Radiation source individual identification method and system based on continuous pulses

Abstract

The invention provides a radiation source individual identification method and system based on continuous pulses, and relates to the technical field of radar signal processing. The method comprises the following steps: acquiring intermediate frequency data of radar signals emitted by different radar radiation source individuals to obtain a plurality of intermediate frequency data, and respectively extracting a plurality of continuous pulse signals of the plurality of intermediate frequency data from the plurality of intermediate frequency data; fuzzy function features of the multiple continuous pulse signals are obtained through a fuzzy function and fused, and fuzzy function feature maps of the multiple pieces of intermediate frequency data are obtained; and after training the constructed radar individual identification model, identifying the plurality of fuzzy function feature maps to generate a plurality of radar radiation source individual identification results, and updating and replacing the radar individual identification model based on the identification results. By extracting the features of a plurality of continuous pulses, the signal time sequence information can be fully utilized to reduce noise interference, and the model is updated in time to improve the recognition rate and generalization ability of the model.

Description

Technical Field

[0001] The present invention mainly relates to the technical field of radar signal processing, and in particular to a method and system for identifying individual radiation sources based on continuous pulses. Background Art

[0002] Radar emitter individual identification is an important technology in the field of radar confrontation in recent years. The development and application of this technology is crucial to improving the efficiency and accuracy of reconnaissance systems in electronic warfare. Early radar emitter identification technology is based on matching and analyzing conventional parameters such as signal carrier frequency, pulse width and repetition period. However, with the development of radar technology, there are more and more radar emitters, and the signals are becoming more and more complex, making it increasingly difficult for traditional methods to work in this situation. Studies have found that the differences in individual radar emitters are caused by the inherent characteristics of the transmitter, and the impact of this characteristic on the signal is subtle, but also relatively stable. These subtle characteristics are called radar fingerprint features. Therefore, the existing radar emitter individual identification method extracts the fingerprint features of the radar signal and then uses deep learning to identify the individual radar emitter. However, this type of method usually uses a single intermediate frequency pulse for identification when identifying the target individual, which is greatly affected by the environment and noise, resulting in identification errors. In addition, if the changes in the environment are not updated to the model in a timely manner for a long time, there will be a large difference between the actual application environment and the training environment, resulting in a decrease in the model recognition rate. Summary of the invention

[0003] The technical problem to be solved by the present invention is to provide a method and system for identifying individual radiation sources based on continuous pulses in view of the deficiencies in the prior art.

[0004] The technical solution of the present invention to solve the above technical problems is as follows:

[0005] A method for identifying individual radiation sources based on continuous pulses comprises the following steps:

[0006] Collecting intermediate frequency data of radar signals emitted by different radar radiation sources to obtain intermediate frequency data of multiple different radar radiation sources, respectively performing pulse extraction on the multiple intermediate frequency data to obtain multiple continuous pulse signals corresponding to the multiple intermediate frequency data;

[0007] Extract features of multiple pulse signals respectively through fuzzy functions to obtain multiple fuzzy function features corresponding to the multiple intermediate frequency data, and fuse the multiple fuzzy function features respectively to obtain multiple fuzzy function feature graphs corresponding to the multiple intermediate frequency data;

[0008] Constructing a radar individual recognition model and training the radar individual recognition model through a pre-constructed radiation source training set, identifying a plurality of fuzzy function feature graphs through the trained radar individual recognition model to obtain a plurality of radar radiation source individual recognition results;

[0009] The trained radar individual recognition model is trained by using a plurality of radar radiation source individual recognition results to obtain a new radar individual recognition model, and the trained radar individual recognition model is replaced with the new radar individual recognition model. The target intermediate frequency data is recognized by using the new radar individual recognition model to obtain the target radar radiation source individual recognition result.

