal matrix distributed in the interval [0,1]:

[0062]

[0063] Among them, pdw′ kmin and pdw′ kmax They are pdw′ ik The minimum and maximum values ​​in .

[0064] In step S3, a convex optimization function corresponding to signal sorting is constructed, wherein the convex optimization function includes a loss function and a regularization term.

[0065] Specifically, the convex optimization objective function of signal sorting is:

[0066]

[0067] Among them, the first term is the loss function, and the second term is the regularization term and generally takes the l1 norm. i is the normalized pulse signal parameter, u i It is pdw i The cluster center, weight parameter ω ij Non-negative and ω ij =ω ji , γ is a non-negative tuning constant.

[0068] In step S4, an alternating minimization method is used to find the minimum value of the convex optimization function.

[0069] Specifically, we first redefine the convex clustering problem as the following equivalence constraint problem:

[0070]

[0071] Among them, l is the constructed centroid pair l = (l1, l2) and l1 <l2,ε={l=(l1,l2):ω l >0} is an edge set defined on non-zero weights, and a new variable is introduced To define the difference between the centroids. The purpose of variable splitting is to simplify the optimization of the regularization term. According to the constraint solving method, the augmented Lagrangian expression is given:

[0072]

[0073] The alternating minimization method is used to optimize and solve the convex optimization objective function. The m-th iteration process is as follows:

[0074]

[0075] in, is z to the closed convex set C l The mapping of C l ={λ l :||λ l || + ≤γω l}, yes The dual norm of , υ is the iteration step size,

[0076] In step S5, the step S4 is iterated until the change of the convex objective function is less than a preset threshold, thereby obtaining the global optimal solution of the convex optimization function.

[0077] In step S6, the Euclidean distances between the vectors of the optimal solution are compared. When the distance is less than a preset threshold, the pulse signals represented by similar vectors are classified into the same cluster to obtain the final signal sorting result.

[0078] like Figure 2 As shown, the present invention sets the simulation data of 6 radar pulse signal parameters, and a certain Gaussian measurement error is added to the simulation of each signal parameter, and compared with the K-means clustering method. The basic idea of ​​the K-means algorithm is to divide the data into K different clusters in an iterative manner, so that the sum of the distances between each data point and the center of the cluster to which it belongs is minimized. The execution process of the K-means algorithm usually includes specifying the number of clusters K, selecting the initial cluster center, iteratively calculating the distance between sample points, updating the cluster center of each cluster, and outputting the final cluster. The K-means algorithm requires manual preset of the number of clusters, and is relatively sensitive to the cluster center, which affects the final signal sorting accuracy and stability.

[0079] The present invention performs signal sorting according to the above policy data, and detects the sorting accuracy of the corresponding pulse signal, sorts the aliased signals of N targets in the environment, and the total number of real signal pulses of the sorted i-th type signal is n i , the number of pulses correctly sorted is m i , then the sorting accuracy of this type of signal pulse is:

[0080]

[0081] When the ratio is greater than a preset threshold (such as 90%), it is determined that the target i is successfully sorted. Therefore, the overall cluster sorting accuracy is the ratio of the number of successfully sorted pulses to the total number of true pulses, and its calculation formula is as follows:

[0082]

[0083] like Figure 3 As shown, the sorting accuracy of different simulated radar signals can be obtained. By comparing the sorting results of each simulated signal, the signal sorting algorithm based on convex clustering proposed in the present invention can still guarantee the correct sorting result when the signal source parameters overlap.

[0084] The stability of the radar signal sorting algorithm based on convex clustering mentioned in the present invention is compared with the stability of the K-means algorithm. The experiment is repeated 50 times. Figure 4As shown in the figure, the experimental results show that when the signal input order is not fixed, the clustering stability of the K-means algorithm is poor, the volatility is large, and the average sorting accuracy is 0.9103. The radar signal sorting algorithm based on convex clustering mentioned in the present invention is not affected by the signal input order, the signal sorting accuracy is 0.9758, and the stability is strong.

