Radar signal intelligent sorting method and device in complex electromagnetic environment and electronic equipment

Abstract

The invention discloses a radar signal intelligent sorting method in a complex electromagnetic environment, and relates to the technical field of radar signal processing, and the method comprises the steps: receiving a pulse signal flow formed by the random overlapping of pulse signals from different radar radiation sources, and the pulse signal flow comprises a plurality of groups of radar pulse sequences; preprocessing each group of radar pulse sequences to generate a signal time-frequency graph of a preset size; and sequentially inputting the plurality of signal time-frequency diagrams into a radar signal sorting model, and classifying the obtained reconstructed signals to obtain a radar signal sorting result. The method can be used for detection and classification of radar signals in a complex electromagnetic environment, and especially has a good classification effect on signals with a low signal-to-noise ratio.

Description

Technical Field

[0001] The present invention belongs to the technical field of radar signal processing, and particularly relates to a method, device and electronic equipment for intelligent sorting of radar signals in a complex electromagnetic environment. Background Art

[0002] Different from the traditional radar signal sorting method based on DPW (Pulse Description Word), the radar signal sorting method based on a convolutional denoising autoencoder utilizes deep and representative features learned from radar signal data, overcoming the limitation of the traditional radar signal sorting method in difficult signal feature extraction in a complex electromagnetic environment. In addition, due to the low signal-to-noise ratio of intercepted radar signals, the sorting and recognition accuracy of the traditional radar signal sorting method is not high, while the convolutional denoising autoencoder has a certain anti-noise ability and can accurately sort radar signals in a noisy environment.

[0003] The radar signal sorting method based on a convolutional denoising autoencoder essentially constructs an unsupervised neural network model. Specifically, the encoder network encodes the input features using an encoding function, maps the input features into feature representations in a low-dimensional space, and obtains the encoded hidden representations; the decoder network is the reverse process, reconstructing the original data in the high-dimensional feature space from the low-dimensional feature representations, that is, using a decoding function to map the hidden representations back to the original space.

[0004] In the prior art, Wen Qian et al. proposed a method for sorting radar signals based on a deep belief neural network. The deep belief network is stacked by multiple restricted Boltzmann machines. First, features are extracted through unsupervised pre-training to determine the initial parameters, and then fine-tuned using the BP algorithm. Hong Shujie et al. proposed a radar signal sorting method combining a convolutional neural network and an encoder. This method first encodes the arrival time of the pulse sequence into a binary vector and inputs it into the convolutional denoising autoencoder to let it learn the time pattern of the target pulse sequence. After the network is trained, it can sort the mixed pulse sequence to extract the target sequence.

[0005] However, the existing radar signal sorting methods often adopt a manual way for local feature extraction when dealing with increasingly complex electromagnetic signals, and the convergence speed is slow. Therefore, it is difficult to sort a large amount of radar data; in addition, under the condition of low signal-to-noise ratio, the sorting accuracy of the existing methods for signals is low. Summary of the Invention

[0006] In order to solve the above problems existing in the prior art, the present invention provides a method, device and electronic equipment for intelligent sorting of radar signals in a complex electromagnetic environment. The technical problems to be solved by the present invention are realized through the following technical solutions:

[0007] In a first aspect, the present invention provides an intelligent sorting method for radar signals in a complex electromagnetic environment, including:

[0008] Receiving a pulse signal stream formed by random overlapping of pulse signals from different radar radiation sources, where the pulse signal stream contains multiple groups of radar pulse sequences;

[0009] Preprocessing each group of radar pulse sequences to generate a signal time-frequency diagram of a preset size;

[0010] Sequentially inputting multiple signal time-frequency diagrams into a radar signal sorting model, classifying the obtained reconstructed signals, and obtaining a radar signal sorting result.

[0011] In an embodiment of the present invention, the step of preprocessing each group of radar pulse sequences to generate a signal time-frequency diagram of a preset size includes:

[0012] Calculating relevant parameters of each group of radar pulse sequences;

[0013] After normalizing the radar pulse sequences, performing short-time Fourier transform to encode the relevant parameters, and generating a two-dimensional signal time-frequency diagram of a preset size through downsampling processing; the relevant parameters include the pulse carrier frequency RF, pulse width PW, and pulse repetition interval PRI of the radar pulse sequences.

