Radiation source identification method, device, system, equipment and program product

Abstract

The invention provides a radiation source identification method, device, system, equipment and program product, and the method comprises the steps: obtaining a pulse characteristic and a time frequency characteristic of a target electromagnetic signal, and determining an identification result of a radiation source corresponding to the target electromagnetic signal based on the pulse characteristic and the time frequency characteristic of the target electromagnetic signal, the pulse features are determined based on intra-pulse features and / or inter-pulse features of the target electromagnetic signal, and the time-frequency features of the target electromagnetic signal are determined based on time-domain features and frequency-domain features of the target electromagnetic signal. According to the radiation source identification method, the identification precision of the radiation source can be effectively improved.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a radiation source identification method, device, system, equipment and program product. Background Art

[0002] With the advancement of science and technology, radiation source identification technology has developed rapidly. Radiation source identification refers to the identification of each radiation source by receiving the electromagnetic signal emitted by the radiation source and analyzing its unique characteristics.

[0003] However, the existing radiation source identification technology usually identifies the radiation source that emits the electromagnetic signal based on the pulse-to-pulse characteristics of the electromagnetic signal emitted by the radiation source, and the radiation source identification accuracy is poor. Summary of the invention

[0004] Based on the above requirements, the present application proposes a radiation source identification method, device, system, equipment and program product to solve the problem that the prior art cannot guarantee accurate identification of the radiation source.

[0005] In order to achieve the above technical objectives, this application specifically proposes the following technical solutions:

[0006] The first aspect of the present application provides a radiation source identification method, comprising:

[0007] Acquire the pulse characteristics and time-frequency characteristics of the target electromagnetic signal; the pulse characteristics of the target electromagnetic signal are determined based on the intra-pulse characteristics and / or inter-pulse characteristics of the target electromagnetic signal; the time-frequency characteristics of the target electromagnetic signal are determined based on the time domain characteristics and frequency domain characteristics of the target electromagnetic signal;

[0008] Based on the pulse characteristics and time-frequency characteristics of the target electromagnetic signal, an identification result of a radiation source corresponding to the target electromagnetic signal is determined.

[0009] The second aspect of the present application provides a radiation source identification device, comprising:

[0010] A first data processing module is used to obtain the pulse characteristics and time-frequency characteristics of the target electromagnetic signal; the pulse characteristics of the target electromagnetic signal are determined based on the intra-pulse characteristics and / or inter-pulse characteristics of the target electromagnetic signal; the time-frequency characteristics of the target electromagnetic signal are determined based on the time domain characteristics and frequency domain characteristics of the target electromagnetic signal;

[0011] The second data processing module is used to determine the identification result of the radiation source corresponding to the target electromagnetic signal based on the pulse characteristics and time-frequency characteristics of the target electromagnetic signal.

[0012] The third aspect of the present application provides a radiation source identification system, including a signal acquisition module and a signal processing module;

[0013] The signal acquisition module is used to acquire target electromagnetic signals;

[0014] The signal processing module is used to execute any one of the radiation source identification methods described above.

[0015] In a fourth aspect, the present application proposes a storage medium, on which a computer program is stored. When the computer program is executed by a processor, the radiation source identification method as described above is implemented.

[0016] A fifth aspect of the present application provides an electronic device, comprising:

[0017] Memory and processor;

[0018] The memory is connected to the processor and is used to store programs;

[0019] The processor is used to implement the above-mentioned radiation source identification method by running the program in the memory.

[0020] A sixth aspect of the present application proposes a computer program product, comprising computer program instructions, which, when executed by a processor, enable the processor to execute the radiation source identification method as described above.

[0021] The radiation source identification method proposed in the present application obtains the pulse characteristics and time-frequency characteristics of the target electromagnetic signal, and determines the identification result of the radiation source corresponding to the target electromagnetic signal based on the pulse characteristics and time-frequency characteristics of the target electromagnetic signal, wherein the pulse characteristics are determined based on the intra-pulse characteristics and / or inter-pulse characteristics of the target electromagnetic signal, and the time-frequency characteristics of the target electromagnetic signal are determined based on the time domain characteristics and frequency domain characteristics of the target electromagnetic signal. Therefore, by comprehensively considering the multi-dimensional characteristics of the target electromagnetic signal, the identification accuracy of the radiation source can be effectively improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0023] Figure 1 A schematic diagram of the structure of a radiation source identification system provided in an embodiment of the present application.

[0024] Figure 2 A schematic diagram of a flow chart of a radiation source identification method provided in an embodiment of the present application.

[0025] Figure 3A schematic diagram of a process for determining model losses of a first initial model and a second initial model provided in an embodiment of the present application.

[0026] Figure 4 A schematic diagram of the structure of a radiation source identification device provided in an embodiment of the present application.

[0027] Figure 5 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0028] The technical solution of the embodiment of the present application is applicable to application scenarios such as communication and environmental monitoring that require radiation source identification. The technical solution of the embodiment of the present application can effectively improve the identification accuracy of the radiation source.

[0029] Conventional radiation source identification schemes are to identify each radiation source by receiving electromagnetic signals emitted by the radiation source and analyzing its unique characteristics. In a simple electromagnetic environment, radiation sources are mainly identified by pulse-to-pulse feature analysis. However, with the rapid increase in the number of radiation sources in the electromagnetic environment and the diversification of the modulation waveforms of radiation source signals, the accuracy of radiation source identification cannot be guaranteed by relying solely on pulse-to-pulse feature analysis.

[0030] Based on the above-mentioned technical status, the inventors of the present application have proposed a new radiation source identification scheme after research. The radiation source identification scheme obtains the pulse characteristics and time-frequency characteristics of the target electromagnetic signal, and determines the identification result of the radiation source corresponding to the target electromagnetic signal based on the pulse characteristics and time-frequency characteristics of the target electromagnetic signal, wherein the pulse characteristics are determined based on the intra-pulse characteristics and / or inter-pulse characteristics of the target electromagnetic signal, and the time-frequency characteristics of the target electromagnetic signal are determined based on the time domain characteristics and frequency domain characteristics of the target electromagnetic signal. Therefore, by comprehensively considering the multi-dimensional characteristics of the target electromagnetic signal, the identification accuracy of the radiation source can be effectively improved.

[0031] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0032] Figure 1 An exemplary implementation environment of the technical solution proposed in the embodiment of the present application is shown, and the implementation environment is a radiation source identification system. Figure 1 As shown, the system at least includes a signal acquisition module 101 and a signal processing module 102. The radiation source identification system can be set at any position, so as to realize automatic identification of the radiation source.

[0033] The signal acquisition module 101 is used to detect and record electromagnetic signals in space. The electromagnetic signal may include a signal emitted by a radiation source, and the type of the electromagnetic signal may depend on the physical properties of the radiation source. For example, the electromagnetic signal may include one or more signals of the type of radio wave signal, microwave signal, infrared signal, ultraviolet signal, X-ray signal, etc. The number of the signal acquisition modules 101 may be one or more, and may be specifically set according to actual needs.

[0034] The signal acquisition module 101 sends the collected electromagnetic signal to the signal processing module 102 , and the signal processing module 102 identifies the radiation source by processing and calculating the electromagnetic signal collected by the signal acquisition module 101 .

[0035] One or more services or applications capable of performing radiation source identification processing are running on the signal processing module 102. As an optional implementation, the signal processing module 102 may be a computer, a processing chip, a server, or other device with data processing capabilities.

[0036] The signal acquisition module 101 and the signal processing module 102 can be connected via a wired or wireless communication link. When the two are connected via a wired communication link, they can be connected via any type of audio signal line such as a coaxial signal line, a digital signal line, an optical cable signal line, a balanced signal line, etc. When the two are connected via a wireless communication link, they can be connected to each other via a network. The network can be any type of network. For example, the network can be a network that can be subdivided into multiple subnetworks. The network or the multiple subnetworks contained in the network can be a local area network (LAN), Ethernet, token ring, wide area network (WAN), Internet, virtual network, virtual private network (VPN), intranet, extranet, public switched telephone network (PSTN), infrared network, wireless network (such as Bluetooth, WIFI), mobile network (such as 4G, 5G), Internet of Things and / or any combination of these and / or other networks.

