Radar composite interference time sequence parameter intelligent extraction method based on frequency agility

Abstract

The invention discloses a radar composite interference time sequence parameter intelligent extraction method based on frequency agility, and relates to the technical field of radars, the radar composite interference time sequence parameter intelligent extraction method based on frequency agility employs an interference separation network fusing convolution enhancement and a Transform attention mechanism to carry out separation processing on radar echo signals, and the radar composite interference time sequence parameter intelligent extraction method based on frequency agility is obtained. The interference separation network strengthens the local feature extraction capability by means of convolution operation, and introduces a cross-scale attention mechanism to capture the global correlation of interference signals, thereby achieving the precise separation of a multi-component composite ISRJ signal, and obtaining a plurality of separated ISRJ component signals. On the basis, through dynamic threshold detection, by means of a robustness measurement algorithm based on pulse detection and clustering optimization, according to time domain characteristics of the ISRJ component signals, accurate measurement of interference parameters such as slice width, forwarding width and forwarding times of the ISRJ component signals is achieved.

Description

Technical Field

[0001] The present invention belongs to the technical field of radar, and particularly relates to an intelligent extraction method for timing parameters of radar composite jamming based on frequency agility. Background Art

[0002] In the current complex and changeable electromagnetic environment, radar systems are facing severe challenges from new jamming threats centered on digital radio frequency memory (DRFM, digital radio frequency memory) technology. Interrupted-sampling Repeater Jamming (ISRJ) intercepts radar signals and performs interrupted sampling, modulation, and forwarding operations to generate a large number of false targets in the time domain and frequency domain, thereby significantly weakening the radar's ability to measure the parameters of real targets. In the composite ISRJ scenario, multiple jamming signals are coupled with each other in the space-time-frequency domain, and their parameter components overlap with each other, making it difficult for traditional parameter estimation methods to achieve effective separation and accurate extraction. Jamming perception, as a key link in the anti-jamming process, has an important impact on subsequent anti-jamming waveform design and strategy selection. Therefore, breaking through the parameter estimation bottleneck of composite ISRJ has become the core prerequisite for anti-jamming waveform design. Existing ISRJ parameter estimation methods can achieve good parameter estimation effects in single jamming scenarios, but there is still a lack of effective and in-depth research on composite ISRJ parameter extraction. Most of the existing few methods involve complex time-frequency domain calculations and are not ideal in the case of low jamming-to-noise ratio and low number of training samples. Summary of the Invention

[0003] In order to solve the above problems existing in the prior art, the present invention provides an intelligent extraction method for timing parameters of radar composite jamming based on frequency agility. The technical problems to be solved by the present invention are realized through the following technical solutions: The present invention provides an intelligent extraction method for timing parameters of radar composite jamming based on frequency agility, including: Step 1: Preprocess the radar echo signal to obtain a preprocessed echo signal, where the radar echo signal is a frequency-agile radar echo signal containing composite ISRJ; Step 2: Input the preprocessed echo signal into a trained composite ISRJ separation network to obtain multiple separated ISRJ component signals, where the composite ISRJ separation network is an interference separation network that integrates convolutional enhancement and Transformer attention mechanism; Step 3: Calculate the dynamic threshold for each ISRJ component signal, and perform binarization processing on each ISRJ component signal according to the dynamic threshold calculation result to obtain a binarized ISRJ signal corresponding to each ISRJ component signal; Step 4: Use the clustering optimization algorithm to perform clustering optimization on each binarized ISRJ signal, and obtain the estimated values of the interference parameters of each ISRJ component signal according to the clustering optimization results.

[0004] Compared with the prior art, the beneficial effects of the present invention are as follows: The intelligent extraction method for the radar composite interference timing parameters based on frequency agile of the present invention uses an interference separation network that combines convolutional enhancement and Transformer attention mechanism to separate the radar echo signal. This interference separation network strengthens the local feature extraction ability through convolutional operations and introduces a cross-scale attention mechanism to capture the global correlation of the interference signal, so as to accurately separate the multi-component composite ISRJ signal into multiple separated ISRJ component signals. On this basis, through dynamic threshold detection, using a robust measurement algorithm based on pulse detection and clustering optimization, according to the time-domain characteristics of the ISRJ component signal, accurate measurement of interference parameters such as the slice width, forwarding width, and forwarding times of the ISRJ component signal is achieved.

