Novel time-frequency network feature extraction method and device, storage medium and electronic equipment

Abstract

The embodiment of the invention provides a novel time-frequency network feature extraction method and device, a storage medium and electronic equipment, and relates to the technical field of time-frequency network feature extraction, and the method comprises the steps: carrying out the CWD transformation, filtering and regularization denoising of a received radiation source signal, and obtaining a two-dimensional CWD transformation distribution value; performing grid division processing and networked modeling on the two-dimensional CWD transformation distribution value to obtain a signal time-frequency network; basic features of the signal time-frequency network are extracted, and a feature set representing radiation source signals is established; and based on the signal time-frequency network, extracting community structure features of the time-frequency network. According to the technical scheme of the invention, better signal characterization capability can be obtained, and the identification accuracy of the radiation source target in a complex environment is improved.

Description

Technical Field

[0001] This application relates to the technical field of time-frequency network feature extraction. Specifically, it relates to a new time-frequency network feature extraction method, device, storage medium, and electronic device. Background Art

[0002] Perceiving the complex and ever-changing battlefield environment is a key link in modern information battlefields. Seizing information superiority is the prerequisite and foundation for gaining the initiative and even winning the war. In future information-based battlefields, the traditional seizure of air superiority depends on seizing information superiority; the foundation for seizing information superiority is seizing electromagnetic superiority. The complex electromagnetic environment is the stage of information-based warfare and the basic feature of information-based battlefields. Electromagnetic superiority has become the foundation for information superiority. With the rapid development of modern electronic technology and its extensive application in air raid operations, the power of electromagnetic attacks has increased unprecedentedly, the attack methods have multiplied, the attack platforms are diverse, and the attack speed has accelerated. It can be imagined that before the start of future air defense operations, both the enemy and us will conduct reconnaissance and anti-reconnaissance, deception and anti-deception; during the operation, interference and anti-interference, destruction and anti-destruction will surely constitute an extremely complex electromagnetic attack environment. Multi-means, all-round, and three-dimensional electromagnetic reconnaissance to perceive the dynamically changing battlefield electromagnetic environment is the key and basic guarantee for seizing "electromagnetic superiority". With the development of radar and communication technologies and the wide application of modern information technologies, the electromagnetic environment in modern information-based battlefields has become increasingly complex. A large number of various radar and communication radiation sources are widely used in the battlefield, electromagnetic signals are extremely dense, and the signal density is as high as millions per second; in order to meet the requirements of various detection and communication functions and anti-jamming needs, the waveform of the radiation source target signal is complex and ever-changing, and the emission law is more random; the existing electromagnetic signal perception methods and means are increasingly difficult to meet the technical requirements of accurate perception of electromagnetic radiation source targets in modern battlefields. There is an urgent need to develop new radiation source identification and perception technologies to cope with the major threats posed by complex electromagnetic targets in the battlefield.

[0003] Radiation source identification is an important function of electromagnetic countermeasure systems such as radar and communication. It is used to compare the characteristic parameters of the radiation source signals obtained from radar, communication, etc. reconnaissance with the technical performance of known radar, communication, and other radiation sources, so as to judge in real time the type of the target radiation source emitting this signal, and determine the use, carrier, threat level, and identification credibility of this radiation source. Among them, feature analysis is the prerequisite and key for radiation source identification. Whether it is classical radiation source identification methods or new radiation source identification methods such as deep learning, ultimately, the features of the signal must be extracted and processed first. However, in the face of a complex electromagnetic environment with highly dense radiation source signals, complex modulation types, and rapidly changing parameters, traditional radiation source signal identification based on conventional parameters cannot meet the requirements of real-time and accuracy, resulting in a low accuracy rate of radiation source target identification. Although feature extraction methods based on time domain, frequency domain, time-frequency domain, and other mathematical transformation domains have a high research popularity and have achieved good results, establishing an efficient and accurate characterization analysis model for radiation source signal sequences is still a very difficult hot issue because it directly affects the radiation source identification performance in complex environments. Summary of the Invention

[0004] Embodiments of the present application provide a novel time-frequency network feature extraction method, device, storage medium, and electronic device to solve the practical needs that traditional characteristic parameters are difficult to meet for high-precision radiation source identification in complex electromagnetic environments.

[0005] Other features and advantages of the present application will become apparent through the following detailed description, or be learned in part through the practice of the present application.

[0006] According to the first aspect of the embodiments of the present application, a novel time-frequency network feature extraction method is provided, including:

[0007] Perform CWD transform, filtering, and regularization denoising processing on the received radiation source signal to obtain two-dimensional CWD transform distribution values;

[0008] Perform grid division processing and networked modeling on the two-dimensional CWD transform distribution values to obtain a signal time-frequency network;

[0009] Extract the basic features of the signal time-frequency network and establish a feature set representing the radiation source signal;

[0010] Based on the signal time-frequency network, extract the time-frequency network community structure features.

