Radar operating mode recognition method and system based on time-frequency analysis feature fusion

Abstract

The present application relates to radar communication technology, specifically to a radar operating mode recognition method and system based on time-frequency analysis feature fusion, which extracts the time-frequency distribution and power spectrum of the radar radiation source through time-frequency analysis and power spectrum estimation processing, constructs a feature fusion model for radar operating mode recognition, and evaluates and rebuilds the recognition result through the training label, constantly obtaining the optimal operating mode recognition result. The method fuses the time-frequency distribution features of the radar signal and the power spectrum features with statistical significance, and the constructed operating mode recognition model has better adaptability and recognition probability for radar operating mode recognition in the presence of missing pulses, missing pulses and false pulses.

Description

Technical Field

[0001] This invention belongs to the field of radar communication technology, and specifically relates to a radar operating mode recognition method and system based on time-frequency analysis feature fusion. Background Technology

[0002] With the continuous development of radar technology, the operating modes of radar systems have become increasingly complex and diverse. Electronic reconnaissance receivers, as a crucial means of electronic countermeasures, are an essential tool in modern electronic warfare. By receiving and analyzing radar signal parameters and based on the design principles of radar waveforms, electronic reconnaissance receivers can determine the radar's operating mode, thereby obtaining the radar's system type and threat level. This provides decision support for jamming releases and destruction strikes, playing a key role in electronic countermeasures in actual battlefield environments.

[0003] Radar signals exhibit diverse characteristics under different operating modes, and similarities exist between different radars. Radar operating mode recognition, a crucial component of electronic reconnaissance systems, requires identifying operating modes by analyzing intercepted signals and uncovering patterns in signal variations. Current radar operating mode recognition methods suffer from insufficient feature parameters, resulting in unsatisfactory recognition performance in the presence of lost, missing, or false pulses. Summary of the Invention

[0004] To address the problems existing in the background technology, the present invention provides a radar operating mode recognition method based on time-frequency analysis feature fusion.

[0005] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: a radar operating mode recognition method based on time-frequency analysis feature fusion, comprising:

[0006] Acquire radar signals;

[0007] By combining synchronous extraction transform, variational mode decomposition, and short-time fractional Fourier transform, a synchronous extraction short-time fractional Fourier transform algorithm Pro-STFrFT based on variational mode decomposition is obtained, which extracts the fractional-order time-frequency distribution features x1 of the radar signal;

[0008] The radar signal is processed by cepstral-based multi-synchronous compression De-shape MSST using the multiple synchronous compression algorithm MSST and cepstral transform to obtain the short-time time-frequency characteristics x2 of the radar signal;

[0009] The power spectrum characteristics of the radar signal are estimated using a power spectrum estimation algorithm, resulting in the power spectrum characteristic x3 of the radar signal.

[0010] A radar operating mode feature fusion model is constructed based on the fractional-order time-frequency distribution characteristics x1, the short-time time-frequency characteristics x2, and the power spectrum characteristics x3 of the radar signal. The model is trained by taking the radar signal features and radar operating mode information as inputs and outputs, and adjusting the weights and connection methods of the internal elements to minimize the loss function of the operating mode feature fusion model.

[0011] The radar signal is processed using a constructed working mode fusion model to identify the radar working mode.

[0012] In the aforementioned radar operating mode recognition method based on time-frequency analysis feature fusion, the synchronous extraction short-time fractional Fourier transform algorithm Pro-STFrFT based on variational mode decomposition extracts the fractional-order time-frequency distribution features x1 of the radar signal, including:

[0013] ① Perform FrFT calculation on the input radar signal;

[0014] ② Calculate the kurtosis coefficients of FrFT at different orders, find the number of peaks K of the kurtosis coefficients, and set the order search range according to the maximum and minimum orders corresponding to the peaks of the kurtosis coefficients;

[0015] ③ Perform VMD decomposition on the input radar signal to obtain the set of components representing the radar signal;

[0016] ④ Perform local optimal STFrFT calculation on each radar signal component to obtain the time-frequency distribution of each radar signal component;

[0017] ⑤ The time-frequency distributions of the radar signal components are superimposed to obtain the time-frequency representation of the original radar input signal;

[0018] ⑥ Calculate the synchronous extraction operator SEO to obtain the fractional-order time-frequency distribution characteristics x1 of the radar signal.

[0019] In the radar operating mode recognition method based on time-frequency analysis feature fusion described above, the cepstral-based multiple synchronous compression De-shape MSST processing for obtaining the short-time time-frequency features x2 of the radar signal includes:

[0020] (1) Perform a short-time Fourier transform (STFT) on the signal to obtain the time-frequency characteristics of the input signal;

[0021] (2) Perform cepstral transformation on the time-frequency features, including short-time cepstral transformation, cepstral domain filtering, and inverse short-time cepstral transformation;

[0022] (3) Multiply the time-frequency characteristics of the signal and the cepstral transform result to obtain the time-frequency distribution characteristics of the short-time Fourier transform (STFT) based on the cepstral transform De-shape;

[0023] (4) Use multiple synchronous compression (MSST) to aggregate time-frequency features, so as to reduce the influence of false signals on time-frequency features and obtain highly aggregated short-time time-frequency features of the signal.