[0005] The above description is only an overview of the technical solution of the present invention. In order to be able to understand the technical means of the present invention more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features, and advantages of the present invention more obvious and understandable, the following specifically gives preferred embodiments and, in conjunction with the drawings, details are described as follows. Description of the Drawings

[0006] Figure 1 is a schematic diagram of an intelligent extraction method for radar composite interference timing parameters based on frequency agile provided by an embodiment of the present invention; Figure 2 is a flow example diagram of an intelligent extraction method for radar composite interference timing parameters based on frequency agile provided by an embodiment of the present invention; Figure 3 is a time-domain model example diagram of a composite ISRJ signal provided by an embodiment of the present invention; Figure 4 is a time-frequency domain model example diagram of a composite ISRJ signal provided by an embodiment of the present invention; Figure 5 is a structural schematic diagram of a composite ISRJ separation network provided by an embodiment of the present invention; Figure 6 is a structural schematic diagram of an attention block provided by an embodiment of the present invention; Figure 7 is a structural schematic diagram of a one-dimensional convolutional layer provided by an embodiment of the present invention; Figure 8 is a comparison diagram of the parameter estimation accuracy of the method proposed by the present invention under different JNR conditions under double interference conditions; Figure 9 It is a comparison chart of the parameter estimation accuracy of the method proposed in the present invention under different JNR conditions with three interference conditions; Figure 10 It is a comparison chart of the parameter estimation accuracy of the method proposed in the present invention under different training set sizes with two interference conditions; Figure 11 It is a comparison chart of the parameter estimation accuracy of the method proposed in the present invention under different training set sizes with three interference conditions. Detailed implementation manners

[0007] In order to further elaborate on the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following combines the accompanying drawings and specific implementation manners to detail a method for intelligent extraction of radar composite interference timing parameters based on frequency agility proposed according to the present invention.

[0008] The foregoing and other technical contents, features and effects of the present invention can be clearly presented in the following detailed description in conjunction with the accompanying drawings. Through the description of the specific implementation manners, a more in-depth and specific understanding of the technical means and effects adopted by the present invention to achieve the predetermined purpose can be obtained. However, the accompanying drawings are only for reference and explanation, and are not used to limit the technical solutions of the present invention.

[0009] The embodiment of the present invention provides a method for intelligent extraction of radar composite interference timing parameters based on frequency agility. Please refer to Figure 1 , Figure 1 which is a schematic diagram of a method for intelligent extraction of radar composite interference timing parameters based on frequency agility provided by the embodiment of the present invention. As Figure 1 shown, the method for intelligent extraction of radar composite interference timing parameters based on frequency agility in this embodiment includes the following steps: Step 1: Preprocess the radar echo signal to obtain a preprocessed echo signal. The radar echo signal is a frequency-agile radar echo signal containing composite ISRJ; Step 2: Input the preprocessed echo signal into the trained composite ISRJ separation network to obtain multiple separated ISRJ component signals. Among them, the composite ISRJ separation network is an interference separation network that fuses convolutional enhancement and Transformer attention mechanism; Step 3: Calculate the dynamic threshold for each ISRJ component signal, and perform binarization processing on each ISRJ component signal according to the dynamic threshold calculation result to obtain the binarized ISRJ signal corresponding to each ISRJ component signal; Step 4: Use the clustering optimization algorithm to perform clustering optimization processing on each binarized ISRJ signal, and obtain the estimated value of the interference parameter of each ISRJ component signal according to the clustering optimization result.

[0010] In this embodiment, the interference parameters include the slice width, the forwarding width, and the number of forwardings.

[0011] Please refer to Figure 2 , Figure 2 which is a flowchart example of an intelligent extraction method for radar composite interference timing parameters based on frequency agility provided by an embodiment of the present invention. As Figure 2 shown, in the intelligent extraction method for radar composite interference timing parameters based on frequency agility of the present invention, an interference separation network that combines convolutional enhancement and Transformer attention mechanism, namely a composite ISRJ separation network, is used to separate the radar echo signal, that is, the composite ISRJ signal. This interference separation network strengthens the local feature extraction ability through convolutional operations and introduces a cross-scale attention mechanism to capture the global correlation of interference signals, thereby achieving precise separation of the multi-component composite ISRJ signal to obtain multiple separated ISRJ component signals. On this basis, through dynamic threshold detection, using a robust measurement algorithm based on pulse detection and clustering optimization, according to the time-domain characteristics of the ISRJ component signals, the slice width of the ISRJ component signals, the forwarding width and other interference parameters such as the number of forwardings are accurately measured.