[0011] In some embodiments of the present application, based on the foregoing solution, performing CWD transform, filtering, and regularization denoising processing on the received radiation source signal to obtain two-dimensional CWD transform distribution values includes:

[0012] Perform the CWD transformation processing on the radiation source signal to obtain two-dimensional time-frequency distribution values;

[0013] Use the time-domain smoothing window function and the frequency-domain smoothing window function to filter the two-dimensional time-frequency distribution values;

[0014] Normalize the two-dimensional time-frequency distribution values after filtering to obtain regularization values;

[0015] Perform thresholding processing on the regularization values to obtain the two-dimensional CWD transformation distribution values.

[0016] In some embodiments of the present application, based on the foregoing solution, performing grid division processing and networked modeling on the two-dimensional CWD transformation distribution values to obtain a signal time-frequency network, including:

[0017] Grid-divide the two-dimensional CWD transformation distribution values into multiple time-frequency blocks with uniform sizes;

[0018] Select time-frequency blocks with characteristic key information amounts from multiple time-frequency blocks, and denote them as characteristic time-frequency blocks;

[0019] Calculate the Pearson correlation coefficients between all characteristic time-frequency blocks to obtain the weight matrix of the connection edges between network nodes;

[0020] Set an edge connection threshold, judge the edge connection situation based on the weight matrix, and construct a signal time-frequency network.

[0021] In some embodiments of the present application, based on the foregoing solution, the calculating the Pearson correlation coefficients between all characteristic time-frequency blocks to obtain the weight matrix of the connection edges between network nodes includes:

[0022] Calculate the energy distribution status of each characteristic time-frequency block;

[0023] Calculate the Pearson correlation coefficients between characteristic time-frequency blocks according to the energy distribution status to obtain the weight matrix.

[0024] In some embodiments of the present application, based on the foregoing solution, the setting an edge connection threshold, judging the edge connection situation based on the weight matrix, and constructing a signal time-frequency network includes:

[0025] Map each characteristic time-frequency block to a network structure node to form a node set;

[0026] Set an edge connection threshold, determine a network connection matrix based on the edge connection threshold and the weight matrix, and construct a signal time-frequency network;

[0027] Set the minimum number of connection edges, judge whether the constructed signal time-frequency network is qualified based on the minimum number of connection variables, and if not, reconstruct it.

[0028] In some embodiments of the present application, based on the foregoing solution, extracting the basic features of the signal time-frequency network and establishing a feature set representing the radiation source signal includes:

[0029] By fully connecting the signal time-frequency network with weights and no direction, calculating the weighted degree of the nodes of the signal time-frequency network, and calculating the average degree feature based on the weighted degree;

[0030] By traversing all the nodes of the signal time-frequency network through the breadth-first search algorithm, obtaining the distances between the nodes, summing up the distances between the nodes and taking the average to obtain the average path length feature, and obtaining the network diameter feature based on the maximum distance between the network nodes;

[0031] Based on the total number of edges and the total number of nodes in the signal time-frequency network, obtaining the network density feature and the clustering coefficient feature;

[0032] Setting a cut set for deleting important nodes of the signal time-frequency network, and calculating the coreness feature by using the cut set;

[0033] Establishing the feature set according to the average degree feature, the average path length feature, the network diameter feature, the network density feature, the clustering coefficient feature and the coreness feature.

[0034] In some embodiments of the present application, based on the foregoing solution, extracting the time-frequency network community structure features based on the signal time-frequency network includes:

[0035] Using the Louvain community detection algorithm to perform modular division on the signal time-frequency network and calculating the modularity to obtain the time-frequency network community structure features.

[0036] According to the second aspect of the embodiments of the present application, a novel time-frequency network feature extraction device is provided, including:

[0037] A first generation unit, configured to perform CWD transformation, filtering, and regularization denoising processing on the received radiation source signal to obtain two-dimensional CWD transformation distribution values;

[0038] A second generation unit, configured to perform grid division processing and networked modeling on the two-dimensional CWD transformation distribution values to obtain a signal time-frequency network;

[0039] A first extraction unit, configured to extract the basic features of the signal time-frequency network and establish a feature set representing the radiation source signal;

[0040] A second extraction unit, configured to extract the time-frequency network community structure features based on the signal time-frequency network.

[0041] According to a third aspect of the embodiments of the present application, a computer-readable storage medium is provided. Computer instructions are stored in the storage medium. When the computer instructions run on a computer, the computer is caused to execute the method described in the first aspect.

[0042] According to a fourth aspect of the embodiments of the present application, an electronic device is provided, including: a memory and a processor;

[0043] The memory is used to store computer instructions;

[0044] The processor is used to call the computer instructions stored in the memory, so that the electronic device executes the method described in the first aspect.

[0045] The technical solution of the present application analyzes the internal structure and timing characteristics of the radiation source signal, performs network modeling on the internal structure or timing characteristics of the signal in the time-frequency domain of the radiation source signal, and extracts corresponding time-frequency network features to obtain better signal characterization ability and improve the recognition accuracy of radiation source targets in complex environments.

