A Design Method for MTD Filters Based on Particle Swarm Optimization Algorithm

Abstract

This invention relates to the field of radar signal processing technology, and particularly to an MTD filter design method based on particle swarm optimization, comprising the following steps: S1: Obtaining the Doppler center frequency range and normalized half-width based on the radar type and clutter type to obtain constraints, and obtaining filter information based on the constraints and the global pilot vector matrix; S2: Designing a fitness function based on a nonlinear function and a hard-coded threshold; S3: Designing the filter based on the fitness function, and iteratively updating the filter using a main lobe width bisection method to obtain the optimal filter. This invention achieves optimized design by constructing linear and nonlinear multi-parameter fitness functions, while satisfying the requirements for stopband attenuation, null depth, and clutter bandwidth. Furthermore, it uses a bisection method for iterative optimization of the main lobe width, achieving high robustness under different clutter environments, improving the design efficiency of the MTD filter, reducing main lobe broadening, and lowering filter gain loss.

Description

Technical Field

[0001] This invention relates to the field of radar signal processing technology, and in particular to an MTD filter design method, device, and storage medium based on particle swarm optimization algorithm. Background Technology

[0002] When detecting, distinguishing, and identifying various aerial targets, the overall detection performance of pulse-Doppler radar is often limited by strong environmental clutter around the target. The MTD filter is a key component in pulse-Doppler radar signal processing, primarily used for detecting and separating moving target signals.

[0003] Traditional MTD filter design methods primarily rely on classic FIR filter design techniques. These methods typically require manual parameter tuning and their performance is less than ideal when handling complex environments (such as multipath propagation and strong clutter interference). Typical clutter interference includes ground clutter, weather clutter, and sea clutter. These clutter signals have a certain radial velocity relative to the radar, thus slightly broadening in the Doppler dimension. Typically, clutter signals are very powerful; without any processing, target echoes will be completely overwhelmed, severely reducing the radar's detection probability. Furthermore, if clutter exhibits strong points in the range dimension, it may be misidentified as a real target, significantly increasing the radar's false alarm rate.

[0004] Conventional adaptive MTD filter design methods can meet the expected frequency response requirements, but they require repeated manual adjustments to the stopband level and clutter power, and cannot automatically and accurately adjust the relevant control parameters. Furthermore, within the weak suppression range, only some sidelobe maxima of the designed filter meet or approach the expected stopband attenuation requirements, while the maxima of other sidelobes are less than the expected attenuation requirements. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies and provide an MTD filter design method based on particle swarm optimization, comprising the following steps:

[0006] S1: Obtain the Doppler center frequency range and normalized half-width based on the radar type and clutter type; obtain the constraint conditions including the center frequency point, sidelobe region, null region and main lobe region based on the Doppler center frequency range and the normalized half-width; and obtain the filter information based on the constraint conditions and the global pilot vector matrix.

[0007] S2: Design a fitness function based on the nonlinear function used to adjust the filter information and the hard-coded threshold;

[0008] S3: Design a filter based on the fitness function, and iteratively update the filter according to the main lobe width bisection method to obtain the optimal filter.

[0009] Preferably, in step S1, obtaining the filter information based on the constraints and the global pilot vector matrix further includes:

[0010] S11: Construct the global pilot vector matrix based on the filter order and all sampling points of the normalized Doppler frequency;

[0011] S12: Calculate the filter frequency response amplitude based on the constraints and the global pilot vector matrix, obtain the frequency response gain based on the filter frequency response amplitude, and obtain the filter information based on the frequency response gain.

[0012] Preferably, in step S11, constructing the global pilot vector matrix based on the filter order and the full sampling points of the normalized Doppler frequency further includes:

[0013] The formula for calculating the row vectors of the global pilot vector matrix is ​​as follows:

[0014]

[0015] in, For the normalized Doppler frequency full sampling points, This represents a row vector whose elements are indices of the filter order, i.e. , The order of the filter is given by the global pilot vector matrix. The The row representation filter in the first row Complex response vector at each frequency point;

[0016] The global pilot vector matrix is ​​obtained from the row vectors of the global pilot vector matrix. The calculation formula is as follows:

[0017] .

