Radar narrow-band interference recognition method and system based on time-frequency energy multi-domain slow-varying characteristics

Abstract

The application relates to a radar narrow-band interference recognition method and system based on a time-frequency energy multi-domain slow change feature. In view of the fact that traditional radar interference type confirmation and anti-interference measure selection more depend on artificial judgment and setting, and there are defects of ambiguity and subjectivity, a method for accurately classifying four types of typical narrow-band interference, i.e. narrow pulse interference, suppression interference, deception interference and other interference is provided. The specific steps include the following processes: interference detection area dynamic planning, envelope level multi-dimensional feature fine perception, point track level parameter robust measurement and self-adaptive enhancement classification recognition. The application effectively enhances the automatic selection ability of the anti-interference measure, and improves the survival ability and combat effectiveness of the radar in a complex electromagnetic environment.

Description

Technical Field

[0001] This invention relates to the field of radar interference signal classification and identification, and in particular to a radar narrowband interference identification method and system based on time-frequency energy multi-domain slow-varying characteristics. Background Technology

[0002] Interference environment cognition requires the extraction, analysis, and classification of interference source characteristics to identify specific interference patterns, thus providing prior information for the optimal selection of anti-interference measures. The patent "Method and System for Interference Type Identification" (CN201711425496.7) uses an interference data training set to train a neural network model for interference type identification, obtaining an interference type identification model and thus achieving interference type identification. This method has high accuracy in identifying interference types for known samples, but it is difficult to adapt to the time-varying nature of the interference environment. The literature "Research and Implementation of Radar Active Interference Signal Identification Method" (Beijing Institute of Technology) uses digital channelization technology to obtain the time-frequency domain characteristic parameters and pulse descriptor parameters of common radar active interference signals, and designs a basic... There are common active jamming signal identification software for radar in DSP, but this method achieves wideband jamming identification from the perspective of electronic reconnaissance, but the number of parameters used is limited, and the application scenarios and recognition accuracy are insufficient. The patent "A method and system for identifying radar active jamming" (CN201810489421.3) performs variational mode decomposition on the received active jamming to obtain different mode components, calculates the corresponding rectangular integral bispectrum of each intrinsic mode component, combines them into feature parameters, and realizes the identification of radar active jamming mode through random forest classifier. However, the mode decomposition method is highly sensitive to the signal. Summary of the Invention

[0003] To overcome the shortcomings of traditional radar interference identification, which relies more on offline intelligent analysis of known samples and cannot adapt to time-varying environments, has high sensitivity of selected features but unstable identification, and has a limited number of parameters and insufficient adaptability, this paper proposes a radar narrowband interference identification method and system based on parameterized decision trees. This method extracts multi-dimensional slowly varying time-frequency energy features at the radar echo envelope level and point trace level.

[0004] This application provides a radar narrowband interference identification method based on time-frequency energy multi-domain slowly varying characteristics, the method including:

[0005] Step 1: Dynamic planning of the interference detection area;

[0006] Step 2, envelope-level multidimensional feature refinement perception, completes the mapping from the low-dimensional echo envelope space to the high-dimensional discrimination domain feature space; among which, envelope-level multidimensional feature refinement perception includes: envelope parameter extraction after intermediate frequency modulus calculation and envelope parameter extraction after pulse compression modulus calculation;

[0007] Step 3: Robust measurement of point-level parameters to complete the mapping from the low-dimensional point space to the high-dimensional discrimination domain parameter space;

[0008] Step 4: Based on the AdaBoost algorithm, perform adaptive enhancement classification and recognition to classify narrow pulse interference, suppression interference, deceptive interference and other types of interference.

[0009] Optionally, step 1, dynamic programming of the interference detection area, includes:

[0010] Set the starting distance of the interference detection zone:

[0011]

[0012] Where R represents the radar line-of-sight range in kilometers; H represents the radar elevation in meters; R set This indicates the distance after secondary correction, in kilometers.

