Electromagnetic signal stream clustering method and device based on high-order features of neural network signals, medium and system

Abstract

The application discloses a kind of electromagnetic signal stream clustering methods, device, medium and system based on neural network signal high-order feature, belong to intelligent signal processing field, including steps: first with the high-order feature of the 1st signal clustering initialization, subsequently to the nth signal respectively signal high-order feature and the cosine similarity of existing clustering high-order feature.If the maximum cosine similarity is greater than threshold, then the signal is attributed to corresponding cluster. Otherwise, a new class is established, until all signals are traversed, to realize electromagnetic signal clustering based on neural network signal high-order feature. The application does not depend on the electromagnetic parameters of radiation source, avoids the problem of electromagnetic signal clustering error caused by the low dimension of traditional PDW parameters.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of intelligent signal processing, and more particularly, to an electromagnetic signal stream clustering method, device, medium and system based on neural network signal high-order features. BACKGROUND

[0002] Traditional electromagnetic signal clustering is mainly based on manually extracted PDW features, but due to the low dimensionality of PDW parameters, the parameters overlap seriously, resulting in poor adaptability. The academic community has proposed to extract medium-frequency high-order features of electromagnetic signals based on neural networks, but so far it has only been used for target identification, and there is still little research on signal clustering. SUMMARY

[0003] The present application aims to overcome the shortcomings of the prior art and provide an electromagnetic signal stream clustering method, device, medium and system based on neural network signal high-order features, which can avoid the problem of electromagnetic signal clustering errors caused by the low dimensionality of traditional PDW parameters.

[0004] The purpose of the present application is achieved by the following scheme:

[0005] An electromagnetic signal stream clustering method based on neural network signal high-order features, comprising the following steps:

[0006] Let the signal high-order feature extracted by the neural network for the nth signal be Feature n , where n = 1, 2, …, N, N represents the total number of signals, and the dimension of Feature n is K; for the mth cluster, its clustering high-order feature is represented as , where m = 1, 2, …, M, M represents the total number of signals, and the dimensions of Feature n and are K; take the 1st signal as the 1st cluster, and its clustering high-order feature is Complete initialization, and then perform the following steps:

[0007] S1: For the nth signal, let there be m cluster results, then calculate the cosine similarity between the nth signal and the mth cluster result in turn, which is:

[0008]

[0009] , where represents the absolute value of the dot product of the nth signal high-order feature and the mth cluster feature, and ||Feature n || represents the Euclidean norm of the nth signal high-order feature, Euclidean norm of the mth cluster high-order feature;

[0010] S2: find R n,m The maximum value is MaxR n,m , and the corresponding serial number is max_m; set the threshold value as Thres, if MaxR n,m ≥ Thres, the nth signal is classified into the max_mth cluster, and S3 is performed; if MaxR n,m < Thres, it is considered that the nth signal belongs to a new cluster, and S4 is performed;

[0011] S3: for the max_mth cluster, update the cluster high-order feature with the nth signal high-order feature,

[0012] That is Wherein α is the cluster high-order feature smoothing coefficient, the dot product of the nth signal high-order feature and the mth cluster feature is greater than or equal to 0, Otherwise

[0013] S4: a new cluster is added, that is, m = m + 1, and the cluster high-order feature is the nth signal high-order feature, that is

[0014] S5: from the second signal, S1-S4 are processed in this way, until all signals are processed, that is, the signal clustering processing based on the neural network signal high-order feature is completed.

[0015] Further, the value of dimension K includes 32.

[0016] Further, the value of threshold Thres includes 0.6.

[0017] Further, the value of the cluster high-order feature smoothing coefficient α includes 1.

[0018] An electromagnetic signal stream clustering device based on neural network signal high-order feature, comprising a processor and a memory, the memory stores a computer program, when the computer program is loaded by the processor, the method as claimed in any one of the above is executed.

[0019] A computer readable storage medium, the computer readable storage medium stores a computer program, when the computer program is loaded by the processor, the method as claimed in any one of the above is executed.

[0020] An electromagnetic signal stream clustering system based on neural network signal high-order feature, comprising the device as claimed above.