[0018] Preferably, in step S12, calculating the filter frequency response amplitude based on the constraints and the global pilot vector matrix further includes:

[0019] According to the normalized Doppler center frequency Calculate the initial pilot vector, the initial pilot vector The calculation formula is as follows:

[0020]

[0021] in, It is the symbol for imaginary numbers;

[0022] The complex weight vector is obtained based on the initial pilot vector. The calculation formula is as follows:

[0023]

[0024] Wherein, the complex weight vector Optimization variables With the initial pilot vector The element-wise product of the complex weight vector The initial pilot vector information, which includes the filter amplitude, filter phase, and filter center frequency, directly affects the frequency response characteristics of the filter.

[0025] The filter frequency response amplitude is obtained based on the constraints, the global pilot vector matrix, and the complex weight vector. The calculation formula is as follows:

[0026] .

[0027] Preferably, in step S12, obtaining the frequency response gain based on the frequency response amplitude of the filter further includes:

[0028] The normal frequency response gain is obtained by normalizing the frequency response amplitude of the filter. The calculation formula is as follows:

[0029]

[0030] According to the normal frequency response gain The frequency response gain is obtained, and the frequency response gain The calculation formula is as follows:

[0031] .

[0032] Preferably, in step S12, obtaining the filter information based on the frequency response gain further includes:

[0033] The maximum level of the sidelobe, the maximum level of the null, and the maximum level of the main lobe are obtained based on the frequency response gain.

[0034] The maximum level of the sidelobe in the sidelobe region is calculated based on the frequency response gain. The calculation formula is as follows:

[0035]

[0036] in, This is a zero-depression region;

[0037] The maximum null level in the null region is calculated based on the frequency response gain. The calculation formula is as follows:

[0038]

[0039] in, This is a zero-depression region;

[0040] The maximum level of the main lobe in the main lobe region is calculated based on the frequency response gain. The calculation formula is as follows:

[0041] .

[0042] Preferably, in step S3, designing the filter based on the fitness function further includes:

[0043] The fitness function designs the filter by adjusting the sidelobe level and the null level. The formula for calculating the fitness function is as follows:

[0044]

[0045] in, The main lobe level is V, and V is the hard-coded threshold. The sidelobe level, The noise suppression zone level is [level value]. For weak attenuation of the sidelobes, This is for clutter band attenuation.

[0046] Preferably, in step S3, obtaining the optimal filter through iteration using the main lobe width bisection method further includes:

[0047] The lower limit is the width of the main lobe corresponding to the unwindowed FFT. The upper limit is the main lobe width corresponding to the Taylor window, with the clutter suppression region level being added. The tolerance ζ is given, and the main lobe width in the first optimization is... =( + ) / 2;

[0048] If the main lobe width meets the requirements in the initial optimization, then If the main lobe width does not meet the requirements in the initial optimization, then... Iterate according to the main lobe width bisection method until... - <ζ;

[0049] The optimal filter is obtained based on the obtained optimal main lobe width.

[0050] Based on the same concept, the present invention also provides a computer device, including a memory and a processor, wherein the memory stores computer-readable instructions, which, when executed by the processor, cause the processor to perform the steps of the MTD filter design method based on particle swarm optimization as described in any one of the embodiments.

[0051] Based on the same concept, the present invention also provides a storage medium storing computer-readable instructions, which, when executed by one or more processors, cause the one or more processors to perform the steps of the MTD filter design method based on particle swarm optimization as described in any one of the embodiments.

[0052] Compared with the prior art, the beneficial effects of the present invention are:

[0053] This invention obtains the Doppler center frequency range and normalized half-width by using radar type and clutter type. Based on the Doppler center frequency range and normalized half-width, it obtains constraints including the center frequency, sidelobe region, null region and main lobe region. Based on the constraints and global pilot vector matrix, it obtains filter information, thus enabling simultaneous optimization of the amplitude and phase of the target filter.