[0012] Furthermore, the intelligent extraction method for radar composite interference timing parameters based on frequency agility in this embodiment will be described in detail.

[0013] First, the frequency-agile radar echo signal containing composite ISRJ is introduced. When multiple jammers are deployed in the interference system, the received radar echo signal contains ISRJ component signals with different interference parameters, randomly overlapping in the time domain and frequency domain. At this time, the received radar echo signal can be expressed as: ; where, is the received radar echo signal, is the amplitude of the radar echo signal, is the target echo signal, represents the th ISRJ component signal, represents time, and respectively represent the time delays of the target signal and the th ISRJ component signal, is the total number of ISRJ component signals included, is Gaussian noise with a mean of 0 and a variance of .

[0014] Please refer to Figure 3 ,Figure 3 This is an example diagram of the time-domain model of a composite ISRJ signal provided by an embodiment of the present invention. As Figure 3 shown, the figure includes a noisy target signal and three ISRJ component signals (ISRJ1, ISRJ2, and ISRJ3). Each ISRJ component signal is represented by different interference parameter characteristics: slice width, forwarding width, and number of forwardings. The composite ISRJ signal formed by the overlap of the above components is the received radar echo signal.

[0015] Please refer to Figure 4 , Figure 4 This is an example diagram of the time-frequency domain model of a composite ISRJ signal provided by an embodiment of the present invention. As Figure 4 shown, it can be seen that each ISRJ component signal is highly overlapped in the time domain and time-frequency domain. At the same time, the influence of noise also makes it difficult to directly measure interference parameters through methods such as envelope detection and time-frequency filtering.

[0016] In an optional embodiment, step 1 includes: performing maximum normalization processing on the radar echo signal, compressing the radar echo signal to the [0, 1] dynamic range, and obtaining a preprocessed echo signal.

[0017] In this embodiment, by performing maximum normalization processing on the radar echo signal, the radar echo signal can be converted into an input format suitable for the composite ISRJ separation network.

[0018] It should be noted that, in order to simplify the processing and improve the accuracy of interference signal separation, this embodiment regards the stepped-frequency radar echo signal and random noise in the received signal as one signal component, and regards each ISRJ component signal as a separate signal component. Therefore, the received time-domain signal can be simplified as: ; where represents the superimposed component of the target echo and noise, that is, the target signal, represents the th ISRJ component signal, is the number of sampling points included in a pulse repetition interval, is the total number of ISRJ component signals included.

[0019] In this embodiment, the composite ISRJ separation network includes: an encoder, a mask network, and a decoder. Among them, the encoder is used to extract features from the preprocessed echo signal to obtain an encoded feature sequence; the mask network is used to implement a non-linear mapping from the encoded feature sequence to multiple time-domain masks to obtain multiple mask vectors; the decoder is used to perform a deconvolution operation on the Kronecker product of the multiple mask vectors and the encoded feature sequence to obtain multiple ISRJ component signals.

[0020] For a radar echo signal containing a composite ISRJ, the independent ISRJ component signals can be estimated based on a deep learning model where , T is a time series. The interference separation network proposed in this embodiment continues the paradigm architecture of an encoder-mask network-decoder. Aiming at the limitations of traditional time-domain convolutional networks in modeling long-range temporal dependencies, this embodiment innovatively constructs a mask network based on a hierarchical Transformer self-attention mechanism. This mask network realizes global context awareness through the multi-head self-attention mechanism, combines a gated convolutional module to enhance local feature extraction capabilities, forms a hybrid architecture with both global-local modeling advantages, and effectively solves the problem of long-term temporal correlation attenuation caused by the local receptive field limitation of traditional convolutional networks.