[0014] In an embodiment of the present invention, the radar signal sorting model includes an encoder, and the encoder includes a convolutional module and a flattening module; wherein,

[0015] The convolutional module is used to extract the feature vector of the radar pulse sequence, including a convolutional layer Conv1, a first pooling layer, a convolutional layer Conv2, a second pooling layer, a convolutional layer Conv3, and a third pooling layer connected in sequence, and the flattening module is used to flatten the feature map vector into a one-dimensional vector.

[0016] In an embodiment of the present invention, the radar signal sorting model further includes a decoder, and the decoder includes a reconstruction module and a transposed convolutional module; wherein,

[0017] The reconstruction module is used to reconstruct the dimension of the one-dimensional vector, and the transposed convolutional module is used to restore the spatial size and number of channels of the one-dimensional vector after reconstruction to obtain a reconstructed signal. The transposed convolutional module includes a first upsampling layer, a first transposed convolutional layer, a second upsampling layer, a second transposed convolutional layer, a third upsampling layer, and a third transposed convolutional layer connected in sequence.

[0018] In an embodiment of the present invention, it further includes a fully connected layer, which is used to map the probability distribution of the reconstructed signal from each radiation source by using the Softmax activation function to obtain a radar signal sorting result.

[0019] In one embodiment of the present invention, the radar signal sorting model is obtained based on the mean square error loss function.

[0020] In one embodiment of the present invention, the mean square error loss function is expressed as:

[0021]

[0022] In the formula, N represents the number of sample signals, x i represents the sorting result of the i-th sample signal, represents the label of the i-th sample signal.

[0023] In a second aspect, the present invention further provides a radar signal intelligent sorting device under a complex electromagnetic environment, including:

[0024] A receiving module, configured to receive a pulse signal stream formed by random overlapping of pulse signals from different radar radiation sources, where the pulse signal stream contains multiple groups of radar pulse sequences;

[0025] A preprocessing module, configured to preprocess each group of radar pulse sequences to generate a signal time-frequency diagram with a preset size;

[0026] A sorting module, configured to sequentially input multiple signal time-frequency diagrams into the radar signal sorting model, and classify the obtained reconstructed signals to obtain a radar signal sorting result.

[0027] In a third aspect, the present invention further provides an electronic device, including a processor, a communication interface, a memory, and a communication bus. Among them, the processor, the communication interface, and the memory complete communication with each other through the communication bus;

[0028] The memory is used to store a computer program;

[0029] The processor, when executing the program stored in the memory, implements the method steps described in the first aspect.

[0030] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0031] (1) In the radar signal intelligent sorting method provided by the present invention under a complex electromagnetic environment, the radar signal sorting model has strong feature extraction ability for radar signals, can learn deep and representative features from a large amount of radar signal data, and overcomes the limitations of manually extracting features in traditional methods. For example, for the complex intra-pulse modulation features in new radar signals, the present invention can automatically capture and learn them, while traditional methods based on pulse parameters are difficult to extract these complex features.

[0032] (2) By learning radar signals through the radar signal sorting model of the present invention, the input time-frequency diagram can be converted into a low-dimensional feature representation, thereby removing redundant information and retaining key features. For example, high-dimensional time-domain or frequency-domain radar signal data is compressed into a more compact feature space while retaining the class discrimination information of the signals. In the subsequent sorting process, this compact feature representation can be used to more quickly and accurately determine the class of the signals, improving the sorting efficiency.

[0033] (3) Through the calculation of the multi-layer perception mechanism, the sorting performance of the radar signal sorting model is less affected by the signal-to-noise ratio. The convolutional neural network can learn different feature patterns of signals and noises during the training process. The filters in its convolutional layer can suppress the interference of noises to a certain extent while extracting signal features. For example, when radar signals are interfered by common noises such as Gaussian white noise, through multi-layer convolution and pooling operations, the network can focus on the feature structure of the signals themselves, reducing the impact of noises on signal sorting, thereby improving the sorting reliability of signals in a noisy environment.