[0021] The beneficial effects of the present application include:

[0022] The application is not dependent on electromagnetic parameters of a radiation source, and avoids the problem of clustering errors of electromagnetic signals caused by low dimensionality of PDW parameters in the prior art. BRIEF DESCRIPTION OF DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description only constitute some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0024] Figure 1 The method step flowchart of the embodiment of the present application;

[0025] Figure 2 The neural network signal high-order feature extraction block diagram of the embodiment of the present application. DETAILED DESCRIPTION

[0026] All features disclosed in the embodiments of the present application, or all steps in the methods or processes impliedly disclosed in the present specification, can be combined and / or extended, replaced, and modified in any manner, except for the features and / or steps that are mutually exclusive.

[0027] In view of the problems in the background, the present application provides an electromagnetic signal stream clustering method based on neural network high-order features. First, the high-order features of the first signal are clustered for initialization, and then the cosine similarity of the high-order features of the nth signal and the existing clustered high-order features is calculated. If the maximum cosine similarity is greater than a threshold, the signal is assigned to the corresponding cluster. Otherwise, a new class is created, and the process is repeated until all signals are traversed, thereby realizing electromagnetic signal clustering based on neural network signal high-order features. Figure 1 and Figure 2 As shown in the specific implementation process of the present application is as follows:

[0028] Suppose for the nth (n = 1, 2,..., N) signal, the signal high-order features extracted by the neural network are represented as Feature n ; where N represents the total number of signals, and the dimension of Feature n is K. For the mth (m = 1, 2,..., M) cluster, the clustered high-order features are represented as where M represents the total number of signals, and the dimensions of Feature n and are both K.

[0029] Initialization: the first signal is taken as the first cluster, and the clustered high-order features are

[0030] Step 1: For the nth signal, assuming there are already m clustering results, calculate the cosine similarity between the nth signal and the mth clustering result in sequence, which is:

[0031]

[0032] where represents the absolute value of the dot product between the high-order features of the nth signal and the mth clustering feature, and ||Feature n || represents the Euclidean norm of the high-order features of the nth signal, represents the Euclidean norm of the high-order features of the mth clustering.

[0033] Step 2: Find the R n,m The maximum value is MaxR n,m , and its corresponding serial number is max_m; set the threshold value as Thres. If MaxR n,m ≥Thres, then classify the nth signal into the max_mth clustering and proceed according to Step 3. If MaxR n,m <Thres, it is considered that the nth signal belongs to a new clustering and proceed according to Step 4.

[0034] Step 3: For the max_mth clustering, update its high-order clustering features with the high-order features of the nth signal, that is where α is the smoothing coefficient of the high-order clustering features. When the dot product between the high-order features of the nth signal and the mth clustering feature is greater than or equal to 0, otherwise

[0035] Step 4: Add a new clustering, that is m = m + 1, and set the high-order clustering features of this clustering as the high-order features of the nth signal, that is

[0036] Step 5: Starting from the 2nd signal, process it according to Step 1 to Step 4 in sequence until all signals are processed, that is, complete the signal clustering processing based on the high-order features of the neural network signals.

[0037] In other embodiments of the present invention, assuming that for the nth (n = 1, 2,..., N) signal, the high-order features of the signal extracted by the neural network are represented as Feature n ; where N represents the total number of signals, and the dimension of Feature n is K-dimensional. For the mth (m = 1, 2,..., M) clustering, its high-order clustering features are represented as where M represents the total number of signals, Feature n and The dimensions of the two are K, and K=32 in this embodiment.

[0038] Initialization: the first signal is taken as the first cluster, and the cluster high-order feature is

[0039] Step 1: for the nth signal, assuming that there are m cluster results, the cosine similarity between the nth signal and the mth cluster result is calculated in turn, and the cosine similarity is

[0040]

[0041] wherein, The absolute value of the dot product of the high-order feature of the nth signal and the mth cluster feature is represented as ||Feature n The Euclidean norm of the high-order feature of the nth signal is represented as ||Feature The Euclidean norm of the mth cluster high-order feature is represented as ||Feature

[0042] Step 2: find the maximum value MaxR n,m of R n,m , and the corresponding serial number is max_m; a threshold value Thres is set, if MaxR n,m ≥ Thres, the nth signal is classified into the max_mth cluster, and the third step is performed. If MaxR n,m < Thres, it is considered that the nth signal belongs to a new cluster, and the fourth step is performed, and Thres = 0.6 in this embodiment.