[0021] Please refer to Figure 5 , Figure 5 which is a schematic structural diagram of a composite ISRJ separation network provided by an embodiment of the present invention. As Figure 5 shown, in this embodiment, the encoder includes a cascaded first one-dimensional convolutional layer (1-D Conv) and a ReLU layer. The encoder is used to process the input sequence and map it to a higher-dimensional space, providing important feature information for subsequent processing. Let the size of the encoder convolution kernel be , the stride be , the number of filters be , and the input sequence be , where b is the batch size, T is the time series. The encoded feature sequence obtained after passing through the encoder can be expressed as: ; wherein, , is the time series after passing through the encoder, , represents the ReLU (Rectified Linear Unit) activation function, represents performing a one-dimensional convolution operation on the input sequence . Subsequently, for the sake of simplicity in derivation, the batch size b is uniformly omitted.

[0022] In this embodiment, the mask network includes a normalization layer (LayerNorm), a position encoding layer, a first pointwise convolutional layer (1-1 Conv), multiple attention blocks, a ReLU layer, a second pointwise convolutional layer (1-1 Conv), and a Sigmoid layer. Among them, in multiple attention blocks, the output of the previous attention block serves as the input of the next attention block; the encoded feature sequence passes through the normalization layer and the position encoding layer to obtain a feature sequence retaining global temporal information, and the feature sequence retaining global temporal information is passed to multiple attention blocks through the first pointwise convolutional layer; multiple attention blocks utilize the self-attention mechanism to capture long-range dependencies in the input feature sequence, and the feature sequence output by the last attention block undergoes dimensionality expansion through the ReLU layer and the second pointwise convolutional layer, and the feature sequence after dimensionality expansion is mapped through the Sigmoid layer to obtain multiple mask vectors.

[0023] The role of the mask network is to implement the non-linear mapping from the encoder output to multiple groups of time-domain masks, and the encoded feature sequence output by the encoder ( N where the feature dimension is consistent with the number of filters, L and it is the time series after passing through the encoder) first undergoes normalization and position encoding to retain global temporal information, and is passed to the attention blocks through pointwise convolution for sequential processing. The attention blocks combine convolutional operations with the attention mechanism to achieve a hybrid process of local feature extraction and global relationship modeling. The attention blocks adopt the self-attention mechanism to capture long-range dependencies. Finally, the attention blocks use residual connections for better training. The output of the current attention block serves as the input of the next attention block, and this is repeated times. The output of the final attention block passes through the ReLU activation function and the pointwise convolutional layer, expanding the dimension of the sequence from to for C ISRJ component signal estimations C mask vectors, and then the size of the mask vectors is mapped to between 0 and 1 through the Sigmoid activation function, and these values between 0 and 1 represent the confidence levels of each dimension of the mask vectors belonging to different ISRJ component signals.

[0024] Please refer to Figure 6 , Figure 6 which is a schematic structural diagram of an attention block provided by an embodiment of the present invention. As shown in Figure 6 , the attention block in this embodiment adopts a hybrid architecture that combines one-dimensional convolution (1-DConv) with the attention mechanism, aiming to collaboratively extract local features and model global dependencies. Among them, the one-dimensional convolutional layer adopts the structure as shown in Figure 7Schematic diagram of the structure of the one-dimensional convolutional layer, which enhances the feature expression ability and gradient flow while maintaining computational efficiency. The attention design of this embodiment is based on the Transform architecture and combines the gated attention unit GAU for long sequence modeling to enhance the processing ability of complex ISRJ signals. The following is the core processing flow of this module: As Figure 6 shown, for the input sequence of the attention block, first, two one-dimensional convolutional layers (1-DConv) are used for preliminary feature extraction. The first convolution focuses on capturing local features (such as short-time spectrum features). Then, the distribution of the convolutional output is adjusted through the scale and offset operation (similar to layer normalization) to ensure feature stability. Subsequently, the rotary position encoding (RoPE) is introduced to embed absolute or relative position information into the feature vector to solve the problem of the lack of position sensitivity in the attention mechanism. Then, based on the normalized features, a query matrix Q and a key matrix K are generated, and an attention matrix is obtained through the improved attention calculation, , where is the query matrix, is the key matrix, is a learnable bias term, represents the rectified linear unit with square activation, and this design enhances the focusing ability of the attention weight on significant features through non-linear enhancement and square operation. Finally, the attention output is processed by the second one-dimensional convolutional layer (1-DConv) to achieve the deep fusion of local features and global context. At the same time, through the residual network, the final output layer generates a discriminative high-dimensional representation, providing an optimized feature space for the interference separation task.