[0034] The following will further elaborate on the present invention in conjunction with the accompanying drawings and embodiments. Description of the Drawings

[0035] Figure 1 is a flowchart of an intelligent radar signal sorting method in a complex electromagnetic environment provided by an embodiment of the present invention;

[0036] Figure 2 is a schematic diagram of radar signal sorting provided by an embodiment of the present invention;

[0037] Figure 3 is a schematic diagram of the structure of a radar signal sorting model provided by an embodiment of the present invention;

[0038] Figure 4 is a relationship curve graph between the sorting accuracy rate and the signal-to-noise ratio of different radar signal sorting methods;

[0039] Figure 5 is a relationship curve graph between the sorting accuracy rate and different pulse loss rates of different radar signal sorting methods;

[0040] Figure 6 is a relationship curve graph between the sorting accuracy rate and different pulse stagger rates of different radar signal sorting methods;

[0041] Figure 7 is a schematic diagram of the structure of an intelligent radar signal sorting device in a complex electromagnetic environment provided by an embodiment of the present invention;

[0042] Figure 8 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. Detailed Embodiments

[0043] The following further describes the present invention in detail with reference to specific embodiments, but the embodiments of the present invention are not limited thereto.

[0044] Figure 1 FIG. is a flowchart of a method for intelligent sorting of radar signals in a complex electromagnetic environment provided by an embodiment of the present invention. An embodiment of the present invention provides a method for intelligent sorting of radar signals in a complex electromagnetic environment, including:

[0045] S1. Receive a pulse signal stream formed by random overlapping of pulse signals from different radar radiation sources, where the pulse signal stream contains multiple groups of radar pulse sequences;

[0046] S2. Preprocess each group of radar pulse sequences to generate a signal time-frequency diagram of a preset size;

[0047] S3. Input multiple signal time-frequency diagrams into a radar signal sorting model in sequence, and classify the obtained reconstructed signals to obtain a radar signal sorting result.

[0048] Figure 2 FIG. is a schematic diagram of radar signal sorting provided by an embodiment of the present invention. As Figure 2 shown, in a complex electromagnetic environment, the radar detection receiver receives a mixed radar signal formed by random overlapping of pulse signal streams from different radar radiation sources. Radar signal sorting refers to the process of separating signals from different radar radiation sources in the mixed radar signal through feature extraction and classification.

[0049] Optionally, in step S1, the step of preprocessing each group of radar pulse sequences to generate a signal time-frequency diagram of a preset size includes:

[0050] S101. Calculate the relevant parameters of each group of radar pulse sequences;

[0051] S102. After normalizing the radar pulse sequences, perform short-time Fourier transform to encode the relevant parameters, and generate a two-dimensional signal time-frequency diagram of a preset size through downsampling processing; the relevant parameters include the pulse carrier frequency RF, pulse width PW, and pulse repetition interval PRI of the radar pulse sequences.

[0052] The normalization process refers to performing amplitude normalization on the collected radar pulse sequences to map the signal amplitude to a specific interval. The method adopted in the present invention is to map the signal amplitude to the interval [0, 1]. Specifically, the original radar pulse sequence is denoted as A raw (t), then the normalized radar pulse sequence is expressed as:

[0053]

[0054] In the formula, A min and A max respectively represent the maximum value and the minimum value in the original radar pulse sequence, and t represents time.

[0055] In this embodiment, the RF (pulse carrier frequency), PW (pulse width), and PRI (pulse repetition interval) parameters of each radar pulse sequence are calculated, and the time-frequency diagram of the radar pulse sequence is obtained through discrete short-time Fourier transform. Further, after downsampling processing, a two-dimensional signal time-frequency diagram with a preset size is obtained. Among them, the discrete short-time Fourier transform is expressed as:

[0056]

[0057] In the formula, x[n] represents the normalized radar pulse sequence, w[n - m] represents the window function centered at the time point m, and j is the imaginary unit.

[0058] In this embodiment, the radar signal sorting model includes an encoder, and the encoder includes a convolution module and a flatten module; among them,

[0059] The convolution module is used to extract the feature vector of the radar pulse sequence, including a convolution layer Conv1, a first pooling layer, a convolution layer Conv2, a second pooling layer, a convolution layer Conv3, and a third pooling layer connected in sequence. The flatten module is used to flatten the feature map vector into a one-dimensional vector.