[0085] The method of adding Gaussian white noise to the simulation data is used to construct simulation data with different signal-to-noise ratios, such as Figure 5 As shown, a curve of the change of sorting accuracy under different signal-to-noise ratios is obtained, and 50 experiments are carried out each time to take the average accuracy. It can be seen that with the decrease of the signal-to-noise ratio, the signal sorting performance decreases, but the sorting accuracy of the present invention under different signal-to-noise ratios is better than K-means. And when the signal-to-noise ratio is higher than 10dB, a signal sorting accuracy of more than 85% can be obtained, which has high noise robustness, indicating that the sorting effect of the present invention is not affected by the order of signal input, has higher stability, and under different signal-to-noise ratio conditions, it still has good noise robustness, which is very suitable for use in electronic reconnaissance tasks that require accurate identification of target radars.

[0086] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A radar signal sorting method based on convex clustering, characterized in that: The following steps are involved: S1. Obtain pulse signals from different radar signal sources detected by the receiver; S2. Standardize the parameters of pulse signals received from different radars; S3, constructing a convex optimization function corresponding to signal sorting; S4, using the alternating minimization method to optimize and solve the convex optimization objective function; S5, iterating the step S4 until the change of the convex objective function is less than a preset threshold, thereby obtaining a global optimal solution of the convex optimization function; S6. Compare the Euclidean distances between the vectors of the optimal solution. When the distance is less than a preset threshold, classify the pulse signals represented by similar vectors into the same cluster to obtain the final signal sorting result.

2. The radar signal sorting method based on convex clustering according to claim 1, characterized in that: In step S2, the detected pulse signal parameters are distributed in the same [0,1] interval and participate in the radar signal sorting task with the same dimension. The pulse signal parameters include carrier frequency, pulse width and arrival angle. The pulse signal parameter matrix is: Among them, cf i ,pw i ,doa i ,i=1,...,N are the frequency, pulse width and arrival direction of the i-th pulse respectively, and N is the total number of pulse signals.

3. A radar signal sorting method based on convex clustering as claimed in claim 2, characterized in that: In step S2, the mean and standard deviation of the k-th dimension data of the pulse signal parameter matrix are: Among them, 1≤k≤3 is the dimension of the pulse signal parameter matrix; the standardized value of the sample pulse vector is obtained: The extreme value normalization formula is used to make the normalized value of the pulse signal matrix distributed in the interval [0,1]. The formula is as follows: Among them, pdw′ kmin and pdw′ kmax pdw′ ik The minimum and maximum values ​​in .

4. The radar signal sorting method based on convex clustering as claimed in claim 3, characterized in that: In step S3, the convex optimization function includes a loss function and a regularization term, and the optimization objective function is: Among them, the first term is the loss function, the second term is the regularization term and generally takes the l1 norm, pdw″ i is the normalized pulse signal parameter, u i It is pdw i The cluster center, weight parameter ω ij Non-negative and ω ij =ω ji , γ is a non-negative tuning constant.

5. The radar signal sorting method based on convex clustering as claimed in claim 4, characterized in that: In step S4, the convex clustering problem is redefined as the following equivalent constraint problem: According to the constraint solving method, the augmented Lagrangian expression is as follows: Among them, l is the constructed centroid pair l = (l1, l2) and l1 <l2,ε={l=(l1,l2):ω l >0} is an edge set defined on non-zero weights, and a new variable is introduced to define the difference between the centroids.

6. The radar signal sorting method based on convex clustering according to claim 5, characterized in that: In step S4, the alternating minimization method is used to solve the minimum value of the convex optimization function, and the convex optimization objective function is optimized and solved. The mth iteration process is as follows: in, is z to the closed convex set C l The mapping of C l ={λ l :||λ l || + ≤γω l }, yes The dual norm of , υ is the iteration step size,

Images