Foldover clutter suppression method for airborne radar based on range-gating and subspace projection

Abstract

The application discloses a method for suppressing airborne radar foldover clutter based on range gating and subspace projection, and relates to the technical field of airborne radar signal processing.The method uses an inter-pulse agile waveform with a range gating characteristic, performs subspace projection on the echo in the range dimension after the echo in the range dimension is gated, obtains a range section clutter reconstruction result, and iterates between range sections to eliminate mutual influence between clutters in the range sections, so that the clutters in the range sections are reconstructed, the clutters in non-range sections are cancelled from the received echo to suppress the foldover clutter, and target detection is performed, thereby effectively improving the detection capability of the airborne radar for a long-distance low-speed weak target.The method has many advantages, such as simple implementation, low calculation complexity, good foldover clutter suppression effect, and the like.

Description

Technical Field

[0001] This invention relates to the field of airborne radar signal processing technology, and specifically to an airborne radar folding clutter suppression method based on range gating and subspace projection. Background Technology

[0002] Ground-based radar is limited in detection range due to the curvature of the Earth. To achieve greater detection range, radar has been mounted on aircraft, leading to the development of airborne radar. However, in look-down search scenarios, airborne radar suffers from severe Doppler expansion of the clutter spectrum due to platform movement, significantly compressing the clear area in the Doppler dimension after Moving Target Detection (MTD) and Pulse Doppler (PD) processing. Therefore, airborne radar mostly operates at medium to high repetition rates (RF). However, because the parameters of the transmitted waveform are fixed, medium to high RF means severe range ambiguity, causing clutter echoes within each pulse repetition time (PRT) to fold along the fast time dimension. This results in distant targets needing to compete with strong nearby clutter, reducing the detection capability of airborne radar for distant, low-speed, and weak targets. To improve this problem, the most widely used method currently is Space-Time Adaptive Signal Processing (STAP) technology. STAP is a space-time two-dimensional filtering technique that constructs an optimal space-time two-dimensional filter by calculating the covariance matrix of clutter plus noise using independent and identically distributed training samples without a target. However, according to the RMB criterion, obtaining a space-time filter with a performance loss of less than 3dB relative to the optimal filter requires independent and identically distributed training samples with twice the system degrees of freedom. But in complex real-world scenarios, non-uniformity, non-stationarity, and system distortion make it difficult to obtain satisfactory data samples, resulting in a significant reduction in STAP performance in practice.

[0003] Therefore, the folding clutter caused by fixed waveform parameters remains a problem that airborne radar urgently needs to solve for long-range detection. Summary of the Invention

[0004] In view of this, the present invention provides an airborne radar folded clutter suppression method based on range gating and subspace projection, which can suppress folded clutter caused by fixed waveform parameters, with low computational complexity and good folded clutter suppression effect.

[0005] To achieve the above objectives, the technical solution of the present invention is an airborne radar clutter suppression method based on range gating and subspace projection, comprising the following steps:

[0006] Step 1: Estimate the distance range of clutter distribution based on the scene.

[0007] Step 2: Determine the clutter Doppler coverage area based on radar parameters.

[0008] Step 3: Vectorize the total echo along the fast time dimension to obtain the vectorized total echo.

[0009] Step 4: Determine the maximum number of range segments P in the clutter echo distribution based on the radar parameters; and assign each range segment to be reconstructed as a variable p, with p taking values ​​of 1, 2, ..., P, and initialize p = 1.

[0010] Step 5: Based on the geometric relationship between the radar antenna and the clutter scattering unit, determine the clutter subspace and construct the clutter projection matrix for the p-th range segment.

[0011] Step 6: Set the receiving filter bank corresponding to the p-th distance segment.

[0012] Step 7: Reconstruct the clutter of each range segment by subspace projection to obtain the clutter reconstruction result of the p-th range segment.

[0013] Step 8: Increment p by 1. If p > P after the update, then reset p = 1 and repeat steps 5 to 7 to achieve inter-segment cyclic iteration until the iteration process converges.

[0014] Step 9: Subtract the reconstructed clutter from the non-range segment clutter using the vectorized total echo for each range segment to obtain the clutter reconstruction result for each range segment. Then, perform inverse vectorization along the fast time dimension to complete the accurate reconstruction of clutter in each range segment and the effective suppression of folded clutter.