[0043] Step 3: for the max_mth cluster, the cluster high-order feature is updated with the high-order feature of the nth signal, that is, Wherein, α is the cluster high-order feature smoothing coefficient, and α = 1 in this embodiment.

[0044] Step 4: a new cluster is added, that is, m = m + 1, and the cluster high-order feature is the high-order feature of the nth signal, that is,

[0045] Step 5: from the second signal, the first step to the fourth step are processed in this way until all the signals are processed, that is, the signal clustering processing based on the neural network signal high-order feature is completed.

[0046] The units described in the embodiments of the present application can be realized by software or hardware, and the described units can also be arranged in a processor. In some cases, the names of these units do not constitute a limitation on the units themselves.

[0047] According to an aspect of an embodiment of the present application, there is provided a computer program product or computer program, comprising computer instructions 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 to cause the computer device to perform the method provided in various optional implementation manners described above.

[0048] As another aspect, the embodiment of the present application further provides a computer readable medium, which can be included in the electronic device described in the above embodiments, or can exist independently without being assembled into the electronic device. The computer readable medium carries one or more programs, which, when executed by the electronic device, cause the electronic device to implement the method described in the above embodiments.

Claims

1. A method for clustering electromagnetic signal streams based on high-order features of neural network signals, characterized in that, The method comprises the following steps: Let the high-order feature of the signal extracted by the neural network for the nth signal be denoted as Feature n , where n = 1, 2, …, N, N represents the total number of signals, the dimension of Feature n is K; for the mth cluster, the cluster high-order feature is denoted as Feature , where m = 1, 2, …, M, M represents the total number of signals, the dimensions of Feature n and Feature are both K; taking the 1st signal as the 1st cluster, the cluster high-order feature is Feature After initialization, the following steps are performed: S1: for the nth signal, if there are m clustering results, then the cosine similarity between the nth signal and the mth clustering result is calculated in turn, and the cosine similarity is: wherein, denotes the absolute value of the dot product of the nth signal high-order feature and the mth cluster feature, ||Feature n denotes the Euclidean norm of the nth signal high-order feature, denotes the Euclidean norm of the mth cluster high-order feature; S2: find R n,m MaxR is the maximum value n,m The corresponding serial number is max_m; set the threshold value Thres, if MaxR n,m ≥ Thres, the nth signal is classified into the max_mth cluster, and S3 is performed; if MaxR n,m < Thres, it is considered that the nth signal belongs to a new cluster, and S4 is performed; S3: for the max_mth clustering, the clustering high-order feature is updated with the nth signal high-order feature, That is wherein a is a clustering high-order feature smoothing coefficient, the dot product of the nth signal high-order feature and the mth clustering feature is greater than or equal to 0, Otherwise S4: a new cluster is added, i.e. m = m + 1, and the high-order feature of the cluster is the nth signal high-order feature, i.e. S5: from the 2nd signal, the S1-S4 are processed in this way until all the signals are processed, that is, the signal clustering processing based on the neural network signal high-order feature is completed.

2. The method of claim 1, wherein, The value of the dimension K includes 32.

3. The method of claim 1, wherein, The value of the threshold Thres includes 0.

6.

4. The method of claim 1, wherein, The value of the clustering high-order feature smoothing coefficient a includes 1.

5. An electromagnetic signal flow clustering device based on high-order features of neural network signals, characterized in that, The device comprises a processor and a memory, and the memory stores a computer program.

6. A computer-readable storage medium, characterized in that, The computer program is stored in a readable storage medium and loaded by a processor to execute the method according to any one of claims 1-4.

7. A neural network signal high-order feature based electromagnetic signal stream clustering system, characterized in that, The device comprises the device according to claim 5.

Images