[0046] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and cannot limit the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The accompanying drawings herein are incorporated into the specification and form a part of the specification, showing embodiments consistent with the present application, and are used together with the specification to explain the principles of the present application. Obviously, the accompanying drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts. In the drawings:

[0048] Figure 1 A flowchart showing a novel time-frequency network feature extraction method according to an embodiment of the present application is shown;

[0049] Figure 2 A flowchart showing the CWD transform and denoising process according to an embodiment of the present application is shown;

[0050] Figure 3 A flowchart showing the signal time-frequency block division and signal network construction according to an embodiment of the present application is shown;

[0051] Figure 4 A schematic diagram of a CW time-frequency network according to an embodiment of the present application is shown;

[0052] Figure 5 A schematic diagram of an LFM time-frequency network according to an embodiment of the present application is shown;

[0053] Figure 6 Shows a schematic diagram of the NLFM time-frequency network according to an embodiment of the present application;

[0054] Figure 7 Shows a schematic diagram of the FSK time-frequency network according to an embodiment of the present application;

[0055] Figure 8 Shows a schematic diagram of the BPSK time-frequency network according to an embodiment of the present application;

[0056] Figure 9 Shows a schematic diagram of the QPSK time-frequency network according to an embodiment of the present application;

[0057] Figure 10 Shows the time-frequency network characteristic distribution diagram of six types of classical radiation source signals with SNR=-5dB according to an embodiment of the present application;

[0058] Figure 11 Shows the fractal dimension characteristic distribution diagram of six types of classical radiation source signals with SNR=-5dB according to an embodiment of the present application;

[0059] Figure 12 Shows the information entropy cascade characteristic distribution diagram of six types of classical radiation source signals with SNR=-5dB according to an embodiment of the present application;

[0060] Figure 13 Shows the schematic diagram of the classification and recognition accuracy of radiation source signals based on three different characteristics under different signal-to-noise ratios according to an embodiment of the present application;

[0061] Figure 14 Shows the block diagram of a novel time-frequency network feature extraction device according to an embodiment of the present application;

[0062] Figure 15 Shows the block diagram of an electronic device according to an embodiment of the present application;

[0063] Figure 16 Shows the schematic diagram of the structure of a computer system of an electronic device suitable for implementing the embodiments of the present application. Detailed implementation manners

[0064] Now, the example embodiments will be described more fully with reference to the accompanying drawings. However, the example embodiments can be implemented in various forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that this application will be more complete and comprehensive, and will fully convey the concept of the example embodiments to those skilled in the art.

[0065] In addition, the described features, structures, or characteristics may be combined in one or more embodiments in any suitable manner. In the following description, numerous specific details are provided to give a thorough understanding of the embodiments of the present application. However, those skilled in the art will realize that the technical solutions of the present application may be practiced without one or more of the specific details, or other methods, components, devices, steps, etc. may be adopted. In other cases, well-known methods, devices, implementations, or operations are not shown or described in detail to avoid obscuring aspects of the present application.

[0066] The block diagrams shown in the drawings are only functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities may be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor devices and / or microcontroller devices.

[0067] The flowcharts shown in the drawings are only illustrative and not necessarily include all the content and operations / steps, nor are they necessarily executed in the described order. For example, some operations / steps can be decomposed, and some operations / steps can be combined or partially combined, so the actual execution order may change according to the actual situation.

[0068] It should be noted that the terms "first", "second", etc. in the specification, claims, and above-mentioned drawings of the present application are used to distinguish similar objects and do not necessarily describe a specific order or sequence. It should be understood that the objects so used can be interchanged under appropriate circumstances so that the embodiments of the present application described here can be implemented in an order other than those illustrated or described.

[0069] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0070] The following will describe in detail some embodiments of the present application in conjunction with the drawings. Without conflict, the following embodiments and the features in the embodiments can be combined with each other.

[0071] See Figure 1 , which shows a schematic flowchart of a novel time-frequency network feature extraction method according to an embodiment of the present application.

[0072] As Figure 1As shown, a new method for extracting time-frequency network features is presented, specifically including steps S100 to S400.

[0073] Reference Figure 1 , in step S100, the received radiation source signal is subjected to CWD transformation, filtering, and regularization denoising to obtain two-dimensional CWD transformation distribution values.

[0074] In some feasible embodiments, based on the foregoing solution, the steps of subjecting the received radiation source signal to CWD transformation, filtering, and regularization denoising to obtain two-dimensional CWD transformation distribution values include:

[0075] Performing CWD transformation on the radiation source signal to obtain two-dimensional time-frequency distribution values;

[0076] Using a time-domain smoothing window function and a frequency-domain smoothing window function to filter the two-dimensional time-frequency distribution values;

[0077] Normalizing the filtered two-dimensional time-frequency distribution values to obtain regularization values;

[0078] Performing thresholding on the regularization values to obtain the two-dimensional CWD transformation distribution values.

[0079] It should be noted that the CWD transformation refers to the Choi-Williams distribution transformation.

[0080] Exemplarily, as Figure 2 shown, the specific process of CWD transformation and denoising is as follows:

[0081] Step S101: Performing CWD transformation on the radiation source signal S:

[0082]

[0083] where A s (u,τ) is the ambiguity function of the radiation source signal S, and Φ(u,τ) is the kernel function, whose value is the exponential kernel function exp(-α(uτ) 2 ), and the attenuation coefficient σ (σ = 1 / α) is set to 0.1. Thus, the two-dimensional time-frequency distribution value D(t,f) of the radiation source signal S after CWD transformation is obtained.