[0010] Another technical solution of the present invention to solve the above technical problems is as follows:

[0011] A radiation source individual identification system based on continuous pulses, comprising:

[0012] A signal acquisition module is used to collect intermediate frequency data of radar signals emitted by different radar radiation sources, obtain intermediate frequency data of multiple different radar radiation sources, and respectively perform pulse extraction on the multiple intermediate frequency data to obtain multiple continuous pulse signals corresponding to the multiple intermediate frequency data;

[0013] A feature extraction module is used to extract features of multiple pulse signals respectively through fuzzy functions to obtain multiple fuzzy function features corresponding to the multiple intermediate frequency data, and to fuse the multiple fuzzy function features respectively to obtain multiple fuzzy function feature graphs corresponding to the multiple intermediate frequency data;

[0014] A target recognition module is used to construct a radar individual recognition model and train the radar individual recognition model through a pre-constructed radiation source training set, and recognize multiple fuzzy function feature graphs through the trained radar individual recognition model to obtain multiple radar radiation source individual recognition results;

[0015] The model updating module is used to train the trained radar individual recognition model through multiple radar radiation source individual recognition results to obtain a new radar individual recognition model, and replace the trained radar individual recognition model with the new radar individual recognition model, and identify the target intermediate frequency data through the new radar individual recognition model to obtain the target radar radiation source individual recognition result.

[0016] The beneficial effects of the present invention are: the characteristic data of multiple continuous pulses are combined for identification, and after the identification is completed, the pulse data for identification is added to the training set according to the probability of the identification result, and the model is updated and trained in time to update the radar individual identification model used for individual identification. Compared with single pulse identification that only uses information at a certain moment, which is easily affected by the environment and noise, resulting in recognition errors, the use of multiple continuous pulses for identification can make full use of signal timing information, effectively avoid the problem that a single pulse is greatly affected by environmental noise, and reduce noise interference. In the case where there is a large difference between the actual application environment and the training environment, resulting in a decrease in the model recognition rate, timely updating of the current model can enable the model to be quickly migrated from the original model space to the current model space, avoiding the problem of gradual deviation of model recognition due to environmental changes, improving the reliability of the model in environmental changes, and improving the recognition rate and generalization ability of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A flow chart of a method for identifying individual radiation sources based on continuous pulses provided in an embodiment of the present invention;

[0018] Figure 2 A structural diagram of a method for identifying individual radiation sources based on continuous pulses provided by an embodiment of the present invention;

[0019] Figure 3 A flowchart of radar individual recognition model recognition processing provided by an embodiment of the present invention;

[0020] Figure 4 A schematic diagram of the convolution process of the radar individual recognition model provided by an embodiment of the present invention;

[0021] Figure 5 A structural diagram of a radar individual identification model provided by an embodiment of the present invention;

[0022] Figure 6 A structural diagram of a radar individual identification model provided by an embodiment of the present invention;

[0023] Figure 7 A module block diagram of a radiation source individual identification system based on continuous pulses provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0024] The principles and features of the present invention are described below in conjunction with the accompanying drawings. The examples given are only used to explain the present invention and are not used to limit the scope of the present invention.

[0025] like Figure 1-Figure 4 As shown, an embodiment of the present invention provides a method for identifying individual radiation sources based on continuous pulses, comprising the following steps:

[0026] Collecting intermediate frequency data of radar signals emitted by different radar radiation sources to obtain intermediate frequency data of multiple different radar radiation sources, respectively performing pulse extraction on the multiple intermediate frequency data to obtain multiple continuous pulse signals corresponding to the multiple intermediate frequency data;

[0027] Extract features of multiple pulse signals respectively through fuzzy functions to obtain multiple fuzzy function features corresponding to the multiple intermediate frequency data, and fuse the multiple fuzzy function features respectively to obtain multiple fuzzy function feature graphs corresponding to the multiple intermediate frequency data;

[0028] Constructing a radar individual recognition model and training the radar individual recognition model through a pre-constructed radiation source training set, identifying a plurality of fuzzy function feature graphs through the trained radar individual recognition model to obtain a plurality of radar radiation source individual recognition results;

[0029] The trained radar individual recognition model is trained by using a plurality of radar radiation source individual recognition results to obtain a new radar individual recognition model, and the trained radar individual recognition model is replaced with the new radar individual recognition model. The target intermediate frequency data is recognized by using the new radar individual recognition model to obtain the target radar radiation source individual recognition result.