[0013] Optionally, in step 2, during the refined perception of envelope-level multidimensional features,

[0014] The parameters extracted from the envelope after intermediate frequency mode calculation include: channel echo average amplitude A1, channel over-threshold interference duty cycle D1, and number of narrow pulses N. np Narrow pulse distance starting B np Narrow pulse distance widening W np Narrow pulse average amplitude A np Number of equally spaced narrow pulses N enp Equal-interval narrow pulse amplitude A enp Number of wide pulses N wp Wide pulse distance starting B wp Wide pulse distance extension W wp Wide pulse average amplitude A wp Number of equally spaced wide pulses N ewp Equal-interval wide pulse amplitude A ewp In a wide pulse vector, the pulse with the median width is interference, and the interference distance is from the initial B. s , Screened interference distance broadening W s The average amplitude of the selected interference A s The main lobe width W of the selected interference autocorrelation function corr The power spectrum energy percentage P within the selected interference band spec The flatness P of the selected interference time-domain waveform s ;

[0015] The parameters extracted from the envelope after pulse compression modeling include: the average amplitude of the echo in the pulse-compressed channel A2 and the duty cycle of the pulse-compressed channel over the threshold interference D2.

[0016] Optionally, step 2, envelope-level multidimensional feature refinement perception, completes the mapping from the low-dimensional echo envelope space to the high-dimensional discrimination domain feature space, including:

[0017] Step 21: First, calculate the average amplitude A1 of the channel echo, then perform threshold envelope extraction:

[0018] The method for extracting the channel echo threshold envelope is as follows: First, obtain the background threshold by fixing, averaging, or using a constant false alarm rate (CFAR) method. Set the current echo amplitude above the background threshold to 1, and the rest to 0. Then, use the N / M criterion to determine the starting and ending points of the envelope.

[0019]

[0020] Where S(n) represents the threshold identifier of the nth distance unit, and Z(n) represents the connectivity identifier of the nth distance unit; the connected data Z(n) is merged and aggregated, that is, the number of data segments with consecutive 1s is counted; the channel threshold interference duty cycle D1 is counted, that is, the ratio of the number of elements with Z as 1 to the total number;

[0021] Step 22: Set narrow pulse width threshold and wide pulse width threshold respectively, and extract parameters at equal intervals;

[0022] Set narrow pulse width threshold T n_w The length of the connected envelope in Z is less than that in T. n_w That is, a narrow pulse; T n_w The pulse width (PW) of the current radar transmitted waveform is set by the following ratio:

[0023]

[0024] Among them, T wave1 This indicates the set pulse width threshold. When the pulse width is less than the threshold, the narrow pulse width threshold is set to a fixed value. When the pulse width is greater than the threshold, the narrow pulse width threshold is set to k1 times the pulse width of the current radar transmitted waveform. Narrow pulse envelope parameters are calculated: number of narrow pulses N. np Narrow pulse distance starting B np Narrow pulse distance widening W np Narrow pulse average amplitude A np ;

[0025] Calculate the difference between the starting positions of any two narrow pulses, project the difference onto a histogram, count the number of differences in the histogram cells, and take the one with the most differences as the result of equal-interval extraction.

[0026] The narrow pulse distance starts as the distance unit I of this envelope. mj Set the distance range for extracting the focus at equal intervals to [I]. min ,Imax If the width of the equally spaced histogram cells is w, then the number of histogram cells is:

[0027] N unit =floor((I max -I min ) / w)+1 (6)

[0028] Where floor(·) represents rounding to negative infinity; the center of each distance interval histogram cell is:

[0029] I c = (c-0.5)×w+I min (7)

[0030] Where c is the histogram grid index, and its range is 1≤c≤N. unit ; Traverse all narrow pulse distance units {I ma ,I mb}, initially set a∈[1,N np ], b = a + 1, calculate the distance interval, i.e., the distance difference I. delat =|I ma -I mb |, if I min ≤I delat ≤I max If the value is not found, it will be mapped to a histogram cell, and the count of the corresponding histogram cell will be incremented by one.

[0031] c = floor((I delta -I min ) / w)+1 (8)

[0032] S(c)=S(c)+1 (9)

[0033] Where S(c) represents the number of elements in the c-th histogram cell;

[0034] Reset a∈[1,N np b = a + d, where d is the distance interval order. Repeat the distance difference calculation and histogram mapping steps in the above narrow pulse equal interval extraction process until the traversal ends. S(c) is the distance difference histogram statistical result. If S(c) is greater than the set threshold, then count the number of equally spaced narrow pulses N. enp Equal-interval narrow pulse amplitude A enp ;

[0035] Set the wide pulse width threshold T w_w The width threshold of the wide pulse is set to be between k2 and k3 times the pulse width of the current radar transmitted waveform.