[0084] Step S102: Time-frequency domain smoothing filtering.

[0085] Select the time-domain smoothing window G = hamming(N / 10 + 1) and the frequency-domain smoothing window function H = hamming(N / 4 + 1) for filtering to obtain D filter (t,f).

[0086] Step S103. Normalize the filtered D filter (t, f), using the z-score method, to obtain the mean value of the two-dimensional distribution and the standard deviation s, and obtain the regularization value from .

[0087] Step S104. Threshold the regularized CWD distribution D normalized (t, f), select the threshold value as 0.15, further eliminate the influence of noise, and obtain the processed two-dimensional CWD transform distribution value D p .

[0088] Continue to refer to Figure 1 , step S200, perform grid division processing and networked modeling on the two-dimensional CWD transform distribution value to obtain a signal time-frequency network

[0089] In some feasible embodiments, based on the foregoing solution, the performing grid division processing and networked modeling on the two-dimensional CWD transform distribution value to obtain a signal time-frequency network includes:

[0090] Grid-divide the two-dimensional CWD transform distribution value into multiple time-frequency blocks with uniform sizes

[0091] Select the time-frequency blocks with characteristic key information amounts from the multiple time-frequency blocks, and denote them as characteristic time-frequency blocks

[0092] Calculate the Pearson correlation coefficients between all characteristic time-frequency blocks to obtain the weight matrix of the connection edges between network nodes

[0093] Set an edge connection threshold, judge the edge connection situation based on the weight matrix, and construct a signal time-frequency network

[0094] In some feasible embodiments, based on the foregoing solution, the calculating the Pearson correlation coefficients between all characteristic time-frequency blocks to obtain the weight matrix of the connection edges between network nodes includes:

[0095] Calculate the energy distribution status of each characteristic time-frequency block

[0096] Calculate the Pearson correlation coefficients between characteristic time-frequency blocks according to the energy distribution status to obtain the weight matrix

[0097] In some feasible embodiments, based on the foregoing solution, the setting an edge connection threshold, judging the edge connection situation based on the weight matrix, and constructing a signal time-frequency network includes:

[0098] Map each characteristic time-frequency block to a network structure node to form a node set

[0099] Set the edge connection threshold, determine the network connection matrix based on the edge connection threshold and the weight matrix, and construct a signal time-frequency network;

[0100] Set the minimum number of connected edges, and judge whether the constructed signal time-frequency network is qualified based on the minimum number of connected variables. If it is unqualified, reconstruct it.

[0101] Exemplarily, refer to Figure 3 , for the two-dimensional CWD transform distribution value D obtained after the CWD time-frequency transform p The specific process of performing grid division, dividing it into time-frequency blocks of uniform size, and performing network modeling on the time-frequency blocks to obtain the signal time-frequency network G(V,E) is as follows:

[0102] Step S201, grid the two-dimensional CWD transform distribution value D p Into multiple uniform time-frequency blocks, and nodes corresponding to each time-frequency block will be constructed in the subsequent steps. In this example, grid division is performed in two dimensions of time and frequency, and the number of divisions N t And N f are determined. Adopting the uniform division method, calculate the time domain interval And the frequency domain interval Thus, the time-frequency block D ij (i = 1, 2,..., N t ) is divided.

[0103] Step S202, screen the time-frequency blocks with characteristic key information. Traverse each time-frequency block D ij , calculate the sparsity coefficient of the time-frequency block Where n0 is the number of zero elements in the time-frequency block D ij , N = N t ·N f is the total number of time-frequency block units. Given the sparsity threshold, select n time-frequency blocks {D1, D2,..., D n}.

[0104] Step S203, calculate the correlation coefficient between the screened time-frequency blocks, and obtain the weight matrix W of the connection edges between network nodes.

[0105] First, calculate the energy distribution status P i of each time-frequency block D i = {p i1 , p i2 ,..., p iN}, where the energy distribution probability p ij is:

[0106]

[0107] Among them (l0,k0) is the index value of the time-frequency block D i (i.e., a specific distribution value within the time-frequency block). Then, calculate the Pearson correlation coefficient W(i,i′) (corresponding to the weight of the connection edge between network nodes) between the time-frequency blocks D i and D i as follows:

[0108]

[0109] Step S204, model the time-frequency block as a network G(V,E).

[0110] Map each time-frequency block D i to a network structure node v i , and form a node set V = {v1,v2,…,v n}. Determine the network adjacency matrix A through the weight matrix W. The edge connection rule between any two nodes v i and v j is expressed as follows:

[0111]

[0112] where th edge is the initial edge connection threshold, and its value reflects the network density change rate, defined as:

[0113]

[0114] Step S205, extract the largest connected subgraph G.

[0115] The network G0 constructed with the adjacency matrix obtained in step d may not be fully connected, so extract the largest connected component of the network G0 to form the subgraph G.

[0116] Step S206, network inspection and exception handling.