[0060] Further, the radar signal sorting model further includes a decoder, and the decoder includes a reconstruction module and a transposed convolution module; among them,

[0061] The reconstruction module is used to reconstruct the dimension of the one-dimensional vector, and the transposed convolution module is used to restore the spatial size and the number of channels of the one-dimensional vector after reconstruction to obtain a reconstructed signal. The transposed convolution module includes a first upsampling layer, a first transposed convolution layer, a second upsampling layer, a second transposed convolution layer, a third upsampling layer, and a third transposed convolution layer connected in sequence.

[0062] Optionally, the above radar signal sorting model further includes a fully connected layer, which is used to map the reconstructed signal to the probability distribution from each radiation source by using the Softmax activation function to obtain the radar signal sorting result.

[0063] Specifically, in the training process of the radar signal sorting model, first, the convolution kernel weights of each convolution layer in the encoder and the transposed weights of each transposed convolution layer in the decoder are initialized by using the normal distribution N(0, σ 2 ), and the bias term is initialized to 0, and σ takes 0.01.

[0064] In this embodiment, the radar signal sorting model is obtained based on the mean square error loss function, which is expressed as:

[0065]

[0066] Wherein, N represents the number of sample signals, and x i represents the sorting result of the i-th sample signal, and y_i represents the label of the i-th sample signal.

[0067] It should be noted that during the training process, the Adam optimizer is selected to adjust the learning rate of each parameter, and its parameter update formula cluster is as follows:

[0068] m t = β1 * m t-1 + (1 - β1) * g t

[0069] v t = β2 * v t-1 + (1 - β2) * g t 2

[0070]

[0071] Wherein, t represents the training step, m t and v t respectively represent the first-order moment estimation and the second-order moment estimation, β1 and β2 are decay coefficients, g t is the gradient, α is the learning rate, ξ is a very small value, and θ t is the model parameter.

[0072] That is to say, in each round of training process, a batch of sample signals will be input. After each batch is processed, the Adam optimizer will estimate the parameters therein. For example, if there are m groups of data in one round, then one epoch needs to perform m / batch_size steps.

[0073] Next, the radar signal intelligent sorting method provided by the present invention under a complex electromagnetic environment will be further described through simulation.

[0074] Specifically, the software platform for the simulation experiment is: Windows 11 operating system, pycharm 2022, and pytorch 2.5. To improve the operation efficiency, the training model is processed using CUDA acceleration, and the version number is 12.1.

[0075] First, simulate and generate the pulse sequences of 5 radars. Each radar generates 10,000 pulse sequences. Sort the pulse sequences of each of the 5 radars according to the arrival time sequence to construct an aliased pulse sequence, and then preprocess the aliased pulse sequence according to the above step S2. Among them, the relevant parameters of the 5 radars are shown in Table 1:

[0076] Table 1

[0077]

[0078] The variation range of the signal-to-noise ratio of the pulse sequence is [-15dB, 15dB]. The traditional SDIF (Sequence Difference Histogram) method and the PRI (Pulse Repetition Interval) transformation method are compared with the present invention.

[0079] Figure 4 is a relationship curve graph between the sorting correct rate of different radar signal sorting methods and the signal-to-noise ratio. From Figure 4 It can be seen that when the signal-to-noise ratio is -15dB, the performance of the present invention is improved by 15% compared with the PRI transformation method and by 28% compared with the SDIF sorting method. Obviously, when the signal-to-noise ratio is low, the intelligent radar signal sorting method provided by the present invention has a significant improvement in the average correct rate of signal sorting compared with the traditional PRI transformation method and SDIF algorithm. When the signal-to-noise ratio is large, the improvement in the average correct rate of sorting is limited.

[0080] Furthermore, the radar signal sorting model provided by the present invention is fine-tuned. After the operation of the first fully connected layer, a dropout layer is added. By setting different dropout parameters, the output is randomly set to zero according to the probability to simulate the situation after pulse loss.