[0015] Step 10: After subtracting the reconstructed clutter from each range segment using the total echo, detect targets in the scene by range segment. For range segments where targets exist, construct the target subspace, update the subspace of the range segment where the target is located, and update the projection matrix of the range segment where the target is located. Repeat steps 7-10 until the total echo minus the reconstruction result of each range segment is less than the set threshold, and the target detection is complete.

[0016] Further, step 1: Estimate the distance range of clutter distribution based on the scene, specifically as follows:

[0017] The maximum detection range of the airborne radar is [R] min ,R max The specific relationship is as shown in equation (1):

[0018]

[0019] Where H is the flight altitude, P t For radar transmission power, G t For the transmit antenna gain, A rS is the effective receiving area of ​​the antenna, σ is the scattering cross-section of the radar illumination area, and S min This is the minimum detectable signal for the receiver;

[0020] [R min ,R max This refers to the distance range of clutter distribution.

[0021] Further, step 2: Determine the clutter Doppler coverage area based on radar parameters, specifically as follows:

[0022] If the airborne radar's flight speed relative to the ground is V and its wavelength is λ, then the clutter Doppler coverage area [f] is... dmin ,f dmax Calculated using equation (2):

[0023]

[0024] Further, step 3: Vectorize the total echo along the fast time dimension, specifically as follows:

[0025] The total echo is O, which is a three-dimensional matrix with dimensions M×N×PRTnum, where M corresponds to the number of pulses, N corresponds to the number of array elements, and PRTnum corresponds to the number of sampling points in a PRT. Vectorize O along the fast time dimension to transform the three-dimensional matrix into a two-dimensional matrix and obtain the vectorized total echo Vec(O).

[0026] The dimension of Vec(O) is MN×PRTnum.

[0027] 5. The airborne radar clutter suppression method based on range gating and subspace projection as described in claim 4, characterized in that step 4: determining the maximum range segment number P of the clutter echo distribution according to radar parameters;

[0028] Based on the distance range [R] of the clutter distribution in step 1 min ,R max The distance range R corresponding to a distance segment c The number of distance segments P to be reconstructed is calculated using equation (3):

[0029]

[0030] in This indicates rounding up to the nearest integer.

[0031] Further, step 5: Based on the geometric relationship between the radar antenna and the clutter scattering unit, determine the clutter subspace and construct the clutter projection matrix for the p-th range segment, specifically as follows:

[0032] It is obtained through the geometric relationship between the radar antenna and the clutter scattering unit.

[0033]

[0034] Where f d The Doppler frequency of the clutter scattering unit, such as Figure 5 As shown, θ is the angle between the clutter scattering element and the radar antenna in the azimuth dimension. Let be the pitch angle of the clutter scattering unit, α be the yaw angle, and ψ be the cone angle, and satisfy the following conditions: Define the angle-Doppler plane as cosψ ~ f d / f r , where f r Let Pulse Repetition Frequency (PRF) represent the frequency of repetition. Then, equation (4) can be further expressed as follows:

[0035]

[0036] Squaring and simplifying both sides of equation (5) yields the following:

[0037]

[0038] As can be seen from equation (6), the clutter spectrum is a set of elliptical equations in the angle-Doppler plane; depending on the different array placement, i.e., the different values ​​of the yaw angle α, the clutter distribution characteristics in the angle-Doppler plane are different.

[0039] Taking the antenna's frontal view, i.e., when the yaw angle α = 0°, as an example, the method for constructing the clutter subspace is introduced. At this time, equation (6) degenerates into a straight line:

[0040]

[0041] Where β = 2VT r / λ corresponds to the slope of the clutter ridge in the angle-Doppler plane;

[0042] Let the degrees of freedom of the constructed clutter subspace be l. Based on experience, l∈[2M,8M], the result Δf calculated by equation (8) is used. d The Doppler coverage range of clutter is defined by the interval [f]. dmin ,f dmax Discretization,

[0043]

[0044] The discrete Doppler frequency F of the constructed clutter subspace is obtained. d =[f d1 ,…,f dn ,…,f dl ], where f d1 =f dmin f dl=f dmax f dn Let represent the nth discretized Doppler frequency. According to the relationship between angle and Doppler in equation (7), the angle cos(ψ) corresponding to each discrete Doppler frequency can be calculated. n As shown in equation (9):

[0045]

[0046] Using equation (9), we can obtain the result related to F. d The corresponding angle F s This allows us to obtain the Doppler steering vector S of the clutter subspace at the p-th distance segment. dt With angular dimension guide vector S at S dt With S at The expression for the nth steering vector in

[0047]

[0048] Where d is the element spacing, S dt If p corresponds to the p-th distance segment, then the estimated clutter subspace Spa is expressed by equation (11):

[0049]

[0050] in Indicates that it is integrable;

[0051] According to the Brennan criterion, the degree of freedom r of clutter under the front-side look condition approximately satisfies equation (12).