[0117] When the threshold is set too large or affected by noise, it may lead to the failure of time-frequency network modeling. The edge set of the network is Set the minimum number of connected edges k, check whether the number of edges in the network edge set E is greater than k. When the number of edges n edge in E is less than k (network construction fails), dynamically reduce the edge connection threshold th edge , and re-perform step S204 to reconstruct the network again. Finally, obtain the constructed signal time-frequency network G(V,E).

[0118] Continue to refer to Figure 1 , step S300, extract the basic features of the signal time-frequency network, and establish a feature set representing the radiation source signal.

[0119] In some feasible embodiments, based on the foregoing solution, extracting the basic features of the signal time-frequency network and establishing a feature set representing the radiation source signal includes:

[0120] By fully connecting the signal time-frequency network with weights and no directions, calculating the weighted degree of the nodes of the signal time-frequency network, and obtaining the average degree feature according to the weighted degree;

[0121] By traversing all nodes of the signal time-frequency network through the breadth-first search algorithm, obtaining the distances between nodes, summing up the distances between nodes and taking the average to obtain the average path length feature, and obtaining the network diameter feature according to the maximum distance between network nodes;

[0122] By the total number of edges and the total number of nodes in the signal time-frequency network, obtaining the network density feature and the clustering coefficient feature;

[0123] Setting a cut set for deleting important nodes of the signal time-frequency network, and calculating the coreness feature by using the cut set;

[0124] Establishing the feature set according to the average degree feature, the average path length feature, the network diameter feature, the network density feature, the clustering coefficient feature and the coreness feature.

[0125] Exemplarily, the process of obtaining the average degree feature is as follows:

[0126] By the weighted undirected complete graph G(V, E), calculating the weighted degree of node u in the node set V:

[0127]

[0128] where V u is the set of adjacent nodes of node u, obtaining the weighted degree set D = {d1, d2,..., d n}, so as to obtain the average degree of graph G

[0129]

[0130] Exemplarily, the process of obtaining the average path length feature and the network diameter feature is as follows:

[0131] By traversing all nodes of network G(V, E) through the breadth-first search algorithm, obtaining the distance d ij between nodes, summing it up and taking the average to obtain the average path length L, that is:

[0132]

[0133] The network diameter D g is the maximum distance between network nodes, that is:

[0134] D g = max ij d ij 。

[0135] Exemplarily, the acquisition processes of the network density feature and the clustering coefficient feature are as follows:

[0136] Based on the total number of edges E and the total number of nodes N in the network, the network density is obtained:

[0137]

[0138] Let E u be the total number of connected edges between node u and its adjacent nodes, and N u be the number of adjacent nodes of node u, and calculate the clustering coefficient of node u:

[0139]

[0140] The clustering coefficient C of network G is the average value of the clustering coefficients of all nodes, and we get:

[0141]

[0142] Exemplarily, the acquisition process of the core degree feature is as follows:

[0143] Given that C(G) represents the cut set for deleting important nodes of graph G, then the core degree h is:

[0144]

[0145] where ω(G) represents the number of connected subgraphs (branches) of graph G, and |S| represents the number of nodes in node set S.

[0146] Continue to refer to Figure 1 , step S400, based on the signal time-frequency network, extract the time-frequency network community structure features.

[0147] In some feasible embodiments, based on the foregoing solution, the extracting the time-frequency network community structure features based on the signal time-frequency network includes:

[0148] Use the Louvain community detection algorithm to perform modular partitioning on the signal time-frequency network and calculate the modularity to obtain the time-frequency network community structure features.

[0149] Exemplarily, the extraction process of the time-frequency network community structure features is specifically as follows:

[0150] Step S401, perform modular partitioning on the signal time-frequency network G(V,E).

[0151] Use the Louvain community detection algorithm, and the specific steps are as follows:

[0152] a. Initialization of community division. Each node v in the initial node set V is regarded as a community. Initially, there are n communities: X1, X2, …, X n .

[0153] b. Locally move nodes. For each node v, try to move it into the community where its neighbor node is located. After the move, calculate the modularity gain ΔQ of the community. Select the community that can maximize the modularity ΔQ and assign the node to this community.

[0154] c. Repeat iteration. Repeat step b until no node moves to further increase ΔQ, so as to obtain the community division.

[0155] Step S402, Modularity calculation.

[0156] The community division obtained from step S401 where X j is the j-th module, and the modularity Q is calculated as:

[0157]

[0158] where E is the total number of edges, c i represents the community to which node i belongs, d i is node i, δ(c i , c j ) = 1 if and only if node i and node j belong to the same community, otherwise δ(c i , c j ) = 0.

[0159] Next, a simulation experiment on the performance comparison between time-frequency network features and other features is provided.

[0160] Extract the time-frequency network features obtained by using the technical solution of this application, as well as the existing typical fractal dimension features and information entropy cascade features, for identifying six types of typical radiation source signals: conventional signal CW, linear frequency modulation signal LFM, non-linear frequency modulation signal NLFM, frequency shift keying signal FSK, binary phase shift keying signal BPSK, and quadrature phase shift keying signal QPSK. Compare the recognition accuracies under different features to reflect the effectiveness of this feature and the performance differences between features. The specific implementation steps are as follows:

[0161] Step1: Simulate and generate six types of typical radiation source signals. In the signal-to-noise ratio range of -10 - 10 dB, generate 250 signals of each type every 1 dB, for a total of 1500 signals. Among them, the number of samples for training the Cost-Support Vector Machine (C-SVM) classifier is 200 for each type, for a total of 1200, and the number of test samples is 50 for each type, for a total of 300.