[0036] k2*PW≤Tw_w ≤k3*PW (10)

[0037] Statistical parameters of wide pulse envelope: number of wide pulses N wp Wide pulse distance starting B wp Wide pulse distance extension W wp Wide pulse average amplitude A wp ;

[0038] The number N of equally spaced wide pulses is extracted using the narrow pulse equally spaced extraction method. ewp Equal-interval wide pulse amplitude A ewp ;

[0039] Interference is filtered out by the median of the wide pulse range broadening vector; that is, the envelope with the median width among all wide pulses is selected as the extracted interference envelope, and the starting B of the filtered interference range is then calculated. s , Screened interference distance broadening W s The average amplitude of the selected interference A s The main lobe width W of the selected interference autocorrelation function corr The power spectrum energy percentage P within the selected interference band spec The flatness P of the selected interference time-domain waveform s The statistics show that the flatness of the time-domain waveform is defined as the variance of the interference time-domain waveform.

[0040] Step 23: Perform pulse compression processing on the radar echo signal; the parameters extracted from the envelope after pulse compression include: the average amplitude A2 of the echo in the pulse-compressed channel and the duty cycle D2 of the pulse-compressed channel when it passes the threshold interference.

[0041] Optionally, after the robust measurement of the point-level parameters in step 3 is performed, the radar echo, after passing constant false alarm rate (CFAR) detection, measures the point-level parameters, including:

[0042] Number of dots N plot The center position of the dot C plot Dot distance widening W plot Average amplitude of the dot A plot Number of equally spaced dots N e_plot Average amplitude A of equally spaced dots e_plot Maximum density of dots D plot .

[0043] Optionally, the adaptive augmentation classification recognition in step 4 includes:

[0044] The envelope-level multidimensional features and point-level parameters extracted in steps 2 and 3 are used as the classification basis, and a decision tree is selected as the base learner. Then, the AdaBoost ensemble optimization algorithm is used to learn and correct the prediction. Finally, the recognition result is obtained by combining the features.

[0045] This application also provides a radar narrowband interference identification system based on time-frequency energy multi-domain slowly varying characteristics, the system including...

[0046] The planning module is used for dynamic planning of the interference detection area;

[0047] The perception module is used for refined perception of envelope-level multidimensional features, completing the mapping from the low-dimensional echo envelope space to the high-dimensional discrimination domain feature space; among which, the refined perception of envelope-level multidimensional features includes: extraction of envelope parameters after intermediate frequency modulus calculation and extraction of envelope parameters after pulse compression modulus calculation;

[0048] The measurement module is used to perform robust measurements of point-level parameters and complete the mapping from the low-dimensional point space to the high-dimensional discrimination domain parameter space.

[0049] The classification module is used to perform adaptive enhanced classification and recognition based on the AdaBoost algorithm, realizing the classification of narrow pulse interference, suppression interference, deceptive interference and other types of interference.

[0050] Optionally, the planning module is specifically used for:

[0051] Set the starting distance of the interference detection zone:

[0052]

[0053] Where R represents the radar line-of-sight range in kilometers; H represents the radar elevation in meters; R set This indicates the distance after secondary correction, in kilometers.

[0054] This application also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor performs the steps of the method of this application.

[0055] This application also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the method of this application.

[0056] The beneficial effects of this invention are as follows: The data is filtered by using a dynamic programming method for interference detection area, eliminating invalid data that is contaminated by clutter in the near range, saving processing resources and improving the purity of the identification samples; the multidimensional slow-varying features of envelope-level and point-level time-frequency energy are used to reduce the sensitivity of the identification algorithm to time-varying environments and improve the stability and adaptability of the identification. Attached Figure Description

[0057] Figure 1 This is a flowchart of a radar narrowband interference identification method based on time-frequency energy multi-domain slow-varying characteristics; wherein, S1-S4 in the figure correspond to steps 1-4 of the process described in the invention.

[0058] Figure 2 The echo modulus values ​​before and after pulse compression are shown for four typical narrowband interference types: narrow pulse interference, suppression interference, deceptive interference, and other types of interference. Detailed Implementation

[0059] The present invention will be further explained and described below with reference to the accompanying drawings and embodiments.