[0025] Please continue to refer to Figure 5 , in this embodiment, the decoder includes a second one-dimensional convolutional layer (1-D Conv), and the second one-dimensional convolutional layer is the transposed convolutional layer of the first one-dimensional convolutional layer.

[0026] In the decoding stage, in order to restore the separated feature sequence to the time-domain signal, a transposed convolutional layer (also called a deconvolution layer) is used as the core component of the decoder. The design of this decoder is symmetric with the encoder and uses the same stride and kernel size as the encoder to ensure that the features can be effectively mapped back to the original input space in reverse. The input sequence of the decoder is multiplied element-wise by each mask vector to obtain the separated feature sequence, and the separated feature sequence is finally decoded into the time-domain signal by the decoder, that is, multiple ISRJ component signals are obtained.

[0027] It should be noted that for the data samples used in the training process of the composite ISRJ separation network in this embodiment, they can be obtained through the following methods. Collect echo signals, which are echo signals containing composite ISRJs, that is, including interference signals and target signals. Independently perform maximum normalization processing on each echo signal to uniformly compress the echo signal and the interference signal to the [0, 1] dynamic range. Stack the interference signal and the target signal along the channel dimension to construct a multi-channel signal matrix containing the target signal of the interference signal, and this multi-channel signal matrix serves as the label of the echo signal. Use the dynamic zero-padding technology to align sequences of different lengths, convert them into PyTorch tensors, and adjust the dimension order to form an input format suitable for the deep learning model.

[0028] In an alternative embodiment, step 3 includes: Step 3.1: According to the preset window length, calculate the dynamic threshold of each ISRJ component signal. The calculation formula for the dynamic threshold is: ; ; In the formula, is the dynamic threshold, is time, is the sensitivity coefficient, is the local statistic of the ISRJ component signal within the window length, represents the median operation, is the window length, represents the ISRJ component signal within the window length.

[0029] It can be understood that since the interference energies of different ISRJ component signals are different, this embodiment designs a dynamic threshold detection method based on interference energy, which tracks the amplitude change through local statistics and avoids the prior dependence of the fixed threshold.

[0030] Step 3.2: Compare each ISRJ component signal with the corresponding dynamic threshold, and obtain the binary ISRJ signal corresponding to each ISRJ component signal according to the comparison result. The binary ISRJ signal is calculated according to the following formula: ; In the formula, represents the binary ISRJ signal, where represents that the signal is in the forwarding state, represents that the signal is in the slicing state.

[0031] In an alternative embodiment, step 4 includes: Step 4.1: Obtain the set of slice widths and the set of forwarding widths corresponding to each ISRJ component signal according to each binarized ISRJ signal.

[0032] In this embodiment, by traversing the binarized ISRJ signal and recording the widths (number of sampling points) of each forwarding (interval of consecutive 1s) and slice (interval of consecutive 0s), the set of slice widths and the set of forwarding widths corresponding to the ISRJ component signal can be obtained.

[0033] Step 4.2: Screen the set of slice widths and the set of forwarding widths through the greedy clustering algorithm to obtain the set of dominant slice widths and the set of dominant forwarding widths corresponding to each ISRJ component signal.

[0034] In this embodiment, to suppress the influence of noise and outliers, define the similarity metric , as two different data values in the set to be screened, where is the similarity threshold, which can take the value of . If the absolute difference between two data points is less than or equal to the threshold , they are considered similar, otherwise they are not. Screen the set of slice widths and the set of forwarding widths respectively through the greedy clustering algorithm to obtain the set of dominant slice widths and the set of dominant forwarding widths.

[0035] Step 4.3: Perform density clustering optimization on the set of dominant slice widths and the set of dominant forwarding widths respectively to obtain the estimated value of the slice width and the estimated value of the forwarding width corresponding to each ISRJ component signal.

[0036] In this embodiment, in the set of dominant slice widths , find a value such that the number of samples contained in the neighborhood centered at with a radius of is the largest. Take the value of at this time as the estimated value of the slice width, which can be expressed as: ; In the formula, represents the estimated value of the slice width, represents the set of dominant slice widths, represents the similarity threshold, which is taken as here, are two different data values in the set of dominant slice widths .