[0030] It should be understood that after collecting multiple different intermediate frequency data, clear data is also included; a CNN network model is constructed based on the CNN network architecture as a radar individual recognition model. The CNN network model performs a convolution operation on the image matrix (i.e., the fuzzy function feature map) through the convolution kernel to obtain an output matrix, and uses the Softmax function to output a probability distribution to obtain the final recognition result (i.e., the radar radiator individual recognition result).

[0031] In an embodiment of the present invention, the use of characteristic data of multiple consecutive pulses can effectively avoid the problem of a single pulse being greatly affected by environmental noise, and the simultaneous processing of multiple pulses can also save the timing change information of the pulses. Timely updating of the model can avoid the problem of gradual deviation of the model due to environmental changes, thereby improving the reliability of the model in environmental changes.

[0032] Preferably, said respectively performing pulse extraction on the plurality of intermediate frequency data to obtain a plurality of continuous pulse signals corresponding to the plurality of intermediate frequency data comprises:

[0033] Perform Hilbert transform processing on the intermediate frequency data to obtain an envelope image, obtain a pulse leading edge and a pulse trailing edge from the envelope image according to an envelope amplitude threshold, obtain a pulse leading edge and a pulse trailing edge of the intermediate frequency data, calculate the pulse leading edge and the pulse trailing edge of the intermediate frequency data by a pulse width calculation expression, obtain multiple signal pulse widths of the intermediate frequency data, filter out multiple continuous pulse signals from the multiple signal pulse widths of the intermediate frequency data according to a filtering condition, and process all intermediate frequency data in this process to obtain multiple continuous pulse signals corresponding to the intermediate frequency data.

[0034] Preferably, the step of selecting a plurality of continuous pulse signals from a plurality of signal pulse widths of the intermediate frequency data according to a screening condition comprises:

[0035] Calculate the difference between the multiple signal pulse widths of the intermediate frequency data and the set acquisition signal pulse width to obtain multiple pulse width differences, delete the signal pulse widths greater than the preset difference among the multiple pulse width differences, calculate the multiple signal pulse widths that are not deleted by using a signal-to-noise ratio calculation expression to obtain multiple signal-to-noise ratios, and the signal-to-noise ratio calculation expression is:

[0036]

[0037] Among them, SNR is the signal-to-noise ratio, Psignal is the power of the signal pulse width, Pnoise is the power of the signal pulse width noise, the signal power is the square sum of the data of the 0.1us part of the pulse, and the noise power is the square sum of the data of the 0.1us before the rising edge;

[0038] The signal pulse widths corresponding to the plurality of signal-to-noise ratios are screened according to the set signal-to-noise ratio parameters to obtain a plurality of continuous pulse signals.

[0039] Specifically, the intermediate frequency data is subjected to Hilbert transform to obtain the envelope image of the intermediate frequency signal. The Hilbert transform expression is:

[0040]

[0041]

[0042] Among them, s(t) is the real part of the intermediate frequency data, is the imaginary part of the intermediate frequency data, and amp is the envelope of the intermediate frequency data;

[0043] The leading and trailing edges of the pulse are found through the envelope amplitude threshold. The difference between the trailing edge position of the pulse and the leading edge position of the pulse is divided by the sampling rate through the pulse width calculation expression to obtain the signal pulse width of the pulse signal. The difference between the signal pulse width and the pulse width of the collected signal is calculated. The pulse signals with a difference greater than 0.5us are removed as noise signals. The signal-to-noise ratio of the remaining pulse signals of the intermediate frequency data is then calculated to filter the pulse signals, leaving ten consecutive pulse signals whose signal-to-noise ratios meet the set conditions. The set condition is that eight of the signal-to-noise ratios of ten consecutive pulse signals meet the set signal-to-noise ratio threshold.