[0054] This invention achieves optimized design by constructing linear and nonlinear multi-parameter fitness functions, while meeting the requirements of stopband attenuation, null depth, and clutter bandwidth.

[0055] This invention optimizes the main lobe width using a bisection method and simulation results show that the algorithm has high robustness in different clutter environments. It also has lower gain loss than the original Taylor window FFT-type MTD filter with direct clutter attenuation level, stronger clutter suppression effect, narrower main lobe width, and improved MTD filter design efficiency. Attached Figure Description

[0056] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention.

[0057] Figure 1 This is a flowchart of an MTD filter design method based on particle swarm optimization algorithm according to the present invention.

[0058] Figure 2This is the amplitude-frequency response diagram of the Ku-band radar ground clutter MTD filter of the present invention;

[0059] Figure 3 This is the amplitude-frequency response diagram of the X-band radar meteorological clutter MTD filter of the present invention;

[0060] Figure 4 This is the amplitude-frequency response diagram of the C-band radar sea clutter MTD filter of the present invention;

[0061] Figure 5 This is a comparison diagram of the results before and after the completion of the bisection method iteration in this invention;

[0062] Figure 6 This is a comparison of the main lobe of the present invention with the main lobe of a conventional Taylor windowed MTD filter. Detailed Implementation

[0063] Example 1

[0064] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. Obviously, the described embodiments are only some, not all, of the embodiments described in this application. All other embodiments obtained by those skilled in the art based on the embodiments in this application without creative effort are within the scope of protection of this application.

[0065] Those skilled in the art will understand that, unless otherwise stated, the singular forms “a” and “an” used herein, and “the”, may also include the plural forms. It should be further understood that the term “comprising” as used in this specification means the presence of the stated features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0066] In recent years, intelligent optimization algorithms (such as Particle Swarm Optimization (PSO) and Genetic Algorithm (GA)) have demonstrated superior performance in signal processing, especially in optimization problems under nonlinear and multi-constraint conditions. These algorithms, by simulating the behavior of biological populations or the process of natural selection, can quickly search for the global optimum of a problem, thereby significantly improving the design efficiency and performance of MTD filters.

[0067] Classical FIR filters often require manual parameter adjustment and perform poorly in complex clutter environments. This embodiment, however, first defines four constraints related to the center frequency, main lobe, strong clutter suppression band, and weak sidelobe suppression before entering the particle swarm optimization algorithm. Then, it sets different parameters such as radar operating frequency, clutter velocity standard deviation, MTD filter order, and pulse repetition period to perform detailed clutter modeling and establish a normalized Doppler full-frequency pilot vector matrix. Next, it constrains the positions of the clutter band and main lobe within the normalized Doppler full frequency range. Finally, based on the set center frequency, it constructs pilot vectors at different MTD filter center frequencies.

[0068] After entering the particle swarm optimization (PSO) design phase, the candidate solutions of the PSO algorithm simultaneously contain the amplitude and phase information of the target filter. The candidate solutions are then multiplied term-by-term with the pilot vector at the center frequency of the MTD filter to obtain a complex weight vector. This vector is then multiplied with the normalized Doppler full-band pilot vector matrix to obtain the frequency response of the target filter. Simultaneously, parameters such as the number of particles, the hybrid optimizer, the maximum number of iterations, and the initial population matrix are adjusted. Furthermore, multi-objective optimization is performed by adding linear constraints on sidelobe and clutter attenuation, and nonlinear constraints on the main lobe center frequency and width, to the fitness function, thus completing the initial design of the filter with a determined center frequency. Finally, the optimal main lobe width is found through bisection iteration, yielding the optimal design result of this invention.