[0037] Furthermore, the above-mentioned radiation source identification system may also include a memory for storing the electromagnetic signal collected by the signal acquisition module 101, or storing the radiation source identification result processed by the signal processing module 102, or storing the intermediate processing result of the signal processing module 102, etc.

[0038] Combine the following Figure 1 The application scenario shown provides an exemplary introduction to the radiation source identification solution proposed in the embodiment of the present application.

[0039] See also Figure 2The embodiment of the present application first proposes a radiation source identification method, which can be executed by a processor and specifically includes:

[0040] S201. Acquire the pulse characteristics and time-frequency characteristics of the target electromagnetic signal; the pulse characteristics of the target electromagnetic signal are determined based on the intra-pulse characteristics and / or inter-pulse characteristics of the target electromagnetic signal; the time-frequency characteristics of the target electromagnetic signal are determined based on the time domain characteristics and frequency domain characteristics of the target electromagnetic signal.

[0041] Specifically, the target electromagnetic signal may include an electromagnetic signal for which radiation source identification is to be performed among the electromagnetic signals collected by the signal collection module, for example, an electromagnetic signal with an abnormal signal, a designated electromagnetic signal, and the like.

[0042] During implementation, the target electromagnetic signal can be a time series of predetermined duration. When the electromagnetic signal collected by the signal acquisition module includes multiple electromagnetic signals, the target electromagnetic signal can also be selected from the electromagnetic signal collected by the signal acquisition module to facilitate identification of the radiation source based on the target electromagnetic signal.

[0043] The pulse characteristics of the target electromagnetic signal can be used to characterize the characteristics of multiple attributes of the pulse in the target electromagnetic signal, and can also include the correlation between the characteristics of different attributes. In implementation, the pulse characteristics of the target electromagnetic signal can be determined based on the intra-pulse characteristics and / or inter-pulse characteristics of the target electromagnetic signal. Among them, the intra-pulse characteristics of the target electromagnetic signal can be used to characterize the internal attributes of a single pulse in the target electromagnetic signal, for example, it can include amplitude characteristics, frequency characteristics, phase characteristics, noise and interference characteristics, etc. The inter-pulse characteristics of the target electromagnetic signal can be used to characterize the attributes between different pulses in the target electromagnetic signal, for example, it can include pulse repetition period, pulse repetition frequency, pulse sequence mode, pulse carrier frequency, modulation type, encoding method, etc. Therefore, in the process of determining the pulse characteristics of the target electromagnetic signal according to the intra-pulse characteristics and / or inter-pulse characteristics of the target electromagnetic signal, the reliability of the pulse characteristics of the target electromagnetic signal can be effectively guaranteed.

[0044] The time-frequency characteristics of the target electromagnetic signal can be used to characterize the time evolution characteristics and frequency composition characteristics of the target electromagnetic signal. In implementation, the time-frequency characteristics of the target electromagnetic signal can be determined based on the time domain characteristics and frequency domain characteristics of the target electromagnetic signal.

[0045] Among them, the time domain characteristics of the target electromagnetic signal can be obtained by performing time domain analysis on the target electromagnetic signal. The time domain analysis can include calculating the statistics of the time attributes of the target electromagnetic signal (such as mean, variance, etc.) to describe the overall time domain characteristics of the target electromagnetic signal based on the statistics of the time attributes. In addition, the time domain analysis can also include extracting the instantaneous characteristics of the target electromagnetic signal (such as rise time, fall time, peak power, etc.) to reflect the pulse shape of the target electromagnetic signal through the instantaneous characteristics.

[0046] In addition, the frequency domain characteristics of the target electromagnetic signal can be obtained by performing frequency domain analysis on the target electromagnetic signal. The frequency domain analysis can include Fourier transform, spectrum estimation, etc. to convert the target electromagnetic signal from a time series into a frequency representation.

[0047] It is understandable that the time-frequency sub-features of the target electromagnetic signal can also be obtained by performing time-frequency analysis on the target electromagnetic signal. The time-frequency analysis can include short-time Fourier transform, wavelet transform, etc.

[0048] During implementation, the time domain characteristics and frequency domain characteristics of the target electromagnetic signal can be fused to obtain the time-frequency characteristics of the target electromagnetic signal. The time domain characteristics, frequency domain characteristics and time-frequency sub-features of the target electromagnetic signal can also be fused to obtain the time-frequency characteristics of the target electromagnetic signal. The time-frequency sub-features of the target electromagnetic signal can also be directly used as the time-frequency characteristics of the target electromagnetic signal, thereby effectively ensuring the reliability of the time-frequency characteristics of the target electromagnetic signal.

[0049] S202: Determine an identification result of a radiation source corresponding to the target electromagnetic signal based on the pulse characteristics and time-frequency characteristics of the target electromagnetic signal.

[0050] Specifically, the type of the identification result of the radiation source can be set according to actual needs, for example, it can be the type of the radiation source, or it can be identification information of the radiation source, etc.

[0051] In implementation, the identification result of the radiation source corresponding to the target electromagnetic signal can be determined based on the pulse characteristics and time-frequency characteristics of the target electromagnetic signal.

[0052] For example, the pulse characteristics and time-frequency characteristics of the target electromagnetic signal can be input into a pre-trained radiation source identification model, so that the radiation source identification model can determine the identification result of the radiation source corresponding to the target electromagnetic signal based on the pulse characteristics and time-frequency characteristics of the target electromagnetic signal; in addition, the pulse characteristics and time-frequency characteristics of the target electromagnetic signal can be fused to obtain the fusion characteristics corresponding to the target electromagnetic signal, so as to determine the identification result of the radiation source corresponding to the target electromagnetic signal based on the predetermined correspondence between the identification result of the radiation source and the fusion characteristics and the fusion characteristics corresponding to the target electromagnetic signal.

[0053] From the above introduction, it can be seen that the radiation source identification method proposed in the embodiment of the present application can effectively improve the identification accuracy of the radiation source by comprehensively considering the pulse characteristics and time-frequency characteristics of the target electromagnetic signal.

[0054] In some embodiments, acquiring a pulse characteristic of a target electromagnetic signal includes:

[0055] The sampling data of the target electromagnetic signal is input into a pre-trained feature extraction model, and the feature extraction model determines the primary features of the target electromagnetic signal based on the sampling data of the target electromagnetic signal, and determines the pulse features of the target electromagnetic signal based on the primary features.

[0056] Specifically, the feature extraction model can be used to extract the pulse characteristics of the target electromagnetic signal. The feature extraction model can adopt a machine learning model, for example, a convolutional neural network model, or a Transformer encoder model, which can be set according to actual needs. In implementation, the feature extraction model can be trained in advance so that in the process of radiation source identification, the pulse characteristics of the target electromagnetic signal can be quickly and accurately obtained through the pre-trained feature extraction model.

[0057] The sampling data of the target electromagnetic signal can be obtained by sampling the time series corresponding to the target electromagnetic signal. For example, the time series corresponding to the target electromagnetic signal can be sampled based on a predetermined sampling interval to obtain the sampling data of the target electromagnetic signal. Therefore, the sampling data of the target electromagnetic signal can effectively characterize the intra-pulse characteristics and inter-pulse characteristics of the target electromagnetic signal, and then the intra-pulse characteristics and inter-pulse characteristics of the target electromagnetic signal can be used to adapt to the identification of radiation sources in more complex environments.

[0058] The primary features of the target electromagnetic signal may be shallow features of the target electromagnetic signal. For example, the primary features may be relatively basic and intuitive features of the target electromagnetic signal, such as amplitude change features, edge features, transition features, periodic features, repetitive patterns, frequency features, waveform features, phase features, texture features, local statistics (such as mean, variance, skewness, kurtosis, etc.), noise features, etc.