[0044] Preferably, before the step of extracting features from the plurality of pulse signals respectively by using a fuzzy function, the method further comprises:

[0045] Extracting signals from a plurality of pulse signals according to the pulse leading edge and pulse trailing edge of the intermediate frequency data to obtain a plurality of initial single pulse signals, and normalizing the plurality of initial single pulse signals to obtain a plurality of single pulse signals;

[0046] Feature extraction is performed on multiple single pulse signals to obtain the time delay difference and frequency shift difference of each single pulse signal, specifically:

[0047] According to the signal pulse width corresponding to any single pulse signal, different time delays are obtained from any single pulse signal, multiple time delays are obtained and the time delay difference is calculated; according to the set frequency shift parameter, different frequency shifts are obtained from any single pulse signal, multiple frequency shifts are obtained and the frequency shift difference is calculated, and so on, the time delay difference and frequency shift difference corresponding to each single pulse signal are obtained;

[0048] In this process, all pulse signals are processed to obtain the time delay difference and frequency shift difference of each single pulse signal corresponding to the plurality of intermediate frequency data.

[0049] It should be understood that the time delay is determined according to the pulse width of the pulse signal, with a value range of (0, T) and a value interval of 1 / fs, where T is the signal pulse width and fs is the signal sampling rate; the frequency shift value range is (-50MHz, 50MHz) and a value interval of 0.1Mhz.

[0050] Preferably, the extracting features of the plurality of pulse signals respectively by using a fuzzy function to obtain a plurality of fuzzy function features corresponding to the plurality of intermediate frequency data comprises:

[0051] The time delay difference and frequency shift difference of the single pulse signal are calculated by the fuzzy function to obtain the fuzzy function characteristics. In this process, all single pulse signals are processed to obtain multiple fuzzy function characteristics corresponding to the multiple intermediate frequency data. The fuzzy function is:

[0052]

[0053] Among them, A X (τ, ξ) is the fuzzy function feature, X(t) is a single pulse signal, τ is the time delay difference, ξ is the frequency shift difference, X * (t+τ) is the conjugate of X(t+τ), j is an imaginary unit, and t is time (i.e., the time domain variation of the pulse signal).

[0054] Specifically, when the radar signal is a single-frequency signal, the expression is:

[0055] x(t)=u(t)cos2πf 0 t,

[0056] Among them, f 0 is the carrier frequency, u(t) is the rectangular window function, t p is the window width;

[0057] For two echo signals with a time delay difference of τ and a frequency shift difference of ξ, the ambiguity function can be defined as follows based on the two-dimensional resolution of distance-velocity:

[0058]

[0059] When the transmitted signal x(t) is a single-frequency real signal x(t) = u(t)cos2πf 0 At t, the fuzzy function energy is concentrated at ξ=0,2f 0 ,-2f 0 nearby;

[0060] When ξ=0,τ=0, the fuzzy function reaches its maximum value.

[0061] |A x (τ,0)|=t p / 2;

[0062] When ξ=0, the zero frequency offset slice of the ambiguity function is:

[0063]

[0064] Its envelope is triangular and there are oscillations inside;

[0065] When ξ=2f 0 Or ξ=-2f 0 When , the fuzzy function is:

[0066]

[0067] Radar signals are complex signals (i.e., the multi-frequency signal generated during simulation), the fuzzy function can be expressed as:

[0068]

[0069] It should be understood that the distance is obtained by multiplying the time delay difference by half the speed of light, the speed is obtained by multiplying the frequency shift difference by half the wavelength, and the ambiguity function is obtained from the two-dimensional resolution of distance-speed; different time delay differences and frequency shift differences will have different resolutions.

[0070] In an embodiment of the present invention, the pulse intermediate frequency data is calculated by a fuzzy function, and the fuzzy function characteristics of the pulse are used as characteristic values ​​for different frequency shifts and time delays, and then the fuzzy function characteristics of the pulse are constructed into a channel feature representation of a feature map that can be recognized by a CNN network.