[0037] In this embodiment, for the set of dominant forwarding widths Perform density clustering, remove clusters with the number of samples less than the threshold (such as single-sample clusters), and take the mean of the remaining samples to obtain an estimated value of the forwarding width. , which can be expressed as: ; In the formula, represents the estimated value of the forwarding width, represents the number of data in the set of dominant forwarding widths , is the data in the set of dominant forwarding widths after removing outliers.

[0038] Step 4.4: Calculate the estimated value of the corresponding forwarding times of each ISRJ component signal based on the estimated value of the slicing width and the estimated value of the forwarding width.

[0039] In this embodiment, according to the characteristics of intermittent sampling and forwarding interference, the estimated value of the forwarding times is calculated according to the following formula: ; In the formula, represents the estimated value of the forwarding times.

[0040] The single ISRJ component signal obtained through the composite ISRJ separation network can effectively suppress the influence of noise, but there will be a situation of time-domain slicing fragmentation. Therefore, in this embodiment, aiming at the characteristics of the separated ISRJ component signal, a robust measurement algorithm based on pulse detection and clustering optimization is proposed. This method combines adaptive dynamic threshold detection, multi-level pulse segmentation and clustering optimization algorithms, and can accurately calculate the interference parameters based on the characteristics of the separated ISRJ component signal.

[0041] To verify the effectiveness and stability of the intelligent extraction method for composite interference timing parameters proposed in the present invention, a series of simulation experiments of complex interference scenarios under different signal-to-interference-plus-noise ratio (SINR) conditions are set up for verification.

[0042] (1) Data set To be as close as possible to the actual situation, the interference parameters of each ISRJ component signal in the composite ISRJ signal are randomly generated, and the JNR is randomly jumped between 0 and 20 dB. The Matlab is used to generate double-interference and triple-interference data sets, among which 5000 are used for training, 1000 are used for verification, and 1000 are used for testing. The specific parameter configurations are shown in Table 1.

[0043] Table 1

[0044] (2) Training settings Optimize using the SI-SDR training loss, which is a loss function widely used to evaluate source separation tasks. Use the Adam optimizer to train the model for a total of 100 rounds, with a learning rate of 1e-3, a batch size of 8, and use gradient clipping to limit the norm of the training gradient to 5.

[0045] (3)Evaluation Metrics The measurement metrics are measured using the ISRJ interference parameter measurement accuracy, i.e., the slice width measurement accuracy , the forwarding width accuracy , and the forwarding count accuracy . The slice width and forwarding width errors are 0.15 and 0.25 . In addition, to measure the overall measurement effect, is defined, which represents the proportion of samples in which all interference parameters are accurately measured.

[0046] (4)Comparison Experiments Experiment ① completes the separation and measurement of compound ISRJ signals in two and three ISRJ scenarios. The simulation experiment test results of the method proposed in the present invention are compared with the methods in reference [1] (L. Liu, Q. Lv, J. Liu, Y. Wu, and Y. Quan, “Adaptive Parameter Estimation for Compound Interrupted Sampling and Repeating Jamming,” IEEE Signal Process. Lett., vol. 31, pp. 496–500, 2024) and reference [2] (Q. Lv, L. Liu, Y. Wu, M. Peng, and Y. Quan, “A Framework for Compound Interrupted Sampling Repeater Jamming Parameter Measurement,” IEEE Trans. Instrum. Meas., vol. 74, pp. 1–4, 2025).

[0047] Experiment ② analyzes the performance advantages of the method proposed in the present invention in measuring interference parameters under different JNR conditions. It includes two scenarios: double interference and triple interference.

[0048] Experiment ③ analyzes the performance advantages of the method proposed in the present invention under different training set ratios.

[0049] ① Experiment on Extracting Interference Parameters of Compound ISRJ Signals The method proposed in the present invention was compared with the DCPE-ISRJ in the ISRJ separation network in Document [1] and the CSNet-ISRJ in the ISRJ separation network in Document [2]. All three methods were tested in two scenarios: double interference and triple interference. Table 2 shows the test results.

[0050] Table 2

[0051] According to the results, compared with DCPE-ISRJ, the method proposed in the present invention achieved a 215.8% performance improvement in the double interference scenario and a 1183% performance improvement in the triple interference scenario. This is mainly because DCPE-ISRJ can only process interference signals with the same JNR, while the method proposed in the present invention can process interference signals with random JNR. Compared with CSNet-ISRJ, it achieved an 11.3% performance improvement in the double interference scenario and a 7.9% performance improvement in the triple interference scenario.