[0081] Figure 5 is a relationship curve graph between the sorting correct rate of different radar signal sorting methods and different pulse loss rates. From Figure 5 It can be seen that when the pulse loss rate is 30%, the performance of the present invention is improved by 12% compared with the PRI transformation method and by 40% compared with the SDIF algorithm. However, as the pulse loss rate continues to increase, the correct rate of radar signal sorting also continues to decline. However, even when the pulse loss rate is 30%, the average sorting correct rate of the present invention still exceeds 60%. It can be seen that the present invention can well improve the low correct rate of radar signal sorting under the condition of a large number of pulse losses.

[0082] In order to test the influence of the pulse jitter rate on the sorting performance of radar emitter signals, in this embodiment, the pulse repetition interval is set to jitter between 1% and 20% around the central value. PRI can be described by the following formula:

[0083]

[0084] Among them, PRI is the center value of the PRI of the jitter radar signal, δ is the specific PRI value, which is uniformly distributed within the range of [-T, T], and γ is the PRI jitter rate.

[0085] Input the above signal into the fine-tuned radar signal sorting model for training, and then input the test set to obtain Figure 6 The relationship curve graph of the sorting accuracy rate of different radar signal sorting methods and different pulse stagger rates as shown. From Figure 6 It can be seen that when the pulse stagger rate of the signal is 20%, the performance of the present invention is improved by 51% compared with the traditional SDIF sorting method, and is improved by 22% compared with the PRI transformation method. Its average sorting accuracy rate is still greater than 70%, and it still has a certain signal sorting ability. However, as the pulse jitter rate continues to increase, the sorting accuracy rate of the radar emitter signal also decreases. It can be concluded that under the condition of pulse stagger, the radar signal intelligent sorting method provided by the present invention can well suppress the situation of low sorting accuracy rate of radar signals.

[0086] In summary, under the complex electromagnetic environment considering the changes of parameters such as signal-to-noise ratio, missing pulse rate, and staggered pulse rate and the existence of multi-functional radar signals, the sorting accuracy rate of the radar signal intelligent sorting method provided by the present invention is significantly better than that of the PRI transformation method and the SDIF method, which proves the effectiveness and superiority of the present invention.

[0087] Figure 7 is a schematic structural diagram of a radar signal intelligent sorting device in a complex electromagnetic environment provided by an embodiment of the present invention. As Figure 7 shown, an embodiment of the present invention provides a radar signal intelligent sorting device in a complex electromagnetic environment, including:

[0088] A receiving module 710, configured to receive a pulse signal stream formed by random overlapping of pulse signals from different radar emitters, where the pulse signal stream includes multiple groups of radar pulse sequences;

[0089] A preprocessing module 720, configured to preprocess each group of radar pulse sequences to generate a signal time-frequency diagram of a preset size;

[0090] A sorting module 730, configured to sequentially input multiple signal time-frequency diagrams into a radar signal sorting model, and classify the obtained reconstructed signals to obtain a radar signal sorting result.

[0091] An embodiment of the present invention also provides an electronic device, as Figure 8 shown, including a processor 801, a communication interface 802, a memory 803, and a communication bus 804. Among them, the processor 801, the communication interface 802, and the memory 803 complete mutual communication through the communication bus 804,

[0092] A memory 803 for storing computer programs;

[0093] A processor 801, when executing the programs stored on the memory 803, implements the following steps:

[0094] Receiving a pulse signal stream formed by random overlapping of pulse signals from different radar radiation sources, where the pulse signal stream contains multiple groups of radar pulse sequences;

[0095] Preprocessing each group of radar pulse sequences to generate a signal time-frequency diagram of a preset size;

[0096] Sequentially inputting multiple signal time-frequency diagrams into a radar signal sorting model, classifying the obtained reconstructed signals, and obtaining a radar signal sorting result.

[0097] The communication bus mentioned in the above electronic device can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This communication bus can be divided into an address bus, a data bus, a control bus, etc. For the sake of representation, only a thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.

[0098] The communication interface is used for communication between the above electronic device and other devices.

[0099] The memory can include a Random Access Memory (RAM), and can also include a Non-Volatile Memory (NVM), such as at least one disk memory. Optionally, the memory can also be at least one storage device located far from the aforementioned processor.

[0100] The above-mentioned processor can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it can also be a Digital Signal Processing (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.

[0101] In the description of the present invention, the description with reference to terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine the different embodiments or examples described in this specification.