[0071] Preferably, the fusing of the plurality of fuzzy function features to obtain a plurality of fuzzy function feature graphs corresponding to the intermediate frequency data comprises:

[0072] Any fuzzy function feature is used as any channel feature of the pre-constructed multi-channel feature map. Similarly, multiple fuzzy function features of the intermediate frequency data are used as corresponding channel features of the multi-channel feature map according to the time sequence of the collected pulse signals. In this way, all fuzzy function features are processed to obtain multiple fuzzy function feature maps corresponding to the intermediate frequency data.

[0073] It should be understood that collecting intermediate frequency data is to collect multiple pulse signals, so there are corresponding different pulse signal collection times (ie, the arrival times of the pulse signals).

[0074] Specifically, since the fuzzy function value (i.e., the fuzzy function feature) is a related calculation quantity of time delay and frequency shift, the fuzzy function feature data of a pulse signal is represented as a three-dimensional data, which can be represented by a two-dimensional table, with the horizontal axis representing the change of time delay, the vertical axis representing the change of frequency shift, and the recorded value being the fuzzy function value under the current frequency shift and time delay. Then, the fuzzy function feature data of a pulse signal can be represented by a channel of a feature graph, and the fuzzy function features of multiple pulses can be represented by a feature graph containing multiple channels. That is, the input is a feature graph, and each channel of the feature graph represents the fuzzy function feature of a pulse signal, and there are N channels for N consecutive pulses.

[0075] In an embodiment of the present invention, after calculating the features of each pulse, the fuzzy function features of each pulse are obtained, and the fuzzy function features of ten pulse signals are fused in the order of the arrival time of the pulses to obtain a feature map with ten channels as input data, which is input into the CNN model for recognition.

[0076] Preferably, if Figure 5 and Figure 6 As shown, the construction of the radar individual recognition model includes:

[0077] Constructing an input layer, multiple convolutional layers, multiple pooling layers, multiple serial connection layers and an output layer, and connecting the output of the convolutional layer and the input of the pooling layer to construct multiple feature extraction units;

[0078] The output of the input layer is connected to the input of the first feature extraction unit, the output of the first feature extraction unit is connected to the input of the second feature extraction unit, and so on, until the output of the last feature extraction unit is connected to the input of the first connection layer in multiple series connection layers, and the output of the last connection layer is connected to the input of the output layer, to obtain a radar individual recognition model.

[0079] The output layer includes a Softmax normalized exponential function, which performs probability distribution calculation on the features output by the last connection layer to obtain the probability values ​​of the samples and the labels in the model. The prediction result (i.e., the individual identification result of the radar emitter) is the label number and the identification probability.

[0080] Specifically, the structure of the constructed CNN network (i.e., radar individual recognition model) is to perform multiple convolutions and pooling, and then connect through the connection layer to finally obtain the output. Since the data input to the network is the fuzzy function feature data of ten pulse signals, the number of input channels of the first convolution layer is designed to be 10. A total of 5 convolution layers and 5 pooling layers are designed to extract deep features of the data, where the convolution kernel size of the convolution layer is 3×3, and the pooling kernel of the pooling layer is 2. After completing all convolutions and pooling, three connection layers are designed to connect the extracted deep features, and finally the softmax function is used to output the distribution probability of the prediction results to obtain the recognition probability of these ten pulses. Setting multiple connection layers enables the model to record complex feature relationships and improve the recognition effect and generalization ability of the model.

[0081] In the embodiment of the present invention, the convolutional neural network (CNN) is a feedforward neural network with a deep structure. The convolutional neural network reduces network parameters and reduces network training time through convolution and pooling operations. Since the convolutional neural network can classify the input data model in a translation-invariant manner, the image is fed into the network input layer, and after convolution and pooling, the specific features of the image are output, and the extracted features have translation invariance, scaling invariance and rotation invariance. Therefore, constructing a convolutional neural network as a radar individual recognition model can accurately identify the radiation source individual in the field of processing radar signals.