[0052] ② Performance comparison under different JNRs To comprehensively evaluate the performance of the method proposed in the present invention, Monte Carlo simulation experiments were carried out under different interference-to-noise ratios (JNRs). In the experimental scenario, the signal-to-noise ratio (SNR) was fixed at 0 dB, and the JNR value range was set to 5 dB, 10 dB, 15 dB, and 20 dB. Tests were carried out in both the double interference scenario and the multi-interference scenario, and a comparative analysis was made with the method proposed in Document [2].

[0053] Please refer to Figure 8 and Figure 9 , Figure 8 which is a comparison chart of the parameter estimation accuracy of the method proposed in the present invention under different JNR conditions in the double interference condition, Figure 9 and Figure 8 is a comparison chart of the parameter estimation accuracy of the method proposed in the present invention under different JNR conditions in the triple interference condition. As Figure 9 shown, in both the double interference and triple interference scenarios, the method proposed in the present invention showed significant performance improvement under low JNR conditions, with the parameter measurement accuracy reaching over 90% and excellent system stability.

[0054] ③ Influence of different training set sizes To evaluate the influence of the training set size on the model performance, the model was trained with sample sizes gradually increasing from 1000 to 5000, and its corresponding performance indicators were tested. In addition, a comparative analysis was made of the obtained results with the method proposed in Document [2]. The specific comparison results are as​​Figure 10 and Figure 11 as shown Figure 10 is a comparison chart of the parameter estimation accuracy of the method proposed in the present invention under different training set sizes under double interference conditions; Figure 11 is a comparison chart of the parameter estimation accuracy of the method proposed in the present invention under different training set sizes under triple interference conditions. It can be seen from the figure that when facing double interference and triple interference scenarios, as the scale of the training data set increases, the performance of both the method proposed in the present invention and the method in reference [2] shows a steady upward trend. However, it is worth noting that when the training sample set is small, the method proposed in the present invention shows higher accuracy compared to reference [2].

[0055] Using the method proposed in the present invention, a training set of only 2000 samples can achieve a performance equivalent to that of reference [2] using a training set of 5000 samples. This shows that the method proposed in the present invention is not only more effective on small-scale data sets, but also can significantly reduce the number of training samples required to achieve the desired model performance.

[0056] It should be noted that in this article, relational terms such as first and second 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 "comprising", "including" or any other variant are intended to cover non-exclusive inclusion, so that an article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed. Without further limitation, an element defined by the statement "including a..." does not exclude the existence of additional identical elements in the article or device including the element. "Connection" or "connected" and other similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The orientation or positional relationship indicated by "up", "down", "left", "right", etc. is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be construed as a limitation of the present invention.

[0057] In the description of this specification, the description referring to terms such as "one embodiment", "some embodiments", "examples", "specific examples", or "some examples", etc. means that the specific features 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 or characteristics described can 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.

[0058] 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. An intelligent extraction method for the timing parameters of radar composite jamming based on frequency agility, characterized in that, Including: Step 1: Preprocess the radar echo signal to obtain a preprocessed echo signal, where the radar echo signal is a frequency-agile radar echo signal containing composite ISRJ; Step 2: Input the preprocessed echo signal into a trained composite ISRJ separation network to obtain multiple separated ISRJ component signals, where the composite ISRJ separation network is an interference separation network that fuses convolutional enhancement and Transformer attention mechanism; Step 3: Calculate the dynamic threshold for each ISRJ component signal, and perform binarization processing on each ISRJ component signal according to the dynamic threshold calculation result to obtain a binarized ISRJ signal corresponding to each ISRJ component signal; Step 4: Use a clustering optimization algorithm to perform clustering optimization processing on each binarized ISRJ signal, and obtain an estimated value of the interference parameter of each ISRJ component signal according to the clustering optimization result.

2. The intelligent extraction method for the timing parameters of the radar composite interference based on frequency agility according to claim 1, wherein The said Step 1 includes: Perform maximum normalization processing on the radar echo signal, compress the radar echo signal to the [0, 1] dynamic range, and obtain the preprocessed echo signal.