[0024] In the radar operating mode recognition method based on time-frequency analysis feature fusion described above, the calculation of the power spectrum feature x3 of the radar signal adopts the AR model algorithm.

[0025] In the radar operating mode recognition method based on time-frequency analysis feature fusion described above, the typical radar operating modes include: velocity search (VS), search-while-ranging (RWS), search-while-tracking (TWS), search-plus-tracking (TAS), single-target tracking (STT), and multi-target tracking (MTT). The loss function value of the trained feature fusion model gradually decreases with the increase of training times and eventually converges to a certain value.

[0026] A radar operating mode recognition system based on time-frequency analysis feature fusion includes a radar signal extraction module for acquiring radar signals; a radar signal feature extraction module for extracting radar signal features; a feature fusion module for constructing a radar operating mode feature model from the extracted radar signal features; and a signal processing module for processing the radar signals and recognizing the radar operating mode.

[0027] An electronic device includes a computer-readable storage medium storing computer-executable instructions; and one or more processors coupled to the computer-readable storage medium and configured to execute the computer-executable instructions to cause the device to perform the method described above.

[0028] A readable storage medium storing computer-executable instructions that, when executed by a processor, configure the processor to perform the method described above.

[0029] Compared with the prior art, the beneficial effects of the present invention are: the present invention extracts three signal features of radar signals: fractional-order time-frequency distribution features, short-time time-frequency distribution features, and power spectrum features, and constructs a radar operating mode feature fusion network for joint feature extraction. This can obtain the feature fusion parameters of the radar operating mode, reduce the impact of signal noise, pulse loss, and interference signals on radar operating mode recognition, and has better adaptability and recognition probability in the presence of lost pulses, missing pulses, and false pulses. Attached Figure Description

[0030] Figure 1 This is a flowchart of a radar operating mode recognition method based on time-frequency analysis feature fusion according to an embodiment of the present invention;

[0031] Figure 2This is a flowchart of the synchronous extraction short-time fractional Fourier transform algorithm (Pro-STFrFT) based on variational mode decomposition according to an embodiment of the present invention.

[0032] Figure 3 This is a flowchart of the cepstral-based multiple synchronous compression (De-shape MSST) embodiment of the present invention;

[0033] Figure 4 This is a schematic diagram illustrating the training of the feature fusion model in an embodiment of the present invention;

[0034] Figure 5 This is a schematic diagram of the loss function values ​​of the feature fusion network after training, according to an embodiment of the present invention.

[0035] Figure 6 This is a schematic diagram of radar operating mode recognition of the radar operating mode fusion network trained according to an embodiment of the present invention. Detailed Implementation

[0036] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0037] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other.

[0038] The present invention will be further described below with reference to specific embodiments, but these are not intended to limit the scope of the invention.

[0039] In the complex electromagnetic environment of the battlefield, various new and unknown electromagnetic threat radiation sources emerge constantly. Improving the radar operating mode recognition capability of electronic reconnaissance systems in complex electromagnetic environments is a key issue for electronic warfare to meet the survivability challenges of future battlefields. This embodiment proposes a radar operating mode recognition method based on time-frequency analysis feature fusion. Through time-frequency analysis and power spectrum estimation, the time-frequency distribution and power spectrum of the radar radiation source are extracted, and a feature fusion model is constructed for radar operating mode recognition. The recognition results are evaluated and improved through training labels to continuously obtain the optimal operating mode recognition results. This embodiment fuses the time-frequency distribution characteristics of radar signals and statistically significant power spectrum characteristics. The constructed operating mode recognition model has better adaptability and recognition probability in the presence of lost pulses, missing pulses, and false pulses.

[0040] This embodiment is achieved through the following technical solution: a radar operating mode recognition method based on time-frequency analysis feature fusion, comprising:

[0041] S1. Acquire radar signals that are conducive to completing the radar operating mode recognition task;

[0042] S2. By combining synchronous extraction transform, variational mode decomposition and short-time fractional Fourier transform, a synchronous extraction short-time fractional Fourier transform algorithm based on variational mode decomposition (Pro-STFrFT) is obtained to extract the fractional-order time-frequency distribution features x1 of the radar signal;

[0043] S3. The radar signal is processed by cepstral-based multi-synchronous compression (De-shape MSST) using the multiple synchronous compression algorithm (MSST) and cepstral transform to obtain the short-time time-frequency characteristics x2 of the radar signal;

[0044] S4. The power spectrum characteristics of the radar signal are estimated using a power spectrum estimation algorithm to obtain the power spectrum characteristics x3 of the radar signal.