[0052] r≈N+β(M-1) (12)

[0053] Since the degrees of freedom l of the constructed clutter subspace are greater than r, some information unrelated to clutter needs to be removed from the constructed subspace Spa. Singular value decomposition is performed on the clutter subspace Spa, as shown in equation (13).

[0054]

[0055] Where Λ1 is a diagonal matrix, and the elements on the diagonal correspond to r. c There are several large singular values ​​greater than 0 dB. U1 and V1 are the left and right singular vectors corresponding to each large singular value in Λ1, respectively. The elements on the diagonal of the diagonal matrix Λ2 correspond to (lr c There are ) small singular values ​​less than 0dB in Λ2, and U2 and V2 are the left singular vector and right singular vector corresponding to each small singular value in Λ2, respectively.

[0056] Remove (lr) through tail-truncated singular value decompositionc After identifying the small singular values ​​and their corresponding subspaces, the space spanned by U1 is the clutter subspace after removing information irrelevant to clutter.

[0057] The projection matrix Q corresponding to the clutter in the p-th range segment is constructed using equation (14). p :

[0058]

[0059] Further, step 6: Set the receiving filter bank corresponding to the p-th distance segment, specifically as follows:

[0060] If the parameters and modulation form of each pulse in the inter-pulse agile waveform are the same, but the initial phase between pulses is different, then the receiving filter bank is designed by a matched filter bank with time delay initial phase agile.

[0061] If the modulation patterns within each pulse are different, the receiving filter bank uses a joint mismatch filter to improve the range sidelobe modulation (RSM) effect caused by the change in modulation patterns between pulses.

[0062] Further, step 7: reconstruct the clutter of each range segment by subspace projection, and obtain the clutter reconstruction result of the p-th range segment; specifically:

[0063] Before reconstructing the clutter in the p-th range segment, the reconstruction results of clutter in other range segments besides the p-th range segment are subtracted from the total echo Vec(O) and used as the input Input(p) of the receiving filter bank in the p-th range segment, as shown in Equation (15).

[0064]

[0065] Initial clutter reconstruction results R for each range segment b (p) = 0, p = 1, 2, ..., P;

[0066] The reconstruction process includes the following steps: setting the receiving filter bank for the p-th range segment, performing pulse compression gating on Input(p) along the fast time dimension to select the echo of the p-th range segment; the echo data of the p-th range segment obtained after gating is Echo(p), and using the constructed projection matrix to perform subspace projection on Echo(p) to obtain the clutter projection result PRO(p) of the p-th range segment, as shown in Equation (16);

[0067] PRO(p) = Q p Echo(p) (16)

[0068] Q p This is the projection matrix corresponding to the clutter in the p-th range segment;

[0069] Performing pulse compression inverse processing on PRO(p) along the fast time dimension yields the clutter reconstruction result R for the p-th range segment. b (p).

[0070] Furthermore, in step 10, the subspace of the distance segment containing the target is updated, and the projection matrix of the distance segment containing the target is updated, specifically as follows:

[0071] The subspace of the target is constructed as U2;

[0072] Update the subspace of the distance segment where the target is located according to equation (17).

[0073] U1=[U1,U2] (17)

[0074] The projection matrix of this distance segment is updated using Equation (14).

[0075] Beneficial effects:

[0076] This invention proposes a clutter suppression method for airborne radar based on range gating and subspace projection. This method utilizes pulse-agile waveforms with range gating characteristics. After gating the echo of the current range segment from the range dimension, subspace projection is performed to obtain the clutter reconstruction result for that range segment. It then iterates between range segments to eliminate mutual interference between clutter in different range segments, achieving segment-specific clutter reconstruction. Subsequently, clutter outside the current range segment is eliminated from the received echo to achieve clutter suppression, followed by target detection. This effectively improves the airborne radar's detection capability for long-range, low-speed, and weak targets. This method has many advantages, including simple implementation, low computational complexity, and good clutter suppression effect. Attached Figure Description