[0059] During implementation, the sampling data of the target electromagnetic signal can be input into a pre-trained feature extraction model, and the feature extraction model can determine the primary features of the target electromagnetic signal based on the sampling data of the target electromagnetic signal, and determine the pulse features of the target electromagnetic signal based on the primary features of the target electromagnetic signal. For example, the primary features of the target electromagnetic signal can be further extracted by the feature extraction model to obtain the pulse features of the target electromagnetic signal.

[0060] The feature extraction model may include a first feature extraction sub-model and a second feature extraction sub-model, and the output end of the first feature extraction sub-model may be connected to the input end of the second feature extraction sub-model. In implementation, the sampling data of the target electromagnetic signal may be input into the first feature extraction sub-model, the first feature extraction sub-model determines the primary features of the target electromagnetic signal according to the sampling data of the target electromagnetic signal and outputs the primary features to the second feature extraction sub-model, and the second feature extraction sub-model extracts the primary features of the target electromagnetic signal to obtain the pulse features of the target electromagnetic signal, thereby being able to quickly and accurately obtain the pulse features of the target electromagnetic signal.

[0061] Optionally, the first feature extraction submodel may use an M-layer convolutional neural network, where M>1. For example, a 7-layer 1-dimensional convolutional neural network may be used, so that the primary features of the target electromagnetic signal can be effectively acquired through the first feature extraction submodel. The second feature extraction submodel may use an N-layer Transformer encoder module, where N>1. For example, a 14-layer Transformer encoder module may be used. The Transformer encoder module may include a self-attention module. The self-attention module may allow the features at each position in the input feature sequence to interact with the features at other positions to capture the global dependencies and contextual information of the input feature sequence, thereby effectively improving the accuracy of the pulse features output by the second feature extraction submodel.

[0062] In some embodiments, the method for training the feature extraction model includes:

[0063] Acquire first sample data; the first sample data includes sampling data and time-frequency characteristics of the first sample electromagnetic signal;

[0064] Based on the first sample data and the first target loss function, performing unsupervised training on the first initial model to obtain the feature extraction model;

[0065] Among them, the first target loss function is used to determine the model loss of the first initial model based on the quantitative characteristics, pulse characteristics and time-frequency characteristics of the first sample electromagnetic signal; the quantitative characteristics are used to characterize the quantization results of the primary characteristics of the first sample electromagnetic signal.

[0066] Specifically, the first sample data can be used to train the first initial model to obtain a feature extraction model. The first sample data may include multiple first samples, one first sample may be determined based on a first sample electromagnetic signal, and the first sample electromagnetic signal may include an electromagnetic signal collected in a historical time interval.

[0067] For example, for any first sample, the first sample may include sampling data of a corresponding first sample electromagnetic signal. The sampling data of the first sample electromagnetic signal may be obtained by sampling a time series corresponding to the first sample electromagnetic signal. For example, the time series corresponding to the first sample electromagnetic signal may be sampled based on a predetermined sampling interval to obtain the sampling data of the first sample electromagnetic signal.

[0068] In addition, the first sample may also include the time-frequency characteristics of the corresponding first sample electromagnetic signal, and the time-frequency characteristics of the first sample electromagnetic signal may be determined based on the time domain characteristics and frequency domain characteristics of the first sample electromagnetic signal. Among them, the time domain characteristics of the first sample electromagnetic signal can be obtained by performing time domain analysis on the first sample electromagnetic signal, and the time domain analysis may include calculating the statistics of the time attributes of the first sample electromagnetic signal (such as mean, variance, etc.) to describe the overall time domain characteristics of the first sample electromagnetic signal according to the statistics of the time attributes. In addition, the time domain analysis may also include extracting the instantaneous characteristics of the first sample electromagnetic signal (such as rise time, fall time, peak power, etc.) to reflect the pulse shape of the first sample electromagnetic signal through the instantaneous characteristics. In addition, the frequency domain characteristics of the first sample electromagnetic signal can be obtained by performing frequency domain analysis on the first sample electromagnetic signal, and the frequency domain analysis may include Fourier transform, spectrum estimation, etc. to convert the first sample electromagnetic signal from a time series to a frequency representation. It is understandable that the time-frequency sub-features of the first sample electromagnetic signal can also be obtained by performing time-frequency analysis on the first sample electromagnetic signal. The time-frequency analysis can include short-time Fourier transform, wavelet transform, etc.

[0069] During implementation, the time domain characteristics and frequency domain characteristics of the first sample electromagnetic signal can be fused to obtain the time-frequency characteristics of the first sample electromagnetic signal. The time domain characteristics, frequency domain characteristics and time-frequency sub-features of the first sample electromagnetic signal can also be fused to obtain the time-frequency characteristics of the first sample electromagnetic signal. The time-frequency sub-features of the first sample electromagnetic signal can also be directly used as the time-frequency characteristics of the first sample electromagnetic signal, thereby effectively ensuring the reliability of the time-frequency characteristics of the first sample electromagnetic signal.

[0070] Wherein, the first initial model can be unsupervisedly trained based on the first sample data and the first target loss function, and the trained first initial model can be used as a feature extraction model. The first target loss function can determine the model loss of the first initial model based on the quantization characteristics, pulse characteristics and time-frequency characteristics of the first sample electromagnetic signal, and the quantization characteristics can be used to characterize the quantization results of the primary characteristics of the corresponding first sample electromagnetic signal. For example, for any first sample electromagnetic signal, the quantization results of the primary characteristics of the first sample electromagnetic signal can be obtained by performing a Gumbel Softmax transformation on the primary characteristics of the first sample electromagnetic signal. Wherein, the first target loss function can be used to characterize the predetermined correspondence between the quantization characteristics, pulse characteristics and time-frequency characteristics of the first sample electromagnetic signal and the model loss of the first initial model.

[0071] Optionally, the first target loss function may include the superposition result of multiple loss functions, for example, each loss function may be superimposed based on the first predetermined coefficients corresponding to different loss functions. Wherein, for any loss function in the first target loss function, the loss function may determine the model loss of the first initial model based on at least one of the quantization feature, pulse feature and time-frequency feature of the first sample electromagnetic signal.

[0072] In implementation, the first sample data may be input into the first initial model, so that the first initial model can determine the primary features of each first sample electromagnetic signal according to the sampling data of each first sample electromagnetic signal in the first sample data, and determine the pulse features of each first sample electromagnetic signal according to the primary features of each first sample electromagnetic signal. Thus, the primary features of each first sample electromagnetic signal can be quantized to obtain the quantized features of each first sample electromagnetic signal, and the quantized features, pulse features and time-frequency features of each first sample electromagnetic signal are input into the first target loss function to obtain the model loss of the first initial model, so as to adjust the model parameters of the first initial model in the direction of reducing the model loss, until the iteration end condition is met, and the first initial model after parameter adjustment is used as the feature extraction model. The iteration end condition may include that the model loss of the first initial model is less than the first predetermined loss, and may also include that the number of iterations in the training process of the first initial model reaches the first predetermined number of iterations.

[0073] Existing radiation source identification methods based on deep learning usually require a large amount of labeled sample data to train the deep learning model. However, in the field of radiation source identification, sample data collection is relatively difficult, which makes the number of labeled sample data limited. As a result, the features learned by the deep learning model are limited, and the accurate identification of the radiation source cannot be guaranteed. However, through the method of the embodiment of the present application, in the training process of the feature extraction model, it is only necessary to obtain the sampling data and time-frequency features of the first sample electromagnetic signal, and there is no need to obtain the labeled data of the radiation source corresponding to each first sample electromagnetic signal, so as to effectively avoid the influence of the small number of labeled sample data on the identification accuracy of the radiation source. At the same time, the method of the embodiment of the present application determines the model loss of the first initial model through the quantitative features, pulse features and time-frequency features of each first sample electromagnetic signal in the first sample data, and trains the first initial model according to the model loss to obtain the feature extraction model, which can further improve the identification accuracy of the radiation source.