[0102] Although the present application has been described herein in connection with various embodiments, however, in implementing the claimed present application, those skilled in the art can understand and realize other variations of the disclosed embodiments by viewing the accompanying drawings, the disclosure, and the appended claims.

[0103] The above content is a further detailed description of the present invention in combination with specific preferred embodiments, and it cannot be determined that the specific implementation of the present invention is only limited to these descriptions. For those of ordinary skill in the technical field to which the present invention pertains, without departing from the concept of the present invention, several simple deductions or substitutions can be made, and all should be regarded as belonging to the protection scope of the present invention.

Claims

1. A method for intelligently sorting radar signals in a complex electromagnetic environment, characterized in that: include: receiving a pulse signal stream formed by random overlap of pulse signals from different radar radiation sources, wherein the pulse signal stream includes a plurality of radar pulse sequences; Preprocess each group of radar pulse sequences to generate a signal time-frequency diagram of a preset size; Multiple signal time-frequency diagrams are sequentially input into the radar signal sorting model, and the obtained reconstructed signals are classified to obtain radar signal sorting results.

2. The method for intelligently sorting radar signals in a complex electromagnetic environment according to claim 1, characterized in that: The step of preprocessing each group of radar pulse sequences to generate a signal time-frequency diagram of a preset size includes: Calculate the relevant parameters of each radar pulse sequence; After the radar pulse sequence is normalized, a short-time Fourier transform is performed to encode the relevant parameters, and a two-dimensional signal time-frequency diagram of a preset size is generated through downsampling processing; the relevant parameters include the pulse carrier frequency RF, pulse width PW and pulse repetition period PRI of the radar pulse sequence.

3. The method for intelligently sorting radar signals in a complex electromagnetic environment according to claim 1, characterized in that: The radar signal sorting model includes an encoder, and the encoder includes a convolution module and a flattening module; wherein, The convolution module is used to extract the feature vector of the radar pulse sequence, including a convolution layer Conv1, a first pooling layer, a convolution layer Conv2, a second pooling layer, a convolution layer Conv3 and a third pooling layer connected in sequence, and the flattening module is used to flatten the feature map vector into a one-dimensional vector.

4. The method for intelligently sorting radar signals in a complex electromagnetic environment according to claim 3 is characterized in that: The radar signal sorting model also includes a decoder, and the decoder includes a reconstruction module and a transposed convolution module; wherein, The reconstruction module is used to reconstruct the dimension of the one-dimensional vector, and the transposed convolution module is used to restore the spatial size and the number of channels of the reconstructed one-dimensional vector to obtain a reconstructed signal. The transposed convolution module includes a first upsampling layer, a first transposed convolution layer, a second upsampling layer, a second transposed convolution layer, a third upsampling layer and a third transposed convolution layer connected in sequence.

5. The method for intelligently sorting radar signals in a complex electromagnetic environment according to claim 4, characterized in that: It also includes a fully connected layer, which is used to use the Softmax activation function mapping to obtain the probability distribution of the reconstructed signal coming from each radiation source, and obtain the radar signal sorting result.

6. The method for intelligently sorting radar signals in a complex electromagnetic environment according to claim 1, characterized in that: The radar signal sorting model is obtained based on a mean square error loss function.

7. The method for intelligently sorting radar signals in a complex electromagnetic environment according to claim 6, characterized in that: The mean square error loss function is expressed as: Where N represents the number of sample signals, x i represents the sorting result of the i-th sample signal, x i Represents the label of the i-th sample signal.

8. An intelligent radar signal sorting device in a complex electromagnetic environment, characterized in that: include: A receiving module, used for receiving a pulse signal stream formed by random overlap of pulse signals from different radar radiation sources, wherein the pulse signal stream includes a plurality of radar pulse sequences; A preprocessing module, used to preprocess each group of radar pulse sequences and generate a signal time-frequency diagram of a preset size; The sorting module is used to input multiple signal time-frequency diagrams into the radar signal sorting model in sequence, and classify the obtained reconstructed signals to obtain radar signal sorting results.

9. An electronic device, characterized in that: It includes a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus; Memory, used to store computer programs; A processor, for implementing the method steps described in any one of claims 1 to 7 when executing a program stored in a memory.

Images