[0069] Please see Figure 1 As shown in the figure, this embodiment provides an MTD filter design method based on particle swarm optimization, which achieves a series of excellent characteristics such as strong attenuation of clutter bands, flexible adjustment of the operating environment, and convenient and fast design process. The method includes the following steps:

[0070] S1: Obtain the Doppler center frequency range and normalized half-width based on the radar type and clutter type. Obtain the constraint conditions including the center frequency, sidelobe region, null region and main lobe region based on the Doppler center frequency range and normalized half-width. Obtain the filter information based on the constraint conditions and the global pilot vector matrix.

[0071] Preferably, the filter order is set to be... The normalized Doppler center frequency is Each MTD filter in the filter bank should have a constant gain of 0 dB at its center frequency, and an initial main lobe width of... Then the main lobe is within the normalized Doppler frequency range. , For the filter frequency response, the frequency response at the center frequency point (i.e., the center frequency region) should meet the following requirements:

[0072]

[0073] To prevent a shift in the center frequency of the filter's main lobe, the main lobe frequency response (i.e., the main lobe region) must satisfy:

[0074]

[0075] Specifically, in this embodiment, the filter's main lobe center frequency is kept constant.

[0076] The clutter bandwidth is calculated based on the radar operating frequency, pulse repetition period, ground clutter velocity standard deviation, and light speed. The calculation formula is as follows:

[0077]

[0078] in, Where is the radar operating frequency, and T is the pulse repetition period. Let c be the standard deviation of ground clutter velocity, and c be the speed of light.

[0079] Normalizing the clutter bandwidth yields the normalized clutter bandwidth. Determine the region with strong clutter suppression. weak inhibition region of the sidelobe and the main petal The minimum distance from the strong suppression region, therefore the clutter band is within the normalized Doppler frequency range. Let the clutter band attenuation be... To meet the most critical requirements of the MTD filter, this embodiment uses a strong constraint to create a null with strong attenuation near the zero frequency band to suppress clutter intensity. The clutter band frequency response satisfies (i.e., the null region):

[0080]

[0081] Specifically, in this embodiment, the null traps that form strong attenuation in the clutter band near zero frequency are strongly constrained. To suppress clutter intensity in the clutter band.

[0082] Besides the constrained main lobe and clutter band regions mentioned above, the remaining range in the fully normalized Doppler frequency range belongs to the weakly suppressed sidelobe portion far from zero frequency. Let the weak sidelobe attenuation be... Therefore, the side lobe region The sidelobe frequency response satisfies (i.e., the sidelobe region):

[0083]

[0084] Specifically, in this embodiment, the sidelobe portion far from zero frequency is weakly suppressed. .

[0085] Having established the preliminary frequency response constraints for each region, this embodiment intends to use the fitness function of the particle swarm optimization algorithm to constrain these conditions. In order to obtain the frequency response amplitude, this invention constructs a complex matrix A to represent the filter's response at different frequency points, i.e., the pilot vector matrix.

[0086] Preferably, in step S1, obtaining filter information based on constraints and the global pilot vector matrix further includes:

[0087] S11: Construct a global pilot vector matrix based on the filter order and all sampling points of the normalized Doppler frequency;

[0088] S12: Calculate the filter frequency response amplitude based on the constraints and the global pilot vector matrix, obtain the frequency response gain based on the filter frequency response amplitude, and obtain the filter information based on the frequency response gain.

[0089] Preferably, in step S11, constructing a global pilot vector matrix based on the filter order and all sampling points of the normalized Doppler frequency further includes:

[0090] The formula for calculating the row vectors of the global pilot vector matrix is ​​as follows:

[0091]

[0092] in, For the normalized Doppler frequency full sampling points, This represents a row vector whose elements are indices of the filter order, i.e. , Let be the order of the filter, and be the global pilot vector matrix. The The row representation filter in the first row Complex response vector at each frequency point;

[0093] The global pilot vector matrix is ​​obtained from the row vectors of the global pilot vector matrix. The calculation formula is as follows:

[0094] .