[0060] The implementation process of this invention is as follows: Figure 1 As shown, the preferred implementation process is specifically described as follows:

[0061] Step 1: Dynamic planning of interference detection area.

[0062] To reduce the impact of strong short-range clutter on the selection of interference samples, the starting distance of the interference detection zone is related to the radar's own elevation. The starting distance of the interference detection zone is set as follows:

[0063]

[0064] Where R represents the radar line-of-sight range in kilometers; H represents the radar elevation in meters; R set This indicates the distance after secondary correction, in kilometers.

[0065] Step 2: Envelope-level multidimensional feature refinement perception, completing the mapping from the low-dimensional echo envelope space to the high-dimensional discrimination domain feature space.

[0066] Envelope-level multidimensional feature refinement perception includes: envelope parameter extraction after mid-frequency modulus calculation and envelope parameter extraction after pulse compression modulus calculation.

[0067] First, the average amplitude A1 of the channel echo is calculated, and then the threshold envelope is extracted. The method for extracting the threshold envelope of the channel echo is as follows: First, the background threshold is obtained by fixing, using the mean, or constant false alarm rate. The current echo amplitude is set to 1 if it is higher than the background threshold, and the rest are set to 0. Then, the N / M criterion is used to determine the starting and ending points of the envelope.

[0068]

[0069] Where S(n) represents the threshold identifier of the nth distance unit, and Z(n) represents the connectivity identifier of the nth distance unit; the connected data Z(n) is merged and aggregated, that is, the number of data segments with consecutive 1s is counted. The channel threshold interference duty cycle D1 is counted, that is, the ratio of the number of elements with Z=1 to the total number.

[0070] Set narrow pulse width threshold T n_w The length of the connected envelope in Z is less than that in T. n_w This is a narrow pulse. T n_wThe pulse width (PW) of the current radar transmitted waveform can be set by adjusting the ratio:

[0071]

[0072] Among them, T wave1 This indicates the set pulse width threshold. When the pulse width is less than a certain threshold, the narrow pulse width threshold is set to a fixed value. When the pulse width is greater than a certain threshold, the narrow pulse width threshold is set to k1 times the pulse width of the current radar transmitted waveform. Narrow pulse envelope parameters are calculated: number of narrow pulses N. np Narrow pulse distance starting B np Narrow pulse distance widening W np Narrow pulse average amplitude A np .

[0073] Calculate the difference between the starting positions of any two narrow pulses, project the difference onto a histogram, count the number of differences in the histogram cells, and take the one with the most differences as the result of equal-interval extraction.

[0074] The narrow pulse distance starts as the distance unit I of this envelope. mj Set the distance range for extracting the focus at equal intervals to [I]. min ,I max If the width of the equally spaced histogram cells is w, then the number of histogram cells is:

[0075] N unit =floor((I max -I min ) / w)+1 (6)

[0076] Where floor(·) represents rounding to negative infinity; the center of each distance interval histogram cell is:

[0077] I c = (c-0.5)×w+I min (7)

[0078] Where c is the histogram grid index, and its range is 1≤c≤N. unit ; Traverse all narrow pulse distance units {I ma ,I mb}, initially set a∈[1,N np ], b = a + 1, calculate the distance interval, i.e., the distance difference I. delat =|I ma -I mb |, if I min ≤I delat ≤I max If the value is not found, it will be mapped to a histogram cell, and the count of the corresponding histogram cell will be incremented by one.

[0079] c = floor((I delta -I min ) / w)+1 (8)

[0080] S(c)=S(c)+1 (9)

[0081] Where S(c) represents the number of elements in the c-th histogram cell;

[0082] Reset b = a + d, where d is the distance interval order. Repeat the distance difference calculation and histogram mapping steps in the narrow pulse equal-interval extraction process for all point pairs until the traversal ends. S(c) is the distance difference histogram statistical result. If S(c) is greater than the set threshold, then count the number N of equal-interval narrow pulses. enp Equal-interval narrow pulse amplitude A enp .

[0083] Set the wide pulse width threshold T w_w The width threshold of the wide pulse is set to be between k2 and k3 times the pulse width of the current radar transmitted waveform.

[0084] k2*PW≤T w_w ≤k3*PW (10)

[0085] Statistical parameters of wide pulse envelope: number of wide pulses N wp Wide pulse distance starting B wp Wide pulse distance extension W wp Wide pulse average amplitude A wp .