[0045] S5. Based on x1, x2, and x3, construct a radar operating mode feature fusion model with high comprehensiveness. By using radar signal features and radar operating mode information as inputs and outputs, train the operating mode feature fusion model, and adjust the internal element weights and connection methods to minimize the model loss function.

[0046] S6. By processing the radar signal using the constructed model, the identification result of the radar operating mode can be obtained.

[0047] Example 1

[0048] like Figure 1 As shown, a radar operating mode recognition method based on time-frequency analysis feature fusion includes: 1) acquiring radar signals; 2) combining synchronous extraction transform, variational mode decomposition, and short-time fractional Fourier transform to obtain a synchronous extraction short-time fractional Fourier transform algorithm based on variational mode decomposition (Pro-STFrFT), extracting fractional-order time-frequency distribution features x1 of the radar signal; 3) using a multiple synchronous compression algorithm (MSST) and cepstral transform to perform cepstral-based multiple synchronous compression (De-shape MSST) processing on the radar signal to obtain high-density time-frequency features x2 of the radar signal; 4) using a power spectrum estimation algorithm to estimate the power spectrum features of the radar signal to obtain the power spectrum features x3 of the radar signal; 5) constructing a radar operating mode feature fusion model with high comprehensiveness based on x1, x2, and x3, training the operating mode feature fusion model by using radar signal features and radar operating mode information as input and output, adjusting the internal element weights and connection methods to minimize the model loss function; 6) using the constructed model to process the radar signal to obtain the radar operating mode recognition result.

[0049] like Figure 2 As shown, in step S2, the specific implementation method of the synchronous extraction short-time fractional Fourier transform algorithm (Pro-STFrFT) based on variational mode decomposition is as follows: ① Perform FrFT calculation on the input signal; ② Calculate the kurtosis coefficients of FrFT at different orders, and find the number of peaks K of the kurtosis coefficients, while setting the order search range according to the maximum and minimum orders corresponding to the peaks of the kurtosis coefficients; ③ Perform VMD decomposition on the input signal to obtain the set of components that can represent the signal; ④ Perform local optimal STFrFT calculation on each component to obtain the time-frequency distribution of each component; ⑤ Superimpose the time-frequency distributions of the components to obtain the time-frequency representation of the original input signal; ⑥ Calculate the synchronous extraction operator SEO to obtain the fractional-order time-frequency distribution characteristics of the signal.

[0050] like Figure 3 As shown, in step S3, the specific implementation method of cepstral-based multiple synchronization compression (De-shape MSST) processing is as follows: ① Perform short-time Fourier transform (STFT) on the signal to obtain the time-frequency characteristics of the input signal; ② Perform cepstral transform processing on the time-frequency characteristics, including short-time cepstral transform, cepstral domain filtering, and inverse short-time cepstral transform; ③ Multiply the signal time-frequency characteristics and the cepstral transform result to obtain the cepstral transform (De-shape) short-time Fourier (STFT) time-frequency distribution characteristics; ④ Use multiple synchronization compression (MSST) to aggregate the time-frequency characteristics to reduce the influence of false signals on the time-frequency characteristics and obtain highly aggregated short-time time-frequency characteristics of the signal.

[0051] In step S4, the power spectrum calculation uses the AR model algorithm to obtain the power spectrum feature x3 of the signal;

[0052] like Figure 4 As shown, in step S5, a radar radiation source working mode feature fusion network based on convolutional neural network is constructed. The extracted radar signal fractional-order time-frequency distribution features x1, short-time time-frequency distribution features x2, and power spectrum features x3 are used as inputs to train the feature fusion model. Information such as the weights of internal elements and connection methods are adjusted to minimize the model loss function.

[0053] like Figure 5 As shown, based on the construction of a training database of typical radar operating modes, including: VS (velocity search), RWS (search-while-ranging), TWS (search-while-tracking), TAS (search-plus-tracking), STT (single target tracking), MTT (multi-target tracking), etc., the loss function value of the feature fusion network after training gradually decreases with the increase of training times, and eventually converges to a certain value.

[0054] like Figure 6As shown, using the radar operating mode fusion network trained above for radar operating mode recognition can achieve a high recognition probability.

[0055] Example 2

[0056] A radar operating mode recognition system based on time-frequency analysis feature fusion includes a radar signal extraction module for acquiring radar signals; a radar signal feature extraction module for extracting radar signal features; a feature fusion module for constructing a radar operating mode feature model from the extracted radar signal features; and a signal processing module for processing the radar signals and recognizing the radar operating modes.