[0074] In some embodiments, determining the model loss of the first initial model based on the quantitative characteristics, pulse characteristics and time-frequency characteristics of the first sample electromagnetic signal includes:

[0075] determining a first loss of the first initial model based on a quantized feature of the first sample electromagnetic signal, determining a second loss of the first initial model based on a quantized feature of the first sample electromagnetic signal and a pulse feature of the first sample electromagnetic signal, and determining a third loss of the first initial model based on a mapping feature of the first sample electromagnetic signal and a time-frequency feature of the first sample electromagnetic signal; the mapping feature of the first sample electromagnetic signal is obtained by performing feature mapping on the pulse feature of the first sample electromagnetic signal;

[0076] A model loss of the first initial model is determined based on the first loss, the second loss, and the third loss of the first initial model.

[0077] Specifically, the first objective loss function may include a first loss function, a second loss function, and a third loss function.

[0078] Wherein, the first loss function can determine the first loss of the first initial model based on the quantitative characteristics of the first sample electromagnetic signal, and the first loss can be a feature diversity loss, which can be used to determine based on the distribution probability of each feature element. Wherein, the quantitative feature can be a V-dimensional feature vector, that is, a quantitative feature can include V elements, and the feature element can be an element in the quantitative characteristics of each first sample electromagnetic signal, so that the first initial model is trained by the feature diversity loss, which can effectively improve the diversity of the feature elements of the primary features extracted by the first feature extraction submodel in the trained feature extraction model, and then enable the first feature extraction submodel to learn as many enhanced representations of the feature elements as possible, thereby, the feature extraction model can effectively improve the recognition accuracy of the pulse feature according to the diversified primary features of the feature elements.

[0079] Optionally, feature diversity loss L d It can be shown as formula (1):

[0080]

[0081] Where G is the number of quantitative features in a batch, and a batch is the number of first samples simultaneously input to the first initial model; V is the number of feature elements in a quantitative feature; p g，v is the probability that the vth feature element in the gth quantitative feature appears among all G*V feature elements.

[0082] The second loss function can determine the second loss of the first initial model based on the quantitative features of the first sample electromagnetic signal and the pulse features of the first sample electromagnetic signal. The second loss can be a contrast loss. The contrast loss can be determined based on the similarity between the pulse features of the first sample electromagnetic signal and each quantitative feature. The similarity between the pulse feature and the quantitative feature can be a predetermined type of similarity, for example, sine similarity, cosine similarity, etc. Thus, the first initial model is trained through the contrast loss, so that the second feature extraction sub-model in the trained feature extraction model can learn the discriminative representation information, thereby effectively improving the stability and generalization ability of the second feature extraction sub-model, thereby improving the recognition accuracy of the feature extraction model for the pulse feature.

[0083] Optionally, contrast loss L c It can be shown as formula (2):

[0084]

[0085] Where, sim() represents cosine similarity; c t is the pulse characteristic of the tth first sample electromagnetic signal; q tis the quantitative feature of the electromagnetic signal of the tth first sample; k is the number of first samples in a batch; Q t is a quantitative feature set, including the quantitative features corresponding to each first sample in a batch; Q t Middle t Other quantitative features.

[0086] The third loss function can determine the third loss of the first initial model based on the mapping features of the first sample electromagnetic signal and the time-frequency features of the first sample electromagnetic signal. The mapping features of the first sample electromagnetic signal can be obtained by feature mapping the pulse features of the first sample electromagnetic signal. For example, the pulse features of the first sample electromagnetic signal can be feature mapped by a feature mapping network. In implementation, a feature mapping network can be connected to the output end of the second feature extraction submodel to feature map the pulse features of the first sample electromagnetic signal output by the second feature extraction submodel through the feature mapping network to obtain the mapping features of the first sample electromagnetic signal, and determine the third loss of the first initial model based on the mapping features and time-frequency features of the first sample electromagnetic signal. It can be understood that the feature mapping network can be used only in the process of training the first initial model, and does not need to be used in the process of obtaining the pulse features of the target electromagnetic signal through the feature extraction model obtained by training.

[0087] Among them, the third loss can be a regression loss, and the regression loss can be determined based on the difference between the mapping feature of the first sample electromagnetic signal and the time-frequency feature of the first sample electromagnetic signal. Therefore, by training the first initial model through regression loss, the second feature extraction sub-model obtained through training can be strengthened to learn deep features, thereby effectively improving the pulse feature recognition accuracy of the second feature extraction sub-model. At the same time, in the process of training the first initial model, the learning of the first initial model is constrained by the time-frequency feature of the first sample electromagnetic signal, which can further improve the pulse feature recognition accuracy of the second feature extraction sub-model obtained through training.

[0088] Optionally, the regression loss L of the first initial model m It can be shown as formula (3):

[0089] L m =(p t ―s t ) 2 (3)

[0090] In the formula, p t is the mapping feature of the tth first sample electromagnetic signal; s t is the time-frequency characteristics of the tth first sample electromagnetic signal.

[0091] Optionally, the determination process of the first loss, the second loss and the third loss of the first initial model may be specifically as follows: Figure 3 shown.

[0092] In implementation, the model loss of the first initial model can be determined based on the first loss, the second loss and the third loss of the first initial model. As a preferred implementation, the first loss, the second loss and the third loss of the first initial model can be superimposed to obtain the model loss of the first initial model, so that according to the model loss of the first initial model, the training efficiency of the first initial model and the robustness of the training results can be effectively improved.

[0093] Optionally, the model loss L1 of the first initial model can be expressed as formula (4):

[0094] L1=L c +αL d +βL m (4)

[0095] Where α and β are adjustable hyperparameters.

[0096] In some embodiments, determining the identification result of the radiation source corresponding to the target electromagnetic signal based on the pulse characteristics and time-frequency characteristics of the target electromagnetic signal includes:

[0097] The pulse characteristics and time-frequency characteristics of the target electromagnetic signal are input into a pre-trained radiation source recognition model, and the radiation source recognition model determines the recognition result of the radiation source corresponding to the target electromagnetic signal based on the pulse characteristics and time-frequency characteristics of the target electromagnetic signal.

[0098] Specifically, the radiation source identification model can be used to automatically identify the radiation source corresponding to the target electromagnetic signal according to the pulse characteristics and time-frequency characteristics of the target electromagnetic signal. Among them, the radiation source identification model can adopt a machine learning model, for example, the radiation source identification model can adopt a classification model, that is, the classification model can predict the radiation source corresponding to the target electromagnetic signal according to the pulse characteristics and time-frequency characteristics of the target electromagnetic signal. In addition, the radiation source identification model can also include a coding model and a classification model connected in sequence. The coding model can further extract features from the pulse characteristics of the target electromagnetic signal to obtain the coding features of the target electromagnetic signal and output them to the classification model. The classification model can predict the radiation source corresponding to the target electromagnetic signal according to the coding features and time-frequency features of the target electromagnetic signal.

[0099] In implementation, the radiation source identification model may be trained in advance, so that in the radiation source identification process, the identification result of the radiation source corresponding to the target electromagnetic signal can be quickly and accurately obtained through the pre-trained radiation source identification model.

[0100] After acquiring the pulse characteristics and time-frequency characteristics of the target electromagnetic signal, the pulse characteristics and time-frequency characteristics of the target electromagnetic signal can be input into a pre-trained radiation source identification model. The radiation source identification model can predict the radiation source corresponding to the target electromagnetic signal based on the pulse characteristics and time-frequency characteristics of the target electromagnetic signal to obtain an identification result of the radiation source corresponding to the target electromagnetic signal, thereby being able to quickly and accurately identify the radiation source corresponding to the target electromagnetic signal. At the same time, by comprehensively considering the pulse characteristics and time-frequency characteristics of the target electromagnetic signal, the accuracy of the identification result of the radiation source corresponding to the target electromagnetic signal can be further improved.