[0162] Step 2: Extract the time-frequency network features of the signal using the technical solution of this application (select some features: the average degree of the time-frequency network, the clustering coefficient, and the core number), and use the feature dimension feature extraction method and the information entropy cascade feature extraction method to obtain the time-frequency network features of the signal Fractal dimension [D b , D I and information entropy cascade [En t , En f , En tf three types of features, which constitute a feature vector for subsequent identification of radiation source signals. During the process of extracting time-frequency network features, it is necessary to construct the network G(V, E) of the signal. The time-frequency networks of the above 6 types of typical signals (signal-to-noise ratio SNR is 0 dB) are as follows Figures 4 to 9 shown

[0163] Step 3: Input the feature vectors corresponding to the training set in Step 2 into the C-SVM classifier for training. Input the feature vectors corresponding to the test set into the trained classifier for prediction, and compare with the actual labels to calculate the recognition accuracy of the radiation source signals corresponding to different features

[0164] Step 4: Compare the recognition accuracies calculated in Step 3 under the three types of feature parameters, and analyze the differences in performance

[0165] Figures 4 to 9 The networks G(V, E) corresponding to different types of signals in have obvious differences, reflecting different types of signals. Based on the time-frequency network, further extract the basic features and community structure features of the signal time-frequency network

[0166] In the experimental verification, when the signal-to-noise ratio SNR = -5 dB, the distributions of the time-frequency network features, fractal dimensions, and information entropy cascade features of the 6 types of classical radiation source signals are respectively as Figure 10 , Figure 11 and Figure 12 shown

[0167] It can be seen from the above experimental results that the time-frequency network features still have good intra-class aggregation and inter-class separability at low signal-to-noise ratios, while the other two types of features perform poorly. At the same time, plot the classification recognition accuracies of the radiation source signals of the three types of features (time-frequency network, fractal dimension, and information entropy cascade feature) within the entire signal-to-noise ratio range as Figure 13 shown

[0168] From Figure 13It can be seen that the recognition ability of time-frequency network features is optimal, and the classification recognition accuracy is generally higher than that of the other two types of features. As the signal-to-noise ratio gradually decreases, the recognition accuracy of each type of feature decreases. Generally, the recognition accuracy corresponding to the time-frequency network features performs better in the entire signal-to-noise ratio change range. In the range where the signal-to-noise ratio is greater than -5 dB, the recognition accuracy of the radiation source signals based on the time-frequency network features is close to or above 95%. At the same time, the recognition accuracy of various radiation source signals of the time-frequency network features in the signal-to-noise ratio change range of -10 - 10 dB with an interval of 5 dB is listed in Table 1.

[0169] Table 1 Recognition accuracy of 6 types of radiation source signals under time-frequency network features (%)

[0170] Recognition accuracy CW LFM NLFM FSK BPSK QPSK Average accuracy -10dB 71 53 35 37 60 38 46.3 -5dB 100 88 82 98 100 97 94.6 0dB 100 98 97 100 99 100 99 5dB 100 100 100 100 100 100 100 10dB 100 100 100 100 100 100 100 Average accuracy 94.2 88 83 87 92 87 \

[0171] In summary, the technical solution of this application starts from the transform domain, uses the CWD transform to divide the time-frequency blocks, constructs a signal time-frequency network diagram, and extracts the time-frequency network features as the features of the radiation signal on this basis, which is used in the field of radiation source perception to meet the requirements of radiation source recognition in complex environments and improve the recognition accuracy of radiation source signals. The experimental results show that in the entire signal-to-noise ratio change range (-5 - 10 dB), the recognition accuracy of the time-frequency network features proposed by the technical solution of this application for classifying 6 types of classical radiation source signals is above 94%, which has obvious advantages compared with traditional features.

[0172] The improvement of the real-time performance and accuracy of radiation source signal recognition has important significance both in military and civilian aspects. In the military aspect, it is mainly aimed at the field of radar radiation source recognition, which helps to improve the electronic reconnaissance ability, obtain the advantage of electronic countermeasure perception, and provide the key perception ability for seizing the "electromagnetic dominance" on the battlefield; in the civilian aspect, it helps to improve and perfect radio frequency identification, network node mutual recognition, information forensics and communication security technologies. The technical solution of this application has broad application prospects in both military and civilian fields.

[0173] The following introduces the device embodiments of this application, which can be used to execute a new time-frequency network feature extraction method in the above embodiments of this application. For the details not disclosed in the device embodiments of this application, please refer to the method embodiments of this application above.

[0174] Refer to Figure 14 As shown, a new time-frequency network feature extraction device 1400 according to an embodiment of this application includes:

[0175] A first generation unit 1401, configured to perform CWD transform, filtering, and regularization denoising processing on the received radiation source signal to obtain two-dimensional CWD transform distribution values;

[0176] A second generation unit 1402, configured to perform grid division processing and networked modeling on the two-dimensional CWD transform distribution values to obtain a signal time-frequency network;

[0177] A first extraction unit 1403, configured to extract basic features of the signal time-frequency network and establish a feature set characterizing the radiation source signal;

[0178] A second extraction unit 1404, configured to extract time-frequency network community structure features based on the signal time-frequency network.