[0086] The number N of equally spaced wide pulses is extracted using the narrow pulse equally spaced extraction method. ewp Equal-interval wide pulse amplitude A ewp .

[0087] Interference can be filtered out by taking the median of the wide pulse range broadening vector. Specifically, the envelope with the median width among all wide pulses is selected as the extracted interference envelope, and the starting distance B for filtering out interference is calculated. s , Screened interference distance broadening W s The average amplitude of the selected interference A s The main lobe width W of the selected interference autocorrelation function corr The power spectrum energy percentage P within the selected interference band spec The flatness P of the selected interference time-domain waveform s The statistics show that the flatness of the time-domain waveform can be defined as the variance of the disturbance time-domain waveform.

[0088] The radar echo signal is processed by pulse compression. The parameters extracted from the envelope after pulse compression and modulus calculation mainly include: the average amplitude A2 of the echo in the pulse-compressed channel and the duty cycle D2 of the pulse-compressed channel when it passes the threshold interference.

[0089] Step 3: Robust measurement of dot-level parameters.

[0090] This achieves mapping from a low-dimensional point space to a high-dimensional discrimination domain parameter space. After constant false alarm rate (CFAR) detection, the radar echoes are used to measure point-level parameters, primarily including the number of points N. plot The center position of the dot C plot Dot distance widening W plot Average amplitude of the dot A plot Number of equally spaced dots N e_plot Average amplitude A of equally spaced dots e_plot Maximum density of dots D plot .

[0091] The maximum density of a dot is defined as the density of a dot centered at a given dot within a radius of K. set The number of all points within a kilometer.

[0092] Step 4: Adaptive Enhanced Classification and Recognition. The envelope-level multidimensional features and point-level parameters extracted in steps S2 and S3 are used as the classification basis. A decision tree is selected as the base learner. Then, based on the AdaBoost ensemble optimization algorithm, a base learner is built to learn and correct predictions. Finally, a more comprehensive recognition result is obtained through combination.

[0093] This application also provides a radar narrowband interference identification system based on time-frequency energy multi-domain slowly varying characteristics, the system including...

[0094] The planning module is used for dynamic planning of the interference detection area;

[0095] The perception module is used for refined perception of envelope-level multidimensional features, completing the mapping from the low-dimensional echo envelope space to the high-dimensional discrimination domain feature space; among which, the refined perception of envelope-level multidimensional features includes: extraction of envelope parameters after intermediate frequency modulus calculation and extraction of envelope parameters after pulse compression modulus calculation;

[0096] The measurement module is used to perform robust measurements of point-level parameters and complete the mapping from the low-dimensional point space to the high-dimensional discrimination domain parameter space.

[0097] The classification module is used to perform adaptive enhanced classification and recognition based on the AdaBoost algorithm, realizing the classification of narrow pulse interference, suppression interference, deceptive interference and other types of interference.

[0098] Optionally, the planning module is specifically used for:

[0099] Set the starting distance of the interference detection zone:

[0100]

[0101] Where R represents the radar line-of-sight range in kilometers; H represents the radar elevation in meters; R set This indicates the distance after secondary correction, in kilometers.

[0102] This application also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor performs the steps of the method of this application.

[0103] This application also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the method of this application.

[0104] like Figure 2 As shown, the basic characteristics of narrow pulse jamming signals in the time domain are "short pulse width and high energy." The energy of noise jamming is uniform in the power spectrum and uncorrelated between any two different times in the time domain. Therefore, the autocorrelation function exhibits impulse characteristics, and pulse compression processing does not provide additional energy gain for noise jamming. Dense false target jamming is generated by electronic warfare receivers using DRFM technology to intercept and relay radar signals. It is similar to real targets in the time and frequency domains and can form stable point tracks, usually evenly distributed. Other types of jamming include dense convolution, which are generated by electronic warfare receivers using DRFM technology to intercept radar signals and modulate them in the time, frequency, amplitude, and phase domains. The method provided in this application has good effects on the above-mentioned jamming. This application uses a dynamic programming method for the jamming detection area to filter data, eliminating invalid data contaminated by clutter in the near range, saving processing resources and improving the purity of the identification samples. It uses envelope-level and point track-level time-frequency energy multidimensional slow-varying features to reduce the sensitivity of the identification algorithm to time-varying environments and improve the identification stability and adaptability.