[0077] Figure 1 (a) in the figure is a three-dimensional view of the ambiguity function of the linear frequency modulated pulse train waveform;

[0078] Figure 1 (b) in the figure is the distance-amplitude projection of the ambiguity function of the linear frequency modulated pulse train waveform;

[0079] Figure 2 (a) in the figure is a three-dimensional view of the ambiguity function of the inter-pulse agile pulse train waveform;

[0080] Figure 2 (b) in the figure is the distance amplitude projection diagram of the ambiguity function of the inter-pulse agile pulse train waveform;

[0081] Figure 3 This is a schematic diagram of the echoes from two range segments based on the inter-pulse agile waveform and the receiving filter bank for each range segment;

[0082] Figure 4This is a flowchart of an airborne radar clutter suppression method based on range gating and subspace projection according to the present invention.

[0083] Figure 5 This is a schematic diagram showing the geometric relationship between the radar antenna array and the clutter scattering unit;

[0084] Figure 6 To simulate the PD processing results of airborne radar clutter and target echo data, Figure 6 (a) is based on the linear frequency modulation transmit waveform. Figure 6 (b) is based on the pulse agile transmit waveform and processed by the method of the present invention. Detailed Implementation

[0085] The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0086] Most current airborne radars employ narrowband PD radar systems, and their transmitted waveforms typically use conventional waveforms with fixed parameters, such as Linear Frequency Modulation (LFM) waveforms. For ease of explanation, let's define a range R corresponding to a PRT. c For a distance segment, i.e. R c = 0.5(C·PRT), where C is the speed of light. From Figure 1 (a) Figure 1 As shown in (b), the ambiguity function of the LFM pulse train waveform reveals that the echoes between each distance segment of the LFM waveform with fixed parameters are completely blurred, which is the fundamental reason for clutter folding. Furthermore, from... Figure 2 (a) Figure 2 As shown in (b), the fuzzy function of the inter-pulse agile pulse train waveform (specifically the inter-pulse spectral agile waveform) indicates that the inter-pulse agile waveform has a range gating performance of approximately 30 dB. The range gating performance refers to the ability to select echoes in the current range segment and suppress echoes in other range segments.

[0087] Based on this, a set of corresponding receiving filters can be designed for the transmitted agile pulse train waveform, and the echoes of each range segment can be selected separately, such as... Figure 3As shown, taking two range segments as an example, a receiving filter bank H corresponding to K waveforms is designed, with PRT as the interval, and the echoes of each range segment are sequentially selected by time shift H. However, due to the limited range gating performance of agile waveforms, H can only achieve decorrelation for echoes outside the current range segment, rather than orthogonality. From the perspective of clutter echoes, under the phased array radar system, the clutter of the current range segment coherently accumulated after range gating exhibits obvious geometric characteristics in the angle-Doppler plane. Taking the frontal side view at a yaw angle of 0° as an example, the clutter is linearly distributed in the angle-Doppler plane; while the clutter outside the current range segment will be scattered across the entire angle-Doppler plane after decorrelation. Therefore, this invention first utilizes the range gating performance of agile waveforms to achieve clutter decorrelation in each range segment. Then, it constructs a clutter subspace based on the characteristics of clutter in angle-Doppler behavior. By projecting the echoes of each range segment into the subspace and iterating between range segments, it eliminates the mutual influence between clutter in each range segment, thereby achieving clutter reconstruction in different range segments. Afterward, it cancels clutter outside the current range segment from the received echoes to achieve folded clutter suppression and performs target detection, effectively improving the detection capability of airborne radar for long-range, low-speed, and weak targets.

[0088] like Figure 4 As shown, the airborne radar clutter suppression method based on range gating and subspace projection of the present invention specifically includes the following steps:

[0089] Step 1: Estimate the range interval of clutter distribution based on the scene. Since the airborne radar has limited power, assume the maximum detection range of the airborne radar is [R]. min ,R max The specific relationship is shown in equation (1).

[0090]

[0091] Where H is the flight altitude, P t For radar transmission power, G t For the transmit antenna gain, A r S is the effective receiving area of ​​the antenna, σ is the scattering cross-section of the radar illumination area, and S min This is the minimum detectable signal for the receiver.

[0092] Step 2: Determine the clutter Doppler coverage area. Let the airborne radar's flight speed relative to the ground be V, and the wavelength be λ. Then the clutter Doppler coverage area [f]... dmin ,f dmax It can be calculated by equation (2).