[0095] In the fitness function of the particle swarm optimization algorithm in this embodiment, X is a candidate solution for the particle swarm optimization, which is an array of length twice the filter order (2N). The upper limit of the first to Nth elements is 1, and the lower limit is -1. The upper limit of the (N+1)th to 2Nth elements is i, and the lower limit is -i. This allows for simultaneous optimization of the amplitude and phase of the target filter.

[0096] Preferably, in step S12, calculating the filter frequency response amplitude based on the constraints and the global pilot vector matrix further includes:

[0097] Based on the normalized Doppler center frequency Calculate the initial pilot vector. The calculation formula is as follows:

[0098]

[0099] in, It is the symbol for imaginary numbers;

[0100] The complex weight vector is obtained from the initial pilot vector. The calculation formula is as follows:

[0101]

[0102] Among them, complex weight vector Optimization variables With the initial pilot vector Element-wise product, complex weight vector The initial pilot vector information, which includes the filter amplitude, filter phase, and filter center frequency, directly affects the frequency response characteristics of the filter. Specifically, in this embodiment, complex weights are introduced and composite weighted phase optimization is used to dynamically evaluate the sidelobe level and clutter suppression region level of the filter.

[0103] The filter frequency response amplitude is obtained based on the constraints, the global pilot vector matrix, and the complex weight vector. The calculation formula is as follows:

[0104] .

[0105] Preferably, in step S12, obtaining the frequency response gain based on the filter frequency response amplitude value further includes:

[0106] The normal frequency response gain is obtained by normalizing the filter's frequency response amplitude. The calculation formula is as follows:

[0107]

[0108] Based on normal frequency response gain The frequency response gain is obtained. The calculation formula is as follows:

[0109] .

[0110] Preferably, in step S12, obtaining filter information based on frequency response gain further includes:

[0111] The maximum level of the sidelobe, the maximum level of the null, and the maximum level of the main lobe are obtained from the frequency response gain.

[0112] Among them, the maximum sidelobe level in the sidelobe region is calculated based on the frequency response gain. The calculation formula is as follows:

[0113]

[0114] in, This is a zero-depression region;

[0115] Calculate the maximum null level in the null region based on the frequency response gain. The calculation formula is as follows:

[0116]

[0117] in, This is a zero-depression region;

[0118] The maximum level of the main lobe in the main lobe region is calculated based on the frequency response gain. The calculation formula is as follows:

[0119] .

[0120] S2: Design a fitness function based on the nonlinear function used to adjust the filter information and a hard-coded threshold. Specifically, in this embodiment, a hard-coded threshold V is introduced to prevent numerical overflow of the nonlinear function during the optimization process.

[0121] S3: Design a filter based on the fitness function, and iteratively update the filter using the main lobe width bisection method to obtain the optimal filter.

[0122] Preferably, in step S3, designing the filter based on the fitness function further includes:

[0123] The fitness function designs the filter by adjusting the sidelobe level and the null level. The formula for calculating the fitness function is as follows:

[0124]

[0125] in, The main lobe level is V, and V is the hard-coded threshold. The sidelobe level, The noise suppression zone level is [level value]. For weak attenuation of the sidelobes, To attenuate clutter bands, specifically in this embodiment, after determining the main lobe level adjustment portion, the side lobe levels are adjusted using a linear function. ) and clutter suppression level ( This is used to obtain multi-objective combined optimization results.

[0126] The fitness function comprehensively considers the different requirements of the filter's main lobe gain and sidelobe and null suppression, thereby ensuring that the designed filter can meet specific performance indicators. The first term aims to control the filter's main lobe gain. By using an exponential function, this term can nonlinearly compress the main lobe gain. When the main lobe gain is high, the value of this term is small, indicating that the filter's main lobe gain is close to the ideal value. To prevent the value of this term from being too large and causing the optimization process to fail, this embodiment sets a hard-coded upper limit V, V≥5. This upper limit ensures that in extreme cases, such as when the main lobe gain is far from the expected value or when the main lobe center frequency shifts, the fitness function can still maintain a reasonable range, avoiding numerical instability caused by excessively large values. The second term is used to evaluate the filter's attenuation performance in the sidelobe and null regions. This term reflects the filter's performance in the sidelobe and null regions by calculating the absolute error between the actual attenuation and the expected target. A smaller error means that the filter design is closer to the predetermined performance indicators.