Claims

1. A radar narrowband interference identification method based on time-frequency energy multi-domain slowly varying characteristics, characterized in that, The method includes: Step 1: Dynamic planning of the interference detection area; Step 2, envelope-level multidimensional feature refinement perception, completes the mapping from the low-dimensional echo envelope space to the high-dimensional discrimination domain feature space; among which, envelope-level multidimensional feature refinement perception includes: envelope parameter extraction after intermediate frequency modulus calculation and envelope parameter extraction after pulse compression modulus calculation; Step 3: Robust measurement of point-level parameters to complete the mapping from the low-dimensional point space to the high-dimensional discrimination domain parameter space; Step 4: Based on the AdaBoost algorithm, perform adaptive enhancement classification and recognition to classify narrow pulse interference, suppression interference, deceptive interference and other types of interference. Adaptive augmented classification and recognition, including: The envelope-level multidimensional features and point-level parameters extracted in steps 2 and 3 are used as the classification basis, and a decision tree is selected as the base learner. Then, the AdaBoost ensemble optimization algorithm is used to learn and correct the prediction. Finally, the recognition result is obtained by combining the features.

2. The radar narrowband interference identification method based on time-frequency energy multi-domain slowly varying characteristics according to claim 1, characterized in that, Step 1, dynamic planning of the interference detection area includes: Set the starting distance of the interference detection zone: (1)； in, Indicates radar line-of-sight range, in kilometers; Indicates radar elevation, unit: meters; This indicates the distance after secondary correction, in kilometers.

3. The radar narrowband interference identification method based on time-frequency energy multi-domain slowly varying characteristics according to claim 1, characterized in that, In step 2, during the refined perception of envelope-level multidimensional features, The parameters extracted from the envelope after intermediate frequency mode calculation include: average amplitude of channel echo. Channel threshold interference duty cycle Number of narrow pulses Narrow pulse distance start Narrow pulse distance widening Narrow pulse average amplitude Number of equally spaced narrow pulses Equal-interval narrow pulse amplitude Number of wide pulses Wide pulse distance start Wide pulse distance extension Wide pulse average amplitude Number of equally spaced wide pulses Equal-interval wide pulse amplitude In a wide pulse vector, the pulse with the median width is considered interference, and the interference distance is initially... , Broadening the selected interference distance Average amplitude of the selected interference The main lobe width of the selected interference autocorrelation function The power spectrum energy ratio within the selected interference band Flatness of the selected interference time-domain waveform ; The parameters extracted from the envelope after pulse compression modeling include: the average amplitude of the echo in the pulse compression channel. Post-pulse compression channel threshold interference duty cycle .