[0101] In some embodiments, the radiation source identification model includes a coding model and a classification model;

[0102] Inputting the pulse characteristics and time-frequency characteristics of the target electromagnetic signal into a pre-trained radiation source recognition model, and the radiation source recognition model determines the recognition result of the radiation source corresponding to the target electromagnetic signal according to the pulse characteristics and time-frequency characteristics of the target electromagnetic signal, including:

[0103] Inputting the pulse characteristics of the target electromagnetic signal into the coding model, and the coding model determines the coding characteristics of the target electromagnetic signal according to the pulse characteristics of the target electromagnetic signal;

[0104] The classification model determines the recognition result of the radiation source corresponding to the target electromagnetic signal according to the coding characteristics of the target electromagnetic signal and the time-frequency characteristics of the target electromagnetic signal.

[0105] Specifically, the radiation source identification model may include a coding model and a classification model. The input end of the coding model can be used to input the pulse characteristics of the target electromagnetic signal, and the output end of the coding model can be connected to the input end of the classification model. At the same time, the input end of the classification model is also used to input the time-frequency characteristics of the target electromagnetic signal.

[0106] The coding model can be used to extract the pulse characteristics of the target electromagnetic signal to obtain the coding characteristics of the target electromagnetic signal and output it to the classification model. For example, the coding model can use a recurrent neural network coding module, a convolutional neural network coding module, or a Transformer coding module, which can be set according to actual needs.

[0107] The classification model can predict the radiation source corresponding to the target electromagnetic signal according to the coding features and time-frequency features of the target electromagnetic signal to obtain the identification result of the radiation source corresponding to the target electromagnetic signal. For example, the classification model can use a softmax classification module, a logistic regression classification module, a fully connected layer, etc.

[0108] During implementation, the pulse characteristics of the target electromagnetic signal can be input into a coding model to determine the coding characteristics of the target electromagnetic signal through the coding model based on the pulse characteristics of the target electromagnetic signal, and output to a classification model to determine the identification result of the radiation source corresponding to the target electromagnetic signal through the classification model based on the coding characteristics of the target electromagnetic signal and the time-frequency characteristics of the target electromagnetic signal.

[0109] Therefore, through the method of the embodiment of the present application, the coding model extracts the pulse characteristics of the target electromagnetic signal to obtain the coding characteristics of the target electromagnetic signal, and then the classification model predicts the radiation source corresponding to the target electromagnetic signal based on the coding characteristics and time-frequency characteristics of the target electromagnetic signal, which can further improve the accuracy of the recognition result of the radiation source corresponding to the target electromagnetic signal.

[0110] In some embodiments, the method for training the radiation source identification model includes:

[0111] Acquire second sample data; the second sample data includes a pulse feature, a time-frequency feature, and a radiation source label corresponding to the second sample electromagnetic signal;

[0112] Based on the second sample data and the second target loss function, supervised training is performed on the second initial model to obtain the radiation source identification model;

[0113] Among them, the two-objective loss function is used to determine the first loss of the second initial model according to the radiation source identification result corresponding to the second sample electromagnetic signal and the radiation source label corresponding to the second sample electromagnetic signal, and determine the model loss of the second initial model based on the first loss of the second initial model.

[0114] Specifically, the second sample data can be used to train the second initial model to obtain a radiation source identification model. The second sample data may include multiple second samples, one second sample may be determined based on a second sample electromagnetic signal, and the second sample electromagnetic signal may include an electromagnetic signal collected in a historical time interval.

[0115] For example, for any second sample, the second sample may include sampling data of a corresponding second sample electromagnetic signal, and the sampling data of the second sample electromagnetic signal may be obtained by sampling a time series corresponding to the second sample electromagnetic signal. For example, the time series corresponding to the second sample electromagnetic signal may be sampled based on a predetermined sampling interval to obtain the sampling data of the second sample electromagnetic signal.

[0116] The second sample may also include the time-frequency characteristics of the corresponding second sample electromagnetic signal, and the time-frequency characteristics of the second sample electromagnetic signal may be determined based on the time domain characteristics and frequency domain characteristics of the second sample electromagnetic signal. The time domain characteristics of the second sample electromagnetic signal may be obtained by performing time domain analysis on the second sample electromagnetic signal, and the time domain analysis may include calculating the statistics of the time attributes of the second sample electromagnetic signal (such as mean, variance, etc.) to describe the overall time domain characteristics of the second sample electromagnetic signal according to the statistics of the time attributes. In addition, the time domain analysis may also include extracting the instantaneous characteristics of the second sample electromagnetic signal (such as rise time, fall time, peak power, etc.) to reflect the pulse shape of the second sample electromagnetic signal through the instantaneous characteristics. In addition, the frequency domain characteristics of the second sample electromagnetic signal may be obtained by performing frequency domain analysis on the second sample electromagnetic signal, and the frequency domain analysis may include Fourier transform, spectrum estimation, etc. to convert the second sample electromagnetic signal from a time series to a frequency representation. It is understandable that the time-frequency sub-features of the second sample electromagnetic signal can also be obtained by performing time-frequency analysis on the second sample electromagnetic signal. The time-frequency analysis can include short-time Fourier transform, wavelet transform, etc.

[0117] In implementation, the time domain features and frequency domain features of the second sample electromagnetic signal can be fused to obtain the time-frequency features of the second sample electromagnetic signal. The time domain features, frequency domain features and time-frequency sub-features of the second sample electromagnetic signal can also be fused to obtain the time-frequency features of the second sample electromagnetic signal. The time-frequency sub-features of the second sample electromagnetic signal can also be directly used as the time-frequency features of the second sample electromagnetic signal, thereby effectively ensuring the reliability of the time-frequency features of the second sample electromagnetic signal. It can be understood that the method of obtaining the time-frequency features of the target electromagnetic signal, the method of obtaining the time-frequency features of the first sample electromagnetic signal, and the method of obtaining the time-frequency features of the second sample electromagnetic signal need to remain the same, so that according to the sampling data and time-frequency features of the target electromagnetic signal, as well as the feature extraction model and the radiation source identification model, the reliability of the radiation source identification result corresponding to the target electromagnetic signal can be effectively improved.

[0118] In addition, the second sample may also include a corresponding radiation source label of the second sample electromagnetic signal. The radiation source label of the second sample electromagnetic signal may be obtained by marking the radiation source corresponding to the second sample electromagnetic signal.

[0119] Among them, the second initial model can be supervised trained based on the second sample data and the second target loss function, and the trained second initial model can be used as a radiation source identification model. The second target loss function can determine the first loss of the second initial model according to the radiation source identification result corresponding to the second sample electromagnetic signal and the radiation source label corresponding to the second sample electromagnetic signal, and determine the model loss of the second initial model based on the first loss of the second initial model. For example, the second target loss function may include a predetermined correspondence between the radiation source identification result corresponding to the second sample electromagnetic signal and the radiation source label corresponding to the second sample electromagnetic signal and the first loss of the second initial model.

[0120] Optionally, the first loss of the second initial model can be a cross-entropy loss. The cross-entropy loss can effectively measure the difference between the radiation source identification result corresponding to the second sample electromagnetic signal and the radiation source label corresponding to the second sample electromagnetic signal, and then train the second initial model through the cross-entropy loss, which can effectively improve the recognition accuracy of the trained radiation source identification model.

[0121] Among them, the cross entropy loss L CE It can be shown as formula (5):

[0122]

[0123] Where R is the number of the second sample; y r is the radiation source label corresponding to the second sample electromagnetic signal in the rth second sample; y′ r is the radiation source identification result corresponding to the second sample electromagnetic signal in the rth second sample.

[0124] In implementation, the pulse characteristics and time-frequency characteristics of each second sample electromagnetic signal in the second sample data can be input into the second initial model, so that the second initial model can determine the radiation source identification results corresponding to each second sample electromagnetic signal according to the pulse characteristics and time-frequency characteristics of each second sample electromagnetic signal, and input the radiation source identification results corresponding to each second sample electromagnetic signal and the radiation source labels corresponding to each second sample electromagnetic signal into the second target loss function to obtain the first loss of the second initial model, so as to determine the model loss of the second initial model according to the first loss, and adjust the model parameters of the second initial model in the direction of reducing the model loss, until the iteration end condition is met, and the second initial model with adjusted parameters is used as the radiation source identification model. Wherein, the iteration end condition can include that the model loss of the second initial model is less than the second predetermined loss, and can also include that the number of iterations in the training process of the second initial model reaches the second predetermined number of iterations, so as to realize the training of the second initial model in a supervised manner.