[0082] Preferably, the radar emitter individual identification result includes an identification result label and an identification probability;

[0083] The step of training the trained radar individual recognition model by using the individual recognition results of the plurality of radar radiation sources to obtain a new radar individual recognition model, and replacing the trained radar individual recognition model with the new radar individual recognition model comprises:

[0084] Add multiple recognition result labels whose recognition probabilities meet the preset probabilities to the radiation source training set to obtain a new radiation source training set, train the trained radar individual recognition model through the new radiation source training set to obtain a new radar individual recognition model, test the new radar individual recognition model through a pre-constructed radiation source test set to obtain a test probability, and if the test probability meets a preset test threshold, replace the trained radar individual recognition model with the new radar individual recognition model.

[0085] Specifically, the pulse intermediate frequency data with a recognition rate greater than 65% (or a recognition rate in the range of 65%-90%) is saved. When the accumulated number of pulse signals reaches 100, the pulse intermediate frequency data and the recognition result label are added to the radiation source training set to obtain a new radiation source training set and train the radar individual recognition model currently used for radiation source individual recognition. After the training is completed, the new model is tested with the test set. If the recognition accuracy of the new model is above 80%, the trained new model is updated to the recognition model.

[0086] It should be understood that the process of training the radar individual recognition model by using the pre-constructed radiation source training set is the same as the process of training the radar individual recognition model currently used for radiation source individual recognition (i.e., the trained radar individual recognition model) by using the new radiation source training set. After the test probability meets the preset test threshold, the new radar individual recognition model is tested using the pre-constructed radiation source verification set, and when the test passes, the trained radar individual recognition model is replaced with the new radar individual recognition model.

[0087] In the embodiment of the present invention, since the new model can adapt to the changes brought about by the environment and improve the individual recognition rate of radar emitters in the actual environment, the model is updated.

[0088] like Figure 7 As shown, an embodiment of the present invention provides a radiation source individual identification system based on continuous pulses, comprising:

[0089] A signal acquisition module is used to collect intermediate frequency data of radar signals emitted by different radar radiation sources, obtain intermediate frequency data of multiple different radar radiation sources, and respectively perform pulse extraction on the multiple intermediate frequency data to obtain multiple continuous pulse signals corresponding to the multiple intermediate frequency data;

[0090] A feature extraction module is used to extract features of multiple pulse signals respectively through fuzzy functions to obtain multiple fuzzy function features corresponding to the multiple intermediate frequency data, and to fuse the multiple fuzzy function features respectively to obtain multiple fuzzy function feature graphs corresponding to the multiple intermediate frequency data;

[0091] A target recognition module is used to construct a radar individual recognition model and train the radar individual recognition model through a pre-constructed radiation source training set, and recognize multiple fuzzy function feature graphs through the trained radar individual recognition model to obtain multiple radar radiation source individual recognition results;

[0092] The model updating module is used to train the trained radar individual recognition model through multiple radar radiation source individual recognition results to obtain a new radar individual recognition model, and replace the trained radar individual recognition model with the new radar individual recognition model, and identify the target intermediate frequency data through the new radar individual recognition model to obtain the target radar radiation source individual recognition result.

[0093] 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.

[0094] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems and modules described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0095] In the several embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are only schematic, for example, the division of modules is only a logical function division, and there may be other division methods in actual implementation, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed.

[0096] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules, that is, they may be located in one place or distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the embodiments of the present invention.

[0097] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for identifying individual radiation sources based on continuous pulses, characterized in that: The steps include: Collecting intermediate frequency data of radar signals emitted by different radar radiation sources to obtain intermediate frequency data of multiple different radar radiation sources, respectively performing pulse extraction on the multiple intermediate frequency data to obtain multiple continuous pulse signals corresponding to the multiple intermediate frequency data; Extract features of multiple pulse signals respectively through fuzzy functions to obtain multiple fuzzy function features corresponding to the multiple intermediate frequency data, and fuse the multiple fuzzy function features respectively to obtain multiple fuzzy function feature graphs corresponding to the multiple intermediate frequency data; Constructing a radar individual recognition model and training the radar individual recognition model through a pre-constructed radiation source training set, identifying a plurality of fuzzy function feature graphs through the trained radar individual recognition model to obtain a plurality of radar radiation source individual recognition results; The trained radar individual recognition model is trained by using a plurality of radar radiation source individual recognition results to obtain a new radar individual recognition model, and the trained radar individual recognition model is replaced with the new radar individual recognition model. The target intermediate frequency data is recognized by using the new radar individual recognition model to obtain the target radar radiation source individual recognition result.