[0179] As Figure 15 shown, an embodiment of the present application further provides an electronic device 1500, including a memory 1510, a processor 1520, and a computer program 1511 stored in the memory 1510 and executable on the processor. When the processor 1520 executes the computer program 1511, the steps of the above-mentioned novel time-frequency network feature extraction method are implemented.

[0180] Since the electronic device introduced in this embodiment is the device used to implement a novel time-frequency network feature extraction device in the embodiments of the present application, based on the method introduced in the embodiments of the present application, those skilled in the art can understand the specific implementation manners and various variations of the electronic device in this embodiment. Therefore, the specific implementation of how this electronic device implements the method in the embodiments of the present application will not be described in detail here. As long as the device used by those skilled in the art to implement the method in the embodiments of the present application belongs to the scope protected by the present application.

[0181] In the specific implementation process, when the computer program 1511 is executed by the processor, it can implement any implementation manner in the corresponding embodiment of the first aspect.

[0182] Figure 16 FIG. shows a schematic structural diagram of a computer system of an electronic device suitable for implementing the embodiments of the present application.

[0183] It should be noted that Figure 16 the computer system 1600 of the electronic device shown is only an example and should not bring any limitation to the functions and usage scopes of the embodiments of the present application.

[0184] As Figure 16As shown, computer system 1600 includes a Central Processing Unit (CPU) 1601, which can perform various appropriate actions and processes according to programs stored in a Read-Only Memory (ROM) 1602 or programs loaded from a storage section 1608 into a Random Access Memory (RAM) 1603, such as executing the methods described in the above embodiments. In the RAM 1603, various programs and data required for system operations are also stored. The CPU 1601, ROM 1602, and RAM 1603 are connected to each other via a bus 1604. An Input / Output (I / O) interface 1605 is also connected to the bus 1604.

[0185] The following components are connected to the I / O interface 1605: an input section 1606 including a keyboard, a mouse, etc.; an output section 1607 including, for example, a Cathode Ray Tube (CRT), a Liquid Crystal Display (LCD), etc., and a speaker, etc.; a storage section 1608 including a hard disk, etc.; and a communication section 1609 including a network interface card such as a LAN (Local Area Network) card, a modem, etc. The communication section 1609 performs communication processing via a network such as the Internet. A drive 1610 is also connected to the I / O interface 1605 as needed. A removable medium 1611, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 1610 as needed so that a computer program read from it can be installed into the storage section 1608 as needed.

[0186] In particular, according to an embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes program codes for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 1609, and / or installed from the removable medium 1611. When the computer program is executed by a Central Processing Unit (CPU) 1601, various functions defined in the system of the present application are executed.

[0187] It should be noted that the computer-readable medium shown in the embodiments of the present application can be a computer-readable signal medium, a computer-readable storage medium, or any combination of the two. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples of the computer-readable storage medium can include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a flash memory, an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, the computer-readable storage medium can be any tangible medium that contains or stores a program, and this program can be used by or in combination with an instruction execution system, apparatus, or device. In the present application, the computer-readable signal medium can include a data signal propagated in a baseband or as part of a carrier wave, which carries computer-readable program code. Such a propagated data signal can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The computer-readable signal medium can also be any computer-readable medium other than the computer-readable storage medium, and this computer-readable medium can send, propagate, or transmit a program for use by or in combination with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any appropriate medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0188] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of the present application. Among them, each block in the flowchart or block diagram can represent a module, a program segment, or a part of the code, and the above-mentioned module, program segment, or part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than marked in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram or flowchart, and the combination of blocks in the block diagram or flowchart, can be implemented by a dedicated hardware-based system for performing the specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0189] The units involved in the embodiments described in this application can be implemented in software or in hardware, and the described units can also be provided in a processor. Among them, the names of these units do not constitute a limitation to the unit itself in certain cases.

[0190] On the other hand, this application also provides a computer program product or a computer program. The computer program product or the computer program includes computer instructions, and the computer instructions are stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes a novel time-frequency network feature extraction method described in the above embodiments.

[0191] On the other hand, this application also provides a computer-readable medium, which may be included in the electronic device described in the above embodiments; or may exist alone without being assembled into the electronic device. The above computer-readable medium carries one or more programs, and when the one or more programs are executed by an electronic device, the electronic device implements a novel time-frequency network feature extraction method described in the above embodiments.

[0192] It should be noted that although several modules or units of a device for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of this application, the features and functions of the two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.

[0193] Through the description of the above embodiments, those skilled in the art can easily understand that the example embodiments described here can be implemented by software or by the means of software in combination with necessary hardware. Therefore, the technical solutions according to the embodiments of this application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, including several instructions to enable a computing device (such as a personal computer, a server, a touch terminal, or a network device, etc.) to execute the method according to the embodiments of this application.