[0125] Therefore, through the method of the embodiment of the present application, the second initial model is supervisedly trained through the radiation source labels corresponding to each second sample electromagnetic signal, which can effectively improve the robustness of the trained radiation source identification model. At the same time, since the radiation source identification model is mainly used to predict the radiation source according to the pulse characteristics and time-frequency characteristics of the target electromagnetic signal, the demand for sample data during its training process is relatively small, so that under the premise of a limited number of labeled second samples, the recognition accuracy of the trained radiation source identification model can still be guaranteed.

[0126] It is understandable that the second initial model can also be connected to the output end of the trained feature extraction model, and the second initial model can be trained and the feature extraction model can be fine-tuned through the third sample data. The third sample data can include multiple third samples, and a third sample can be determined based on a third sample electromagnetic signal, and the third sample electromagnetic signal can include an electromagnetic signal collected in a historical time interval. Among them, the third sample can include the sampling data, time-frequency features and radiation source labels corresponding to the third sample electromagnetic signals of the corresponding third sample electromagnetic signals. In implementation, the sampling data of each third sample electromagnetic signal in the third sample data can be input into the feature extraction model, and the feature extraction model can determine the pulse features of each third sample electromagnetic signal according to the sampling data of each third sample electromagnetic signal, and the pulse features and time-frequency features of each third sample electromagnetic signal are input into the second initial model, so as to supervise the second initial model through the radiation source labels corresponding to each third sample electromagnetic signal, and fine-tune the feature extraction model, so as to further improve the accuracy of radiation source identification.

[0127] In some embodiments, the radiation source identification model further includes a decoding model, the input end of the decoding model is connected to the output end of the encoding model; the decoding model is used to determine the regression feature of the second sample electromagnetic signal according to the encoding feature of the second sample electromagnetic signal output by the encoding model;

[0128] The two-objective loss function is also used to determine the second loss of the second initial model based on the regression characteristics of the second sample electromagnetic signal and the time-frequency characteristics of the second sample electromagnetic signal, and to determine the model loss of the second initial model based on the first loss and the second loss of the second initial model.

[0129] Specifically, the radiation source identification model may further include a decoding model, the input end of the decoding model is connected to the output end of the encoding model, that is, the encoding features output by the encoding model may be output to the classification model and the decoding model at the same time.

[0130] In the process of training the second initial model, the coding features of the second sample electromagnetic signal output by the coding model can be simultaneously output to the decoding model and the classification model, and the decoding model determines the regression features of the second sample electromagnetic signal according to the coding features of the second sample electromagnetic signal, and the classification model determines the radiation source identification result corresponding to the second sample electromagnetic signal according to the coding features and time-frequency features of the second sample electromagnetic signal.

[0131] Thus, the second target loss function can determine the first loss of the second initial model based on the radiation source identification result of the second sample electromagnetic signal and the radiation source label corresponding to the second sample electromagnetic signal. At the same time, the second target loss function can also determine the second loss of the second initial model based on the regression characteristics of the second sample electromagnetic signal and the time-frequency characteristics of the second sample electromagnetic signal. It can be understood that the decoding model can be used only in the process of training the second initial model, and does not need to be used in the process of predicting the radiation source identification result corresponding to the target electromagnetic signal by the radiation source identification model obtained through training.

[0132] Optionally, the second loss of the second initial model can be a regression loss. The regression loss can effectively evaluate the difference between the coding characteristics and the time-frequency characteristics of the second sample electromagnetic signal, and then the second initial model is trained by the regression loss, which can effectively improve the recognition accuracy of the trained radiation source recognition model.

[0133] Among them, the regression loss L of the second initial model is MSE It can be shown as formula (6):

[0134] L MSE =(f t ―s t ) 2 (6)

[0135] In the formula, f t is the regression feature of the electromagnetic signal of the second sample in the t-th second sample; s t is the time-frequency characteristic of the second sample electromagnetic signal in the tth second sample.

[0136] Optionally, the determination process of the first loss and the second loss of the second initial model may be specifically as follows: Figure 3 shown.

[0137] During implementation, the second objective loss function may superimpose the first loss and the second loss of the second initial model based on a second predetermined coefficient to obtain the model loss of the second initial model.

[0138] Optionally, the model loss L2 of the second initial model can be expressed as formula (7):

[0139] L2=LCE +ηL MSE (7)

[0140] In the formula, η is an adjustable hyperparameter.

[0141] Therefore, through the method of the embodiment of the present application, the second initial model can be trained according to the first loss and the second loss at the same time, so that the radiation source identification accuracy of the trained radiation source identification model can be further improved.

[0142] Corresponding to the above-mentioned radiation source identification method, the present application embodiment also provides a radiation source identification device, see Figure 4 As shown, the device comprises:

[0143] The first data processing module 401 is used to obtain the pulse characteristics and time-frequency characteristics of the target electromagnetic signal; the pulse characteristics of the target electromagnetic signal are determined based on the intra-pulse characteristics and / or inter-pulse characteristics of the target electromagnetic signal; the time-frequency characteristics of the target electromagnetic signal are determined based on the time domain characteristics and frequency domain characteristics of the target electromagnetic signal;

[0144] The second data processing module 402 is used to determine the identification result of the radiation source corresponding to the target electromagnetic signal based on the pulse characteristics and time-frequency characteristics of the target electromagnetic signal.

[0145] In a possible implementation manner, the first data processing module 401 is specifically configured to:

[0146] The sampling data of the target electromagnetic signal is input into a pre-trained feature extraction model, and the feature extraction model determines the primary features of the target electromagnetic signal based on the sampling data of the target electromagnetic signal, and determines the pulse features of the target electromagnetic signal based on the primary features.

[0147] In a possible implementation, a third data processing module is further included, and the third data processing module is used to:

[0148] Acquire first sample data; the first sample data includes sampling data and time-frequency characteristics of the first sample electromagnetic signal;

[0149] Based on the first sample data and the first target loss function, performing unsupervised training on the first initial model to obtain the feature extraction model;

[0150] Among them, the first target loss function is used to determine the model loss of the first initial model based on the quantitative characteristics, pulse characteristics and time-frequency characteristics of the first sample electromagnetic signal; the quantitative characteristics are used to characterize the quantization results of the primary characteristics of the first sample electromagnetic signal.

[0151] In a possible implementation, the third data processing module is specifically used to:

[0152] determining a first loss of the first initial model based on a quantized feature of the first sample electromagnetic signal, determining a second loss of the first initial model based on a quantized feature of the first sample electromagnetic signal and a pulse feature of the first sample electromagnetic signal, and determining a third loss of the first initial model based on a mapping feature of the first sample electromagnetic signal and a time-frequency feature of the first sample electromagnetic signal; the mapping feature of the first sample electromagnetic signal is obtained by performing feature mapping on the pulse feature of the first sample electromagnetic signal;

[0153] A model loss of the first initial model is determined based on the first loss, the second loss, and the third loss of the first initial model.

[0154] In a possible implementation manner, the second data processing module 402 is specifically configured to:

[0155] The pulse characteristics and time-frequency characteristics of the target electromagnetic signal are input into a pre-trained radiation source recognition model, and the radiation source recognition model determines the recognition result of the radiation source corresponding to the target electromagnetic signal based on the pulse characteristics and time-frequency characteristics of the target electromagnetic signal.