3. The intelligent extraction method for the timing parameters of the radar composite interference based on frequency agile as claimed in claim 1, wherein, The composite ISRJ separation network includes: an encoder, a mask network, and a decoder, where The encoder is used to extract features from the preprocessed echo signal to obtain an encoded feature sequence; The mask network is used to implement a non-linear mapping from the encoded feature sequence to multiple groups of time-domain masks to obtain multiple mask vectors; The decoder is used to perform a deconvolution operation on the Kronecker product of the multiple mask vectors and the encoded feature sequence to obtain the multiple ISRJ component signals.

4. The intelligent extraction method for the timing parameters of the radar composite interference based on frequency agile as claimed in claim 3, wherein, The encoder includes a cascaded first one-dimensional convolutional layer and a ReLU layer.

5. The intelligent extraction method for the timing parameters of the radar composite interference based on frequency agility according to claim 3, wherein The mask network includes a normalization layer, a position encoding layer, a first pointwise convolutional layer, multiple attention blocks, a ReLU layer, a second pointwise convolutional layer, and a Sigmoid layer, where Among the multiple attention blocks, the output of the previous attention block is used as the input of the next attention block; The encoded feature sequence passes through the normalization layer and the position encoding layer to obtain a feature sequence retaining global temporal information, and the feature sequence retaining global temporal information is passed to the multiple attention blocks through the first pointwise convolutional layer; The multiple attention blocks use the self-attention mechanism to capture long-range dependencies in the input feature sequence, and the feature sequence output by the last attention block passes through the ReLU layer and the second pointwise convolutional layer to achieve dimension expansion, and the feature sequence after dimension expansion passes through the mapping of the Sigmoid layer to obtain the multiple mask vectors.

6. The intelligent extraction method for the timing parameters of radar composite interference based on frequency agility according to claim 4, wherein The decoder includes a second one-dimensional convolutional layer, and the second one-dimensional convolutional layer is the transposed convolutional layer of the first one-dimensional convolutional layer.

7. The intelligent extraction method for the timing parameters of radar composite jamming based on frequency agile as claimed in claim 1, wherein The said Step 3 includes: Step 3.1: Calculate the dynamic threshold for each ISRJ component signal according to a preset window length, and the calculation formula of the dynamic threshold is: ； ； wherein, is the dynamic threshold, is the time, is the sensitivity coefficient, is the local statistic of the ISRJ component signal within the window length, represents the median operation, is the window length, represents the ISRJ component signal within the window length; Step 3.2: Compare each ISRJ component signal with the corresponding dynamic threshold, and obtain the binarized ISRJ signal corresponding to each ISRJ component signal according to the comparison result. The binarized ISRJ signal is calculated according to the following formula: ； In the formula, represents the binarized ISRJ signal.

8. The intelligent extraction method for radar composite interference timing parameters based on frequency agility according to claim 1, characterized in that The interference parameters include the slice width, the forwarding width, and the number of forwarding times.

9. The intelligent extraction method for the timing parameters of radar composite jamming based on frequency agile as claimed in claim 8, wherein The step 4 includes: Step 4.1: Obtain the slice width set and the forwarding width set corresponding to each ISRJ component signal according to each binarized ISRJ signal; Step 4.2: Screen the slice width set and the forwarding width set through a greedy clustering algorithm to obtain the dominant slice width set and the dominant forwarding width set corresponding to each ISRJ component signal; Step 4.3: Perform density clustering optimization on the dominant slice width set and the dominant forwarding width set respectively to obtain the estimated value of the slice width and the estimated value of the forwarding width corresponding to each ISRJ component signal; Step 4.4: Calculate the estimated value of the number of forwarding times corresponding to each ISRJ component signal according to the estimated value of the slice width and the estimated value of the forwarding width.

10. The intelligent extraction method for the timing parameters of radar composite jamming based on frequency agile as claimed in claim 9, wherein The estimated value of the slice width is calculated according to the following formula: ； In the formula, represents an estimated value of the slice width, represents a set of dominant slice widths, represents a similarity threshold, is a set of dominant slice widths and two different data values in it; The estimated value of the forwarding width is calculated according to the following formula: ； In the formula, represents the estimated value of the forwarding width, represents the number of data in the set of dominant forwarding widths , is the data after removing outliers from the set of dominant forwarding widths . The estimated value of the number of forwarding times is calculated according to the following formula: ； In the formula, represents the estimated value of the number of forwarding times.

Images