[0093]

[0094] Step 3: Vectorize the total echo along the fast time dimension. Let the total echo be O, where O is a three-dimensional matrix with dimensions M×N×PRTnum, where M corresponds to the number of pulses, N corresponds to the number of array elements, and PRTnum corresponds to the number of sampling points within a PRT. Since the projection matrices for each range segment constructed in Step 5 are two-dimensional matrices with dimensions MN×MN, to match the dimensions of the echo data with the projection matrix, O needs to be vectorized along the fast time dimension, transforming the three-dimensional matrix into a two-dimensional matrix, resulting in Vec(O), with dimensions MN×PRTnum.

[0095] Step 4: Determine the number of distance segments to be reconstructed. Based on the clutter distribution range [R] in Step 1. min ,R max The distance range R corresponding to a distance segment c The number of distance segments P to be reconstructed is calculated using equation (3).

[0096]

[0097] in This indicates rounding up. Each distance segment to be reconstructed is numbered as a variable p (p = 1, 2, ..., P), and initialized to p = 1.

[0098] Step 5: Based on the geometric relationship between the radar antenna and the clutter scattering unit, determine the clutter subspace and construct the clutter projection matrix for the p-th range segment. For example... Figure 5 As shown, the geometric relationship between the radar antenna and the clutter scattering unit can be used to obtain...

[0099]

[0100] Where f d Let θ represent the Doppler frequency of the clutter scattering element, and θ be the angle between the clutter scattering element and the radar antenna in the azimuth dimension. Let be the pitch angle of the clutter scattering unit, α be the yaw angle, and ψ be the cone angle, and satisfy the following conditions: Define the angle-Doppler plane as cosψ ~ f d / f r , where f r Let Pulse Repetition Frequency (PRF) represent the pulse repetition frequency. Then, equation (4) can be further expressed as follows:

[0101]

[0102] Squaring and simplifying both sides of equation (5) yields the following:

[0103]

[0104] As can be seen from equation (6), the clutter spectrum is a set of elliptical equations in the angle-Doppler plane. Depending on the array placement, i.e., different values ​​of the yaw angle α, the distribution characteristics of clutter in the angle-Doppler plane vary.

[0105] Taking the antenna's frontal view, i.e., when the yaw angle α = 0°, as an example, the method for constructing the clutter subspace is introduced. At this time, equation (6) degenerates into a straight line.

[0106]

[0107] Where β = 2VT r / λ corresponds to the slope of the clutter ridge in the angle-Doppler plane.

[0108] Let the degrees of freedom of the constructed clutter subspace be l, where, empirically, l ∈ [2M, 8M]. The calculation result Δf from equation (8) is used. d The Doppler coverage range of clutter is defined by the interval [f]. dmin ,f dmax Discretization,

[0109]

[0110] The discrete Doppler frequency F of the constructed clutter subspace is obtained. d =[f d1 ,...,f dn ,…,f dl ], where f d1 =f dmin f dl =f dmax f dn Let represent the nth discretized Doppler frequency. According to the relationship between angle and Doppler in equation (7), the angle cos(ψ) corresponding to each discrete Doppler frequency can be calculated. n As shown in equation (9),

[0111]

[0112] Using equation (9), we can obtain the result related to F. d The corresponding angle F s This allows us to obtain the Doppler steering vector S of the clutter subspace at the p-th distance segment. dt With angular dimension guide vector S at S dt With S at The expression for the nth steering vector in

[0113]

[0114] Where d is the element spacing, S dtIf p corresponds to the p-th range segment, then the estimated clutter subspace Spa can be represented by equation (11).

[0115]

[0116] in This indicates that the clutter is integrable. According to the Brennan criterion, the degree of freedom r of the clutter under the front-side look condition approximately satisfies equation (12).

[0117] r≈N+β(M-1) (12)

[0118] The degrees of freedom l of the constructed clutter subspace are generally greater than r. Therefore, some information unrelated to clutter needs to be removed from the constructed subspace Spa. Singular value decomposition is performed on the clutter subspace Spa, as shown in equation (13).