[0156] In a possible implementation manner, the radiation source identification model includes a coding model and a classification model; the second data processing module 402 is specifically used to:

[0157] Inputting the pulse characteristics of the target electromagnetic signal into the coding model, and the coding model determines the coding characteristics of the target electromagnetic signal according to the pulse characteristics of the target electromagnetic signal;

[0158] The classification model determines the recognition result of the radiation source corresponding to the target electromagnetic signal according to the coding characteristics of the target electromagnetic signal and the time-frequency characteristics of the target electromagnetic signal.

[0159] In a possible implementation, a fourth processing module is further included, and the fourth processing module is specifically configured to:

[0160] Acquire second sample data; the second sample data includes a pulse feature, a time-frequency feature, and a radiation source label corresponding to the second sample electromagnetic signal;

[0161] Based on the second sample data and the second target loss function, supervised training is performed on the second initial model to obtain the radiation source identification model;

[0162] Among them, the two-objective loss function is used to determine the first loss of the second initial model according to the radiation source identification result corresponding to the second sample electromagnetic signal and the radiation source label corresponding to the second sample electromagnetic signal, and determine the model loss of the second initial model based on the first loss of the second initial model.

[0163] In a possible implementation, the radiation source identification model further includes a decoding model, the input end of the decoding model is connected to the output end of the encoding model; the decoding model is used to determine the regression feature of the second sample electromagnetic signal according to the encoding feature of the second sample electromagnetic signal output by the encoding model;

[0164] The two-objective loss function is also used to determine the second loss of the second initial model based on the regression characteristics of the second sample electromagnetic signal and the time-frequency characteristics of the second sample electromagnetic signal, and to determine the model loss of the second initial model based on the first loss and the second loss of the second initial model.

[0165] The radiation source identification device provided in this embodiment belongs to the same application concept as the radiation source identification method provided in the above embodiments of this application, and can execute the radiation source identification method provided in any of the above embodiments of this application, and has the corresponding functional modules and beneficial effects of executing the radiation source identification method. For technical details not fully described in this embodiment, please refer to the specific processing content of the radiation source identification method provided in the above embodiments of this application, and will not be repeated here.

[0166] The functions implemented by the above modules can be implemented by the same or different processors respectively, and the embodiments of the present application are not limited thereto.

[0167] It should be understood that each module in the above device can be implemented in the form of a processor calling software. For example, the device includes a processor, the processor is connected to a memory, and instructions are stored in the memory. The processor calls the instructions stored in the memory to implement any of the above methods or realize the functions of each unit of the device, wherein the processor can be a general-purpose processor, such as a CPU or a microprocessor, etc., and the memory can be a memory in the device or a memory outside the device. Alternatively, the unit in the device can be implemented in the form of a hardware circuit, and the functions of some or all units can be realized by designing the hardware circuit. The hardware circuit can be understood as one or more processors; for example, in one implementation, the hardware circuit is an ASIC, and the functions of some or all of the above units are realized by designing the logical relationship of the components in the circuit; for another example, in another implementation, the hardware circuit can be implemented by PLD, taking FPGA as an example, which can include a large number of logic gate circuits, and the connection relationship between the logic gate circuits is configured by a configuration file, so as to realize the functions of some or all of the above units. All units of the above device can be implemented in the form of a processor calling software, or in the form of a hardware circuit, or in part by a processor calling software, and the remaining part is implemented in the form of a hardware circuit.

[0168] In an embodiment of the present application, a processor is a circuit with the ability to process signals. In one implementation, the processor may be a circuit with the ability to read and run instructions, such as a CPU, a microprocessor, a GPU, or a DSP; in another implementation, the processor may implement certain functions through the logical relationship of a hardware circuit, and the logical relationship of the hardware circuit is fixed or reconfigurable, such as a hardware circuit implemented by an ASIC or PLD, such as an FPGA. In a reconfigurable hardware circuit, the process of the processor loading a configuration document to implement the hardware circuit configuration can be understood as the process of the processor loading instructions to implement the functions of some or all of the above units. In addition, it can also be a hardware circuit designed for artificial intelligence, which can be understood as an ASIC, such as an NPU, TPU, DPU, etc.

[0169] It can be seen that each unit in the above device can be one or more processors (or processing circuits) configured to implement the above method, such as: CPU, GPU, NPU, TPU, DPU, microprocessor, DSP, ASIC, FPGA, or a combination of at least two of these processor forms.

[0170] In addition, all or part of the units in the above device can be integrated together, or can be implemented independently. In one implementation, these units are integrated together and implemented in the form of a SOC. The SOC may include at least one processor for implementing any of the above methods or implementing the functions of each unit of the device. The type of the at least one processor may be different, for example, including a CPU and an FPGA, a CPU and an artificial intelligence processor, a CPU and a GPU, etc.

[0171] The embodiment of the present application also provides a radiation source identification system, including a signal acquisition module and a signal processing module;

[0172] The signal acquisition module is used to acquire target electromagnetic signals;

[0173] The signal processing module is used to execute the radiation source identification method described in any of the above embodiments of this specification.

[0174] Another embodiment of the present application also provides an electronic device, see Figure 5 As shown, the electronic device includes:

[0175] Memory 200 and processor 210;

[0176] The memory 200 is connected to the processor 210 and is used to store programs;

[0177] The processor 210 is used to implement the radiation source identification method described in any of the above embodiments of this specification by running the program stored in the memory 200.

[0178] Specifically, the electronic device may further include: a bus, a communication interface 220 , an input device 230 and an output device 240 .

[0179] The processor 210, the memory 200, the communication interface 220, the input device 230 and the output device 240 are connected to each other via a bus.

[0180] A bus may include a pathway that transfers information between components of a computer system.

[0181] The processor 210 may be a general-purpose processor, such as a general-purpose central processing unit (CPU), a microprocessor, etc., or an application-specific integrated circuit (ASIC), or one or more integrated circuits for controlling the execution of the program of the scheme of the present invention. It may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0182] The processor 210 may include a main processor, and may also include a baseband chip, a modem, and the like.

[0183] The memory 200 stores a program for executing the technical solution of the present invention, and may also store an operating system and other key services. Specifically, the program may include a program code, and the program code includes a computer operation instruction. More specifically, the memory 200 may include a read-only memory (ROM), other types of static storage devices that can store static information and instructions, a random access memory (RAM), other types of dynamic storage devices that can store information and instructions, a disk storage, a flash, and the like.

[0184] The input device 230 may include a device for receiving data and information input by a user, such as a keyboard, a mouse, a camera, a scanner, a light pen, a voice input device, a touch screen, a pedometer, or a gravity sensor.

[0185] Output device 240 may include devices that allow information to be output to a user, such as a display screen, printer, speaker, etc.

[0186] The communication interface 220 may include any transceiver or the like to communicate with other devices or communication networks, such as Ethernet, Radio Access Network (RAN), Wireless Local Area Network (WLAN), etc.

[0187] The processor 210 executes the program stored in the memory 200 and calls other devices, which can be used to implement each step of any radiation source identification method provided in the above embodiments of the present application.

[0188] The method in the present application can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented by software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer programs or instructions, and when the computer programs or instructions are loaded and executed on a computer, the process or function described in the present application is executed in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, a network device, a user device, a core network device, an OAM or other programmable device.

[0189] The computer program product may be written in any combination of one or more programming languages ​​to write program codes for performing the operations of the embodiments of the present application, including object-oriented programming languages, such as Java, C++, etc., and conventional procedural programming languages, such as "C" language or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as an independent software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0190] The computer program or instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer program or instructions may be transmitted from one website, computer, server or data center to another website, computer, server or data center by wired or wireless means. The computer-readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server or data center that integrates one or more available media. The available medium may be a magnetic medium, such as a floppy disk, a hard disk, or a magnetic tape; it may also be an optical medium, such as a digital video disk; it may also be a semiconductor medium, such as a solid-state drive. The computer-readable storage medium may be a volatile or non-volatile storage medium, or may include both volatile and non-volatile types of storage media.