[0127] This fitness function combines an exponential decay term and an absolute error term to simultaneously optimize the main lobe gain and side lobe attenuation performance of the filter. The main lobe gain is optimized through nonlinear function compression, while the attenuation performance of the side lobes and nulls is ensured by minimizing the absolute error. Together, these two aspects enable the final designed filter to achieve a balanced performance across all dimensions. The introduction of V ensures that the numerical range of the fitness function does not become excessively extreme during optimization, thereby enhancing the stability and convergence of the algorithm.

[0128] Preferably, in step S3, obtaining the optimal filter through the main lobe width bisection method further includes:

[0129] The lower limit is the width of the main lobe corresponding to the unwindowed FFT. The upper limit is the main lobe width corresponding to the Taylor window, with the clutter suppression region level as the upper limit. The tolerance ζ is given, and the main lobe width in the first optimization is... =( + ) / 2;

[0130] If the main lobe width meets the requirements in the initial optimization, then If the main lobe width does not meet the requirements in the initial optimization, then... Iterate according to the main lobe width bisection method until... - <ζ;

[0131] The optimal main lobe width is obtained, and the optimal filter is derived.

[0132] Please see Figure 2 The figure shows the amplitude-frequency response of the Ku-band radar ground clutter MTD filter. Let the filter order be... The normalized Doppler center frequency is 64. The value is 0.3, and the lower limit of the main lobe width is... =0.0140, upper limit is =0.0245, initial main lobe width is =0.0193, radar operating frequency The frequency is 14 GHz, the pulse repetition period (PRT) is 70 μs, and the standard deviation of the ground clutter velocity is selected. The normalized clutter bandwidth is obtained as 0.32 m / s. Given a value of 0.0065, assume clutter band attenuation. -65dB, weak sidelobe attenuation With a tolerance of -30dB and a bisection method iteration tolerance of ζ=0.0001, the final design result after bisection method iteration is shown in the figure. Calculations verify that the main lobe width of the filter designed in this embodiment is 0.0175, which is better than the main lobe width of 0.0232 with a -65dB Taylor window in the conventional MTD filter design, and also better than the main lobe width of 0.0176 with a -30dB Taylor window in the conventional MTD filter design. The signal-to-noise ratio (SNR) gain of the filter designed in this embodiment is 29.1664 dB, the SNR gain with a -30dB Taylor window is 29.2609 dB, and the SNR gain with a -65dB Taylor window is 28.1503 dB. Both the main lobe width and SNR gain of this embodiment are superior to the conventional design with a -65dB Taylor window.

[0133] Please see Figure 3 The image shows the amplitude-frequency response of the MTD filter for meteorological clutter in an X-band radar. The radar operating frequency is... 10GHz, normalized Doppler center frequency The value is -0.4, the pulse repetition period (PRT) is 60 μs, and the standard deviation of the meteorological clutter velocity is... 4 m / s, normalized clutter bandwidth The clutter band attenuation is 0.0503. The value is set to -50dB, with other parameters remaining unchanged. The final design result after bisection iteration is shown in the figure. Calculations verify that the main lobe width of the filter designed in this embodiment is 0.0175, which is better than the main lobe width of 0.0210 achieved with a -50dB Taylor window in the conventional MTD filter design, and also better than the main lobe width of 0.0176 achieved with a -30dB Taylor window in the conventional MTD filter design. The signal-to-noise ratio (SNR) gain of the filter designed in this embodiment is 27.6036 dB, the SNR gain with a -30dB Taylor window is 27.8664 dB, and the SNR gain with a -50dB Taylor window is 27.4658 dB. Both the main lobe width and SNR gain of this embodiment are superior to the conventional design with a -50dB Taylor window.