2. The method for identifying individual radiation sources according to claim 1, characterized in that: The step of respectively performing pulse extraction on the plurality of intermediate frequency data to obtain a plurality of continuous pulse signals corresponding to the plurality of intermediate frequency data comprises: Perform Hilbert transform processing on the intermediate frequency data to obtain an envelope image, obtain a pulse leading edge and a pulse trailing edge from the envelope image according to an envelope amplitude threshold, obtain a pulse leading edge and a pulse trailing edge of the intermediate frequency data, calculate the pulse leading edge and the pulse trailing edge of the intermediate frequency data by a pulse width calculation expression, obtain multiple signal pulse widths of the intermediate frequency data, filter out multiple continuous pulse signals from the multiple signal pulse widths of the intermediate frequency data according to a filtering condition, and process all intermediate frequency data in this process to obtain multiple continuous pulse signals corresponding to the intermediate frequency data.

3. The method for identifying individual radiation sources according to claim 2, characterized in that: The step of selecting a plurality of continuous pulse signals from a plurality of signal pulse widths of the intermediate frequency data according to the selection condition comprises: The differences between the multiple signal pulse widths of the intermediate frequency data and the set acquisition signal pulse width are calculated to obtain multiple pulse width differences, the signal pulse widths greater than the preset difference among the multiple pulse width differences are deleted, the multiple signal pulse widths that are not deleted are calculated by a signal-to-noise ratio calculation expression to obtain multiple signal-to-noise ratios, and the signal pulse widths corresponding to the multiple signal-to-noise ratios are screened according to the set signal-to-noise ratio parameters to obtain multiple continuous pulse signals.

4. The method for identifying individual radiation sources according to claim 2, characterized in that: Before the step of extracting features from the plurality of pulse signals respectively by using the fuzzy function, the method further comprises: Extracting signals from a plurality of pulse signals according to the pulse leading edge and pulse trailing edge of the intermediate frequency data to obtain a plurality of initial single pulse signals, and normalizing the plurality of initial single pulse signals to obtain a plurality of single pulse signals; Feature extraction is performed on multiple single pulse signals to obtain the time delay difference and frequency shift difference of each single pulse signal, specifically: According to the signal pulse width corresponding to any single pulse signal, different time delays are obtained from any single pulse signal, multiple time delays are obtained and the time delay difference is calculated; according to the set frequency shift parameter, different frequency shifts are obtained from any single pulse signal, multiple frequency shifts are obtained and the frequency shift difference is calculated, and so on, the time delay difference and frequency shift difference corresponding to each single pulse signal are obtained; In this process, all pulse signals are processed to obtain the time delay difference and frequency shift difference of each single pulse signal corresponding to the plurality of intermediate frequency data.

5. The method for identifying individual radiation sources according to claim 4, characterized in that: The method of extracting features of the plurality of pulse signals respectively by using a fuzzy function to obtain a plurality of fuzzy function features corresponding to the plurality of intermediate frequency data includes: The time delay difference and frequency shift difference of the single pulse signal are calculated by the fuzzy function to obtain the fuzzy function characteristics. In this process, all single pulse signals are processed to obtain multiple fuzzy function characteristics corresponding to the multiple intermediate frequency data. The fuzzy function is: Among them, A x (τ, ξ) is the fuzzy function characteristic, X(t) is a single pulse signal, τ is the time delay difference, and ξ is the frequency shift difference.