[0191] In addition, the embodiment of the present application may also be a storage medium on which a computer program is stored, and the computer program is executed by a processor in the steps of the radiation source identification method described in any of the above embodiments of this specification, and specifically the following steps may be implemented:

[0192] S201, acquiring pulse characteristics and time-frequency characteristics of a target electromagnetic signal; the pulse characteristics of the target electromagnetic signal are determined based on intra-pulse characteristics and / or inter-pulse characteristics of the target electromagnetic signal; the time-frequency characteristics of the target electromagnetic signal are determined based on time domain characteristics and frequency domain characteristics of the target electromagnetic signal;

[0193] S202: Determine an identification result of a radiation source corresponding to the target electromagnetic signal based on the pulse characteristics and time-frequency characteristics of the target electromagnetic signal.

[0194] For the aforementioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the order of the actions described, because according to the present application, some steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0195] It should be noted that each embodiment in this specification is described in a progressive manner, and each embodiment focuses on the differences from other embodiments, and the same or similar parts between the embodiments can be referred to each other. For the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0196] The steps in the methods of the embodiments of the present application can be adjusted in order, combined and deleted according to actual needs, and the technical features recorded in the embodiments can be replaced or combined. The modules and submodules in the devices and terminals in the embodiments of the present application can be combined, divided and deleted according to actual needs.

[0197] In the several embodiments provided in the present application, it should be understood that the disclosed terminals, devices and methods can be implemented in other ways. For example, the terminal embodiments described above are only schematic, for example, the division of modules or submodules is only a logical function division, and there may be other division methods in actual implementation, for example, multiple submodules or modules can be combined or integrated into another module, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or modules, which can be electrical, mechanical or other forms.

[0198] The modules or submodules described as separate components may or may not be physically separated, and the components of the modules or submodules may or may not be physical modules or submodules, that is, they may be located in one place, or they may be distributed on multiple network modules or submodules. Some or all of the modules or submodules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0199] In addition, each functional module or submodule in each embodiment of the present application may be integrated into one processing module, or each module or submodule may exist physically separately, or two or more modules or submodules may be integrated into one module. The above-mentioned integrated modules or submodules may be implemented in the form of hardware or in the form of software functional modules or submodules.

[0200] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0201] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be implemented directly by hardware, software units executed by a processor, or a combination of the two. The software units may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0202] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0203] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A radiation source identification method, characterized in that: include: Acquire the pulse characteristics and time-frequency characteristics of the target electromagnetic signal; the pulse characteristics of the target electromagnetic signal are determined based on the intra-pulse characteristics and / or inter-pulse characteristics of the target electromagnetic signal; the time-frequency characteristics of the target electromagnetic signal are determined based on the time domain characteristics and frequency domain characteristics of the target electromagnetic signal; Based on the pulse characteristics and time-frequency characteristics of the target electromagnetic signal, an identification result of a radiation source corresponding to the target electromagnetic signal is determined.

2. The method according to claim 1, characterized in that Acquire the pulse characteristics of the target electromagnetic signal, including: The sampling data of the target electromagnetic signal is input into a pre-trained feature extraction model, and the feature extraction model determines the primary features of the target electromagnetic signal based on the sampling data of the target electromagnetic signal, and determines the pulse features of the target electromagnetic signal based on the primary features.

3. The method according to claim 2, characterized in that The training method of the feature extraction model includes: Acquire first sample data; the first sample data includes sampling data and time-frequency characteristics of the first sample electromagnetic signal; Based on the first sample data and the first target loss function, performing unsupervised training on the first initial model to obtain the feature extraction model; Among them, the first target loss function is used to determine the model loss of the first initial model based on the quantitative characteristics, pulse characteristics and time-frequency characteristics of the first sample electromagnetic signal; the quantitative characteristics are used to characterize the quantization results of the primary characteristics of the first sample electromagnetic signal.

4. The method according to claim 3, characterized in that Determining the model loss of the first initial model based on the quantitative characteristics, pulse characteristics and time-frequency characteristics of the first sample electromagnetic signal includes: determining a first loss of the first initial model based on a quantized feature of the first sample electromagnetic signal, determining a second loss of the first initial model based on a quantized feature of the first sample electromagnetic signal and a pulse feature of the first sample electromagnetic signal, and determining a third loss of the first initial model based on a mapping feature of the first sample electromagnetic signal and a time-frequency feature of the first sample electromagnetic signal; the mapping feature of the first sample electromagnetic signal is obtained by performing feature mapping on the pulse feature of the first sample electromagnetic signal; A model loss of the first initial model is determined based on the first loss, the second loss, and the third loss of the first initial model.

5. The method according to any one of claims 1 to 4, characterized in that: The determining, based on the pulse characteristics and time-frequency characteristics of the target electromagnetic signal, an identification result of a radiation source corresponding to the target electromagnetic signal comprises: The pulse characteristics and time-frequency characteristics of the target electromagnetic signal are input into a pre-trained radiation source recognition model, and the radiation source recognition model determines the recognition result of the radiation source corresponding to the target electromagnetic signal based on the pulse characteristics and time-frequency characteristics of the target electromagnetic signal.

6. The method according to claim 5, characterized in that The radiation source identification model includes a coding model and a classification model; Inputting the pulse characteristics and time-frequency characteristics of the target electromagnetic signal into a pre-trained radiation source recognition model, and the radiation source recognition model determines the recognition result of the radiation source corresponding to the target electromagnetic signal according to the pulse characteristics and time-frequency characteristics of the target electromagnetic signal, including: Inputting the pulse characteristics of the target electromagnetic signal into the coding model, and the coding model determines the coding characteristics of the target electromagnetic signal according to the pulse characteristics of the target electromagnetic signal; The classification model determines the recognition result of the radiation source corresponding to the target electromagnetic signal according to the coding characteristics of the target electromagnetic signal and the time-frequency characteristics of the target electromagnetic signal.

7. The method according to claim 6, characterized in that The training method of the radiation source identification model includes: Acquire second sample data; the second sample data includes a pulse feature, a time-frequency feature, and a radiation source label corresponding to the second sample electromagnetic signal; Based on the second sample data and the second target loss function, supervised training is performed on the second initial model to obtain the radiation source identification model; Among them, the two-objective loss function is used to determine the first loss of the second initial model according to the radiation source identification result corresponding to the second sample electromagnetic signal and the radiation source label corresponding to the second sample electromagnetic signal, and determine the model loss of the second initial model based on the first loss of the second initial model.

8. The method according to claim 7, characterized in that The radiation source identification model also includes a decoding model, the input end of the decoding model is connected to the output end of the encoding model; the decoding model is used to determine the regression feature of the second sample electromagnetic signal according to the encoding feature of the second sample electromagnetic signal output by the encoding model; The two-objective loss function is also used to determine the second loss of the second initial model based on the regression characteristics of the second sample electromagnetic signal and the time-frequency characteristics of the second sample electromagnetic signal, and to determine the model loss of the second initial model based on the first loss and the second loss of the second initial model.

9. A radiation source identification device, characterized in that: include: A first data processing module is used to obtain the pulse characteristics and time-frequency characteristics of the target electromagnetic signal; the pulse characteristics of the target electromagnetic signal are determined based on the intra-pulse characteristics and / or inter-pulse characteristics of the target electromagnetic signal; the time-frequency characteristics of the target electromagnetic signal are determined based on the time domain characteristics and frequency domain characteristics of the target electromagnetic signal; The second data processing module is used to determine the identification result of the radiation source corresponding to the target electromagnetic signal based on the pulse characteristics and time-frequency characteristics of the target electromagnetic signal.

10. A radiation source identification system, characterized in that: It includes a signal acquisition module and a signal processing module; The signal acquisition module is used to acquire target electromagnetic signals; The signal processing module is used to execute the radiation source identification method as described in any one of claims 1 to 8.

11. An electronic device, characterized in that: include: Memory and processor; The memory is connected to the processor and is used to store programs; The processor is used to implement the radiation source identification method according to any one of claims 1 to 8 by running the program in the memory.

12. A computer program product, characterized in that The method comprises computer program instructions, which, when executed by a processor, enable the processor to perform the radiation source identification method according to any one of claims 1 to 8.

Images