[0119]

[0120] Where Λ1 is a diagonal matrix, and the elements on the diagonal correspond to r. c There are several large singular values ​​greater than 0 dB. U1 and V1 are the left and right singular vectors corresponding to each large singular value in Λ1, respectively. The elements on the diagonal of the diagonal matrix Λ2 correspond to (lr c There are ) small singular values ​​less than 0 dB in Λ2, and U2 and V2 are the left and right singular vectors corresponding to each small singular value in Λ2, respectively. The (lr) is removed by truncated singular value decomposition. c After identifying the p-th singular values ​​and their corresponding subspaces, the space spanned by U1 is the clutter subspace after removing information irrelevant to the clutter. Therefore, the projection matrix Q corresponding to the clutter in the p-th range segment can be constructed using equation (14). p ,

[0121]

[0122] Step 6: Set up the receiving filter bank corresponding to the p-th range segment. If the parameters and modulation forms within each pulse of the inter-pulse agile waveform are the same, and only the initial phase between pulses is different, then the receiving filter bank can be designed using a matched filter bank with time delay initial phase agile. If the modulation forms within each pulse are different, then the receiving filter bank must use a joint mismatch filter to improve the range sidelobe modulation (RSM) effect caused by the change in the modulation form between pulses. Several commonly used joint mismatch filter design algorithms are currently available.

[0123] Step 7: Reconstruct clutter for each range segment by subspace projection. Before reconstructing the clutter for the p-th range segment, the reconstruction results of the clutter for other range segments besides the p-th range segment need to be subtracted from the total echo Vec(O) and used as the input Input(p) of the receiving filter bank for the p-th range segment, as shown in Equation (15).

[0124]

[0125] Initial clutter reconstruction results R for each range segment b (p) = 0, p = 1, 2, ..., P.

[0126] The reconstruction submodule includes the following steps: setting the receiving filter bank for the p-th range segment, performing pulse compression gating on Input(p) along the fast time dimension to select the echo of the p-th range segment; the echo data of the p-th range segment obtained after gating is Echo(p), and using the constructed projection matrix to perform subspace projection on Echo(p) to obtain the clutter projection result PRO(p) of the p-th range segment, as shown in Equation (16);

[0127] PRO(p) = Q p Echo(p) (16)

[0128] Performing pulse compression inverse processing on PRO(p) along the fast time dimension yields the clutter reconstruction result R for the p-th range segment. b (p).

[0129] Step 8: Iterate between range segments to eliminate the mutual influence between clutter in each range segment. Let p = p + 1 (if p > P after the update, then let p = 1 again), and repeat steps 5 to 7 until the iteration process converges. The convergence condition can be determined by the difference between two adjacent iterations. If the absolute value of the difference is less than a certain threshold, the algorithm can be considered to have converged.

[0130] Step 9: Subtract the reconstructed clutter from the non-local clutter for each range segment using the total echo Vec(O), as shown in Equation (15), to obtain the clutter reconstruction result for each range segment. Then, perform inverse vectorization along the fast time dimension to complete the accurate reconstruction of clutter for each range segment and the effective suppression of folded clutter.

[0131] Step 10: Detect targets and add them to the reconstruction process. After subtracting the reconstructed range clutter from the total echo O, targets in the scene are detected by range segment. For range segments where targets exist, a target subspace U2 is constructed, and the subspace of the range segment containing the target is updated according to equation (17).

[0132] U1=[U1,U2] (17)

[0133] Then, use Equation (14) to update the projection matrix of this range segment, and repeat steps 7-10 until the total echo minus the reconstruction result of each range segment is less than the set threshold, at which point the target detection can be considered complete.

[0134] The beneficial effects of the invention will be verified and explained:

[0135] To test the algorithm's performance, a comprehensive simulation test was conducted on an airborne radar detection scenario. In the simulation, the radar's unambiguous ranging range was set to 75km. The distances and velocities of four distant low-speed point targets were 494km and 56m / s; 502km and -67m / s; 567km and 32m / s; and 572km and -24m / s, respectively, which are located in the 7th and 8th range segments. The positive and negative signs of the velocities represent the target moving away from or towards the radar, respectively. The target signal-to-noise ratio was approximately 20dB. The simulated airborne radar clutter velocity distribution ranged from approximately -90m / s to 90m / s, the distance distribution ranged from approximately 9km to 450km, and the signal-to-clutter ratio was approximately -50dB. Figure 6 (a) in the figure shows the simulation results based on the linear frequency modulation transmission waveform. As can be seen from the figure, both clutter and target are folded into the same range segment, resulting in severe range ambiguity. Furthermore, due to the folded clutter covering low-speed point targets, they cannot be effectively detected. Figure 6 Figure (b) shows the simulation results based on the pulse agile transmission waveform and processed by the method of this invention. As can be seen from the figure, the folded clutter in each range segment is completely reconstructed, effectively removing the influence of the folded clutter on the target detection in the 7th and 8th range segments, and realizing unambiguous ranging and velocity measurement of the target.