4. The radar narrowband interference identification method based on time-frequency energy multi-domain slow-varying characteristics according to claim 1, characterized in that, Step 2, envelope-level multidimensional feature refinement perception, completes the mapping from the low-dimensional echo envelope space to the high-dimensional discrimination domain feature space, including: Step 21: First, calculate the average amplitude of the channel echo. Then perform threshold envelope extraction: The method for extracting the channel echo threshold envelope is as follows: First, obtain the background threshold using a fixed, average, or constant false alarm rate (CFAR) method. Set the current echo amplitude above the background threshold to 1, and the rest to 0. Then, use... The criteria for determining the starting and ending points of the envelope are as follows: (4)； in, Indicates the first Each distance unit exceeds the threshold indicator. Indicates the first Each distance unit connectivity identifier; for the connected data Merge and consolidate data, i.e., count the number of consecutive 1s in a data segment; count the duty cycle of the channel when it exceeds the threshold interference. ,Right now The ratio of the number of elements equal to 1 to the total number of elements; Step 22: Set narrow pulse width threshold and wide pulse width threshold respectively, and extract parameters at equal intervals; Set narrow pulse width threshold , The length of the connected envelope is less than That is, a narrow pulse; The pulse width of the current radar transmission waveform Scale settings: (5)； in, This indicates the set pulse width threshold. When the pulse width is less than the threshold, the narrow pulse width threshold is set to a fixed value. When the pulse width is greater than the threshold, the narrow pulse width threshold is set to a value equal to the pulse width of the current radar transmitted waveform. Times; Statistical narrow pulse envelope parameters: number of narrow pulses Narrow pulse distance start Narrow pulse distance widening Narrow pulse average amplitude ; The difference between the starting positions of any two narrow pulses is calculated, and the difference is projected onto a histogram. The number of differences in the histogram cells is counted, and the result with the largest difference is taken as the result of equal interval extraction. The narrow pulse distance starts as the distance unit of this envelope. Set the distance range for extracting the focus at equal intervals as follows: The width of the equally spaced histogram grid is Then the number of cells in the histogram is: (6)； in, This indicates rounding towards negative infinity; the center of each histogram cell at each distance interval is: (7)； Where c is the histogram grid index, ranging from... ; Traverse all narrow pulse distance unit pairs Initial settings , Calculate the distance interval, i.e., the distance difference. ,like If the value is mapped to a histogram cell, the count of the corresponding histogram cell is incremented by one. (8)； (9)； in, Indicates the first The number of elements in each histogram cell; Reset , , To determine the distance interval order, iterate through all point pairs again, repeating the distance difference calculation and histogram mapping steps in the narrow pulse equal interval extraction process until the traversal is complete. This is the statistical result of the distance difference histogram; if If the value exceeds the set threshold, the number of equally spaced narrow pulses will be counted. Equal-interval narrow pulse amplitude ; Set wide pulse width threshold The width threshold of the wide pulse is set to the pulse width of the current radar transmitted waveform. Doubled Between times: (10)； Statistical parameters of wide pulse envelope: number of wide pulses Wide pulse distance start Wide pulse distance extension Wide pulse average amplitude ; The number of equally spaced wide pulses is extracted using the narrow pulse equally spaced extraction method. Equal-interval wide pulse amplitude ; Interference is filtered out by using the median of the wide-pulse range-stretched vector. Specifically, the envelope with the median width among all wide pulses is selected as the extracted interference envelope, and the starting range of the filtered interference is determined. , Broadening the selected interference distance Average amplitude of the selected interference The main lobe width of the selected interference autocorrelation function The power spectrum energy ratio within the selected interference band Flatness of the selected interference time-domain waveform The statistics show that the flatness of the time-domain waveform is defined as the variance of the interference time-domain waveform; Step 23: Perform pulse compression processing on the radar echo signal; the parameters extracted from the envelope after pulse compression and modulus calculation include: the average amplitude of the channel echo after pulse compression. Post-pulse compression channel threshold interference duty cycle .

5. The radar narrowband interference identification method based on time-frequency energy multi-domain slowly varying characteristics according to claim 1, characterized in that, After the robust measurement of the point-level parameters in step 3 is performed, the radar echo, after passing constant false alarm rate (CFAR) detection, measures the point-level parameters, including: Number of dots The center position of the dot Widening of dot distance Average amplitude of dots Number of equally spaced dots Average amplitude of equally spaced dots Maximum density of dots .

6. A radar narrowband interference identification system based on time-frequency energy multi-domain slow-varying characteristics, characterized in that, The system includes The planning module is used for dynamic planning of the interference detection area; The perception module is used for refined perception of envelope-level multidimensional features, completing the mapping from the low-dimensional echo envelope space to the high-dimensional discrimination domain feature space; among which, the refined perception of envelope-level multidimensional features includes: extraction of envelope parameters after intermediate frequency modulus calculation and extraction of envelope parameters after pulse compression modulus calculation; The measurement module is used to perform robust measurements of point-level parameters and complete the mapping from the low-dimensional point space to the high-dimensional discrimination domain parameter space. The classification module is used to perform adaptive enhanced classification and recognition based on the AdaBoost algorithm, realizing the classification of narrow pulse interference, suppression interference, deceptive interference and other types of interference; Adaptive augmented classification and recognition, including: The envelope-level multidimensional features and point-level parameters extracted by the perception and measurement modules are used as the classification basis, and a decision tree is selected as the base learner. Then, the AdaBoost ensemble optimization algorithm is used to learn and correct the prediction. Finally, the recognition result is obtained by combining the features.

7. The radar narrowband interference identification system based on time-frequency energy multi-domain slow-varying characteristics according to claim 6, characterized in that, The planning module is specifically used for: Set the starting distance of the interference detection zone: (1)； in, Indicates radar line-of-sight range, in kilometers; Indicates radar elevation, unit: meters; This indicates the distance after secondary correction, in kilometers.

8. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 5.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 5.

Images