[0057] Example 3

[0058] An electronic device includes a computer-readable storage medium storing computer-executable instructions; and one or more processors coupled to the computer-readable storage medium and configured to execute the computer-executable instructions such that the device performs the method described in Embodiment 1.

[0059] Example 4

[0060] A readable storage medium storing computer-executable instructions that, when executed by a processor, configure the processor to perform the method described in Embodiment 1.

[0061] The above are merely preferred embodiments of the present invention and are not intended to limit the implementation methods and protection scope of the present invention. Those skilled in the art should recognize that any equivalent substitutions and obvious changes made based on the content of this specification should be included within the protection scope of the present invention.

Claims

1. A radar operating mode recognition method based on time-frequency analysis feature fusion, characterized in that, include: Acquire radar signals; By combining synchronous extraction transform, variational mode decomposition, and short-time fractional Fourier transform, a synchronous extraction short-time fractional Fourier transform algorithm Pro-STFrFT based on variational mode decomposition is obtained, which extracts the fractional-order time-frequency distribution features x1 of the radar signal; The radar signal is processed using the multiple synchronous compression algorithm (MSST) and cepstral transform (CST) to obtain the short-time time-frequency characteristics (x2) of the radar signal. The specific steps are as follows: ① Perform a short-time Fourier transform (STFT) on the signal to obtain the time-frequency characteristics of the input signal; ② Perform cepstral transformation on the time-frequency features, including short-time cepstral transformation, cepstral domain filtering, and inverse short-time cepstral transformation; ③ Multiply the signal time-frequency characteristics and the cepstral transform result to obtain the short-time Fourier transform (STFT) time-frequency distribution characteristics based on the cepstral transform De-shape; ④ Use multiple synchronous compression (MSST) to aggregate time-frequency features, so as to reduce the influence of false signals on time-frequency features and obtain highly aggregated short-time time-frequency features of the signal; The power spectrum characteristics of the radar signal are estimated using a power spectrum estimation algorithm, resulting in the power spectrum characteristic x3 of the radar signal. A radar operating mode feature fusion model is constructed based on the fractional-order time-frequency distribution characteristics x1, the short-time time-frequency characteristics x2, and the power spectrum characteristics x3 of the radar signal. The model is trained by taking the radar signal features and radar operating mode information as inputs and outputs, and adjusting the weights and connection methods of the internal elements to minimize the loss function of the operating mode feature fusion model. The radar signal is processed using a constructed working mode fusion model to identify the radar working mode.

2. The radar operating mode recognition method based on time-frequency analysis feature fusion according to claim 1, characterized in that, The synchronous extraction short-time fractional Fourier transform algorithm Pro-STFrFT based on variational mode decomposition extracts the fractional-order time-frequency distribution features x1 of the radar signal, including: ① Perform FrFT calculation on the input radar signal; ② Calculate the kurtosis coefficients of FrFT at different orders, find the number of peaks K of the kurtosis coefficients, and set the order search range according to the maximum and minimum orders corresponding to the peaks of the kurtosis coefficients; ③ Perform VMD decomposition on the input radar signal to obtain the set of components representing the radar signal; ④ Perform local optimal STFrFT calculation on each radar signal component to obtain the time-frequency distribution of each radar signal component; ⑤ The time-frequency distributions of the radar signal components are superimposed to obtain the time-frequency representation of the original radar input signal; ⑥ Calculate the synchronous extraction operator SEO to obtain the fractional-order time-frequency distribution characteristics x1 of the radar signal.

3. The radar operating mode recognition method based on time-frequency analysis feature fusion according to claim 1, characterized in that, The power spectrum characteristic x3 of the radar signal is calculated using the AR model algorithm.

4. The radar operating mode recognition method based on time-frequency analysis feature fusion according to claim 1, characterized in that, The radar's operating modes include: velocity search (VS), search-while-ranging (RWS), search-while-tracking (TWS), search-plus-tracking (TAS), single-target tracking (STT), and multi-target tracking (MTT). The loss function value of the trained feature fusion model gradually decreases with the increase of training times, eventually converging to a certain value.

5. A system for the radar operating mode recognition method based on time-frequency analysis feature fusion as described in any one of claims 1-4, characterized in that, It includes a radar signal extraction module for acquiring radar signals and a radar signal feature extraction module for extracting radar signal features; The feature fusion module is used to construct a radar operating mode feature model from the extracted radar signal features; the signal processing module is used to process the radar signals and identify the radar operating mode.

6. An electronic device, characterized in that, A computer-readable storage medium storing computer-executable instructions; and one or more processors coupled to the computer-readable storage medium and configured to execute the computer-executable instructions to cause the device to perform the method according to any one of claims 1-4.

7. A readable storage medium, characterized in that, The system stores computer-executable instructions that, when executed by a processor, configure the processor to perform the method according to any one of claims 1-4.

Images