[0136] In summary, the above are merely preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. An airborne radar clutter suppression method based on range gating and subspace projection, characterized in that, Includes the following steps: Step 1: Estimate the distance range of clutter distribution based on the scene; Step 2: Determine the clutter Doppler coverage area based on radar parameters; Step 3: Vectorize the total echo along the fast time dimension to obtain the vectorized total echo; Step 4: Determine the maximum range segment number of the clutter echo distribution based on radar parameters. Each distance segment to be reconstructed is numbered as a variable. , Values ,initialization ; Step 5: Based on the geometric relationship between the radar antenna and the clutter scattering unit, determine the clutter subspace and construct the first... Range clutter projection matrix; Step 6: Set the first The receiving filter bank corresponding to the range segment; Step 7: Reconstruct the clutter of each range segment by subspatial projection, and obtain the clutter of each range segment. Clutter reconstruction results for the range segment; Step 8: Let Increment by 1 if After the update Then reorder Repeat steps 5 to 7 to perform inter-segment cyclic iteration until the iteration process converges; Step 9: Subtract the reconstructed clutter from the non-local clutter in each range segment using the vectorized total echo to obtain the clutter reconstruction result for each range segment. Then, perform inverse vectorization along the fast time dimension to complete the accurate reconstruction of clutter in each range segment and the effective suppression of folded clutter. Step 10: After subtracting the reconstructed clutter from each range segment using the total echo, detect targets in the scene by range segment. For range segments where targets exist, construct the target subspace, update the subspace of the range segment where the target is located, and update the projection matrix of the range segment where the target is located. Repeat steps 7-10 until the total echo minus the reconstruction result of each range segment is less than the set threshold, and the target detection is complete.

2. The airborne radar clutter suppression method based on range gating and subspace projection as described in claim 1, characterized in that, Step 1: Estimating the distance range of clutter distribution based on the scene, specifically: The maximum detection range of airborne radar is The specific relationship is as shown in equation (1): （1） in For flight altitude, For radar transmission power, For the transmit antenna gain, The effective receiving area of ​​the antenna. The radar cross section is the area of ​​the radar illumination region. This is the minimum detectable signal for the receiver; This refers to the distance range of clutter distribution.

3. The airborne radar clutter suppression method based on range gating and subspace projection as described in claim 1 or 2, characterized in that, Step 2: Determine the clutter Doppler coverage area based on radar parameters, specifically as follows: The airborne radar's speed relative to the ground is , wavelength is Then the Doppler coverage area of ​​the clutter wave Calculated using equation (2): （2）。 4. The airborne radar folding clutter suppression method based on range gating and subspace projection as described in claim 3, characterized in that, Step 3: Vectorize the total echo along the fast time dimension, specifically as follows: Total echo is , It is a three-dimensional matrix with dimension 1. ,in Corresponding to the number of pulses, Corresponding to the number of array elements, The number of sampling points corresponding to a PRT; along the fast time dimension Vectorization is performed, transforming the three-dimensional matrix into a two-dimensional matrix to obtain the vectorized total echo. ; The dimension is .

5. The airborne radar folding clutter suppression method based on range gating and subspace projection as described in claim 4, characterized in that, Step 4: Determine the maximum range segment P of the clutter echo distribution based on radar parameters; Based on the distance range of clutter distribution in step 1 Distance range corresponding to a distance segment The number of distance segments to be reconstructed is calculated using equation (3). : （3）， in This indicates rounding up to the nearest integer.