[0194] Other embodiments of the present application will be readily envisioned by those skilled in the art upon consideration of the specification and practice of the embodiments disclosed herein. The present application is intended to cover any variations, uses, or adaptations of the present application, which follow the general principles of the present application and include known common general knowledge or conventional technical means in the technical field not disclosed in the present application. It should be understood that the present application is not limited to the precise structures described above and shown in the drawings, and various modifications and changes can be made without departing from its scope. The scope of the present application is only limited by the appended claims.

Claims

1. A novel time-frequency network feature extraction method, characterized in that Including: Performing CWD transformation, filtering, and regularization denoising on the received radiation source signal to obtain two-dimensional CWD transformation distribution values; Performing grid division processing and networked modeling on the two-dimensional CWD transformation distribution values to obtain a signal time-frequency network; Extracting the basic features of the signal time-frequency network and establishing a feature set representing the radiation source signal; Based on the signal time-frequency network, extracting the time-frequency network community structure features.

2. The method according to claim 1, wherein The performing CWD transformation, filtering, and regularization denoising on the received radiation source signal to obtain two-dimensional CWD transformation distribution values includes: Performing CWD transformation processing on the radiation source signal to obtain two-dimensional time-frequency distribution values; Filtering the two-dimensional time-frequency distribution values using a time-domain smoothing window function and a frequency-domain smoothing window function; Performing normalization processing on the filtered two-dimensional time-frequency distribution values to obtain regularization values; Performing thresholding processing on the regularization values to obtain the two-dimensional CWD transformation distribution values.

3. The method according to claim 1, wherein The performing grid division processing and networked modeling on the two-dimensional CWD transformation distribution values to obtain a signal time-frequency network includes: Grid-dividing the two-dimensional CWD transformation distribution values into multiple time-frequency blocks with uniform sizes; Selecting time-frequency blocks with characteristic key information amounts from multiple time-frequency blocks, denoted as characteristic time-frequency blocks; Calculating the Pearson correlation coefficients between all characteristic time-frequency blocks to obtain a weight matrix for the connection edges between network nodes; Setting an edge connection threshold, judging the edge connection situation based on the weight matrix, and constructing a signal time-frequency network.

4. The method according to claim 3, wherein The calculating the Pearson correlation coefficients between all characteristic time-frequency blocks to obtain a weight matrix for the connection edges between network nodes includes: Calculating the energy distribution status of each characteristic time-frequency block; Calculating the Pearson correlation coefficients between characteristic time-frequency blocks according to the energy distribution status to obtain the weight matrix.

5. The method according to claim 3, characterized in that, The setting an edge connection threshold, judging the edge connection situation based on the weight matrix, and constructing a signal time-frequency network includes: Mapping each characteristic time-frequency block to a network structure node to form a node set; Setting an edge connection threshold, determining a network connection matrix based on the edge connection threshold and the weight matrix, and constructing a signal time-frequency network; Setting a minimum number of connection edges, judging whether the constructed signal time-frequency network is qualified based on the minimum number of connection variables, and if not, reconstructing it.

6. The method according to claim 1, characterized in that, The extracting the basic features of the signal time-frequency network and establishing a feature set representing the radiation source signal includes: By fully connecting the signal time-frequency network with weights and undirected edges, calculating the weighted degrees of the nodes of the signal time-frequency network, and calculating the average degree feature according to the weighted degrees; Traversing all nodes of the signal time-frequency network through a breadth-first search algorithm, obtaining the distances between nodes, summing up the distances between nodes and taking the average to obtain the average path length feature, and obtaining the network diameter feature according to the maximum distance between network nodes; Based on the total number of edges and total number of nodes in the signal time-frequency network, obtaining the network density feature and the clustering coefficient feature; Setting a cut set for deleting important nodes of the signal time-frequency network and calculating the coreness feature using the cut set; The feature set is established according to the average degree feature, the average path length feature, the network diameter feature, the network density feature, the clustering coefficient feature, and the coreness feature.

7. The method according to claim 1, wherein Based on the signal time-frequency network, the time-frequency network community structure features are extracted, including: The Louvain community detection algorithm is used to perform modular partitioning on the signal time-frequency network, and the modularity is calculated to obtain the time-frequency network community structure features.

8. A novel time-frequency network feature extraction device, characterized in that Including: The first generation unit is configured to perform CWD transformation, filtering, and regularization denoising processing on the received radiation source signal to obtain two-dimensional CWD transformation distribution values; The second generation unit is configured to perform grid partitioning processing and networked modeling on the two-dimensional CWD transformation distribution values to obtain a signal time-frequency network; The first extraction unit is configured to extract the basic features of the signal time-frequency network and establish a feature set representing the radiation source signal; The second extraction unit is configured to extract the time-frequency network community structure features based on the signal time-frequency network.

9. A computer-readable storage medium, characterized in that, Computer instructions are stored in the storage medium, and when the computer instructions are run on a computer, the computer is caused to execute the method according to any one of claims 1-7.

10. An electronic device, characterized in that, Including: A memory and a processor; The memory is used to store computer instructions; The processor is configured to call the computer instructions stored in the memory, so that the electronic device executes the method according to any one of claims 1-7.

Images