6. The method for identifying individual radiation sources according to claim 1, characterized in that: The fusion processing is performed on the multiple fuzzy function features respectively to obtain the fuzzy function feature graphs corresponding to the multiple intermediate frequency data, including: Any fuzzy function feature is used as any channel feature of the pre-constructed multi-channel feature map. Similarly, multiple fuzzy function features of the intermediate frequency data are used as corresponding channel features of the multi-channel feature map according to the time sequence of the collected pulse signals. In this way, all fuzzy function features are processed to obtain multiple fuzzy function feature maps corresponding to the intermediate frequency data.

7. The method for identifying individual radiation sources according to claim 1, characterized in that: The step of constructing a radar individual recognition model comprises: Constructing an input layer, multiple convolutional layers, multiple pooling layers, multiple serial connection layers and an output layer, and connecting the output of the convolutional layer and the input of the pooling layer to construct multiple feature extraction units; The output of the input layer is connected to the input of the first feature extraction unit, the output of the first feature extraction unit is connected to the input of the second feature extraction unit, and so on, until the output of the last feature extraction unit is connected to the input of the first connection layer in multiple series connection layers, and the output of the last connection layer is connected to the input of the output layer, to obtain a radar individual recognition model.

8. The method for identifying individual radiation sources according to claim 1, characterized in that: The radar radiation source individual identification result includes an identification result label and an identification probability; The step of training the trained radar individual recognition model by using the individual recognition results of the plurality of radar radiation sources to obtain a new radar individual recognition model, and replacing the trained radar individual recognition model with the new radar individual recognition model comprises: Add multiple recognition result labels whose recognition probabilities meet the preset probabilities to the radiation source training set to obtain a new radiation source training set, train the trained radar individual recognition model through the new radiation source training set to obtain a new radar individual recognition model, test the new radar individual recognition model through a pre-constructed radiation source test set to obtain a test probability, and if the test probability meets a preset test threshold, replace the trained radar individual recognition model with the new radar individual recognition model.

9. A radiation source individual identification system based on continuous pulses, characterized in that: include: A signal acquisition module is used to collect intermediate frequency data of radar signals emitted by different radar radiation sources, obtain intermediate frequency data of multiple different radar radiation sources, and respectively perform pulse extraction on the multiple intermediate frequency data to obtain multiple continuous pulse signals corresponding to the multiple intermediate frequency data; A feature extraction module is used to extract features of multiple pulse signals respectively through fuzzy functions to obtain multiple fuzzy function features corresponding to the multiple intermediate frequency data, and to fuse the multiple fuzzy function features respectively to obtain multiple fuzzy function feature graphs corresponding to the multiple intermediate frequency data; A target recognition module is used to construct a radar individual recognition model and train the radar individual recognition model through a pre-constructed radiation source training set, and recognize multiple fuzzy function feature graphs through the trained radar individual recognition model to obtain multiple radar radiation source individual recognition results; The model updating module is used to train the trained radar individual recognition model through multiple radar radiation source individual recognition results to obtain a new radar individual recognition model, and replace the trained radar individual recognition model with the new radar individual recognition model, and identify the target intermediate frequency data through the new radar individual recognition model to obtain the target radar radiation source individual recognition result.

10. The radiation source individual identification system according to claim 9, characterized in that: In the signal acquisition module, pulse extraction is performed on the plurality of intermediate frequency data respectively to obtain a plurality of continuous pulse signals corresponding to the plurality of intermediate frequency data, including: Perform Hilbert transform processing on the intermediate frequency data to obtain an envelope image, obtain a pulse leading edge and a pulse trailing edge from the envelope image according to an envelope amplitude threshold, obtain a pulse leading edge and a pulse trailing edge of the intermediate frequency data, calculate the pulse leading edge and the pulse trailing edge of the intermediate frequency data by a pulse width calculation expression, obtain multiple signal pulse widths of the intermediate frequency data, filter out multiple continuous pulse signals from the multiple signal pulse widths of the intermediate frequency data according to a filtering condition, and process all intermediate frequency data in this process to obtain multiple continuous pulse signals corresponding to the intermediate frequency data.

Images