6. The airborne radar clutter suppression method based on range gating and subspace projection as described in claim 5, characterized in that, Step 5: Based on the geometric relationship between the radar antenna and the clutter scattering unit, determine the clutter subspace and construct the first... The range clutter projection matrix is ​​as follows: It is obtained through the geometric relationship between the radar antenna and the clutter scattering unit. （4） in The Doppler frequency of the clutter scattering unit is represented. The angle between the clutter scattering element and the radar antenna in the azimuth dimension. The elevation angle of the clutter scattering unit. Yaw angle It is a cone angle, and satisfies Define the angle-Doppler plane as... ,in Let PRF represent the pulse repetition frequency. Then, equation (4) can be further expressed as follows: （5） Squaring and simplifying both sides of equation (5) yields the following: （6） As can be seen from equation (6), the clutter spectrum is a set of elliptical equations in the angle-Doppler plane; depending on the array placement, i.e., the yaw angle The distribution characteristics of clutter in the angle-Doppler plane differ depending on the value taken; When yaw angle At this point, equation (6) degenerates into a straight line: （7） in The slope of the corresponding clutter ridge in the angle-Doppler plane; Let the degrees of freedom of the constructed clutter subspace be . Based on experience The calculation result of equation (8) The interval is used to determine the Doppler coverage of clutter. Discretization, （8） The discrete Doppler frequency of the constructed clutter subspace is obtained. ,in , , Indicates the first Based on the relationship between angle and Doppler frequency in equation (7), the angle corresponding to each discrete Doppler frequency can be calculated. As shown in equation (9): （9） Using equation (9), we can obtain the result of equation (9). Corresponding angle Therefore, we can obtain the first... Range segment clutter subspace Doppler steering vector With angular dimension guide vector , and The first in Expression of a guide vector （10） in For the spacing between array elements, middle Corresponding to the For each distance segment, the estimated clutter subspace is... This can be expressed using equation (11): （11） in Indicates that it is integrable; According to the Brennan criterion, the degrees of freedom of clutter under front-side look conditions It approximately satisfies equation (12). （12） The degrees of freedom of the constructed clutter subspace Greater than Therefore, the constructed subspace There is some information unrelated to clutter that needs to be removed; for the clutter subspace Perform singular value decomposition, as shown in equation (13). （13） in It is a diagonal matrix, and the elements on the diagonal correspond to A large singular value greater than 0 dB , They are respectively The left and right singular vectors corresponding to each large singular value in the matrix, and the diagonal matrix The elements on the diagonal correspond A small singular value less than 0 dB , They are respectively The left and right singular vectors corresponding to each small singular value in the vector; Removed by tail-truncated singular value decomposition After considering the small singular values ​​and their corresponding subspaces, Zhang Cheng's space is the clutter subspace after removing information irrelevant to clutter. Construct the first through equation (14) Projection matrix corresponding to clutter at each range segment : （14）。 7. The airborne radar clutter suppression method based on range gating and subspace projection as described in any one of claims 1, 2, 4, 5, or 6, characterized in that, Step 6: Set the first The receiving filter bank corresponding to the range segment is as follows: If the parameters and modulation form of each pulse in the inter-pulse agile waveform are the same, but the initial phase between pulses is different, then the receiving filter bank is designed by a matched filter bank with time delay initial phase agile. If the modulation patterns within each pulse are different, the receiving filter bank uses a joint mismatch filter to improve the range sidelobe modulation (RSM) effect caused by the inter-pulse modulation pattern variation.

8. The airborne radar clutter suppression method based on range gating and subspace projection as described in claim 6, characterized in that, Step 7: Reconstruct clutter for each range segment by subspace projection, obtaining the clutter of the first range segment. Clutter reconstruction results for each range segment; specifically: When for the first Before reconstructing the range clutter, except for the first segment... Clutter reconstruction results for range segments other than the total echo Subtract from the middle, as the first Input of range segment receiving filter bank As shown in equation (15), （15） Initial clutter reconstruction results for each range segment ; The refactoring process includes the following steps: setting the first... The receiving filter bank for each range segment, along the fast time dimension... Perform pulse compression gating The echo of the first distance segment; the echo obtained after gating. The echo data for each distance segment is Using the constructed projection matrix to Perform subspace projection to obtain the first... Clutter projection results in the range segment As shown in equation (16); （16） in For the first The projection matrix corresponding to clutter in each range segment; right Performing pulse compression inverse processing along the fast time dimension yields the th... Clutter reconstruction results for each range segment .

9. The airborne radar clutter suppression method based on range gating and subspace projection as described in claim 8, characterized in that, In step 10, the subspace of the distance segment containing the target is updated, and the projection matrix of the distance segment containing the target is updated, specifically as follows: The subspace of the constructed target is ; Update the subspace of the distance segment where the target is located according to equation (17). （17） The projection matrix of this distance segment is updated using equation (14).

Images