[0134] Please see Figure 4 The image shows the amplitude-frequency response of the MTD filter for sea clutter in a C-band radar. The radar operating frequency is... 7GHz, normalized Doppler center frequency The value is -0.3, the pulse repetition period (PRT) is 80 μs, and the standard deviation of sea clutter velocity is... The normalized clutter bandwidth is 1.2 m / s. The value is 0.0141, indicating clutter band attenuation. The value is -60dB, with other parameters remaining unchanged. The final design result after bisection iteration is shown in the figure. Calculations verify that the main lobe width of the filter designed in this embodiment is 0.0174, which is significantly better than the 0.0228 main lobe width achieved by adding a -60dB Taylor window in conventional MTD filter design, and also better than the 0.0176 main lobe width achieved by adding a -30dB Taylor window in conventional MTD filter design. The signal-to-noise ratio (SNR) gain of the filter designed in this embodiment is 29.1717 dB, the SNR gain with a -30dB Taylor window is 29.2476 dB, and the SNR gain with a -60dB Taylor window is 28.5235 dB. Both the main lobe width and SNR gain of this embodiment are superior to the conventional design with a -60dB Taylor window.

[0135] Please see Figure 5 The image shows a comparison of the results before and after the completion of the bisection method iteration for the Ku-band radar ground clutter MTD filter. It can be seen that through iteration, this embodiment designs an optimal main lobe width that satisfies all constraints.

[0136] Please see Figure 6 As shown, the main lobe of the MTD filter designed in this embodiment is compared with that of a conventional MTD filter with a Taylor window. It can be seen that the main lobe width of the MTD filter designed in this embodiment is better than that of the conventionally designed MTD filter.

[0137] Example 2

[0138] In some embodiments of this application, a computer device is also provided, including a memory and a processor, wherein the memory stores computer-readable instructions, which, when executed by the processor, cause the processor to perform the steps of the MTD filter design method based on particle swarm optimization as described in any one of Embodiments 1.

[0139] The present invention also provides a storage medium storing computer-readable instructions, which, when executed by one or more processors, cause the one or more processors to perform the steps of the MTD filter design method based on particle swarm optimization as described in any one of Embodiments 1.

[0140] It is understood that, for the aforementioned MTD filter design methods based on particle swarm optimization, if all are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer server or a network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes: USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, optical disks, and other media capable of storing program code.

[0141] Computer-readable storage media may include data signals propagated in baseband or as part of a carrier wave, carrying readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable storage medium may also be any readable medium other than a readable storage medium that can transmit, propagate, or transfer a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof.

[0142] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A method for designing a MTD filter based on a particle swarm optimization algorithm, characterized in that, The method comprises the following steps: S1: obtaining a Doppler center frequency range and a normalized half-width according to a radar type and a clutter type, obtaining a constraint condition comprising a center frequency point, a sidelobe region, a null region and a main lobe region according to the Doppler center frequency range and the normalized half-width, and obtaining the filter information according to the constraint condition and a global pilot vector matrix; S2: designing a fitness function according to a non-linear function for adjusting the filter information and a hard-coded threshold; S3: designing a filter according to the fitness function, and iteratively updating the filter according to a main lobe width bisection method to obtain an optimal filter; In step S3, the filter is designed according to the fitness function, and the method further comprises: The filter is designed by adjusting a sidelobe level and a null level according to the fitness function, and a calculation formula of the fitness function is as follows: wherein, Vp is the main lobe level, V is the hard-coded threshold, Vp is the main lobe level, V is the hard-coded threshold, Vp is the main lobe level, V is the hard-coded threshold, Vp is the main lobe level, V is the hard-coded threshold, Vp is the main lobe level, V is the hard-coded threshold, In step S3, the optimal filter is obtained by iteratively according to the main lobe width bisection method, and the method further comprises: with the main lobe width corresponding to the unwindowed FFT as the lower bound with the main lobe width corresponding to the Taylor window with the added clutter suppression zone level as the upper bound and the tolerance ζ, the main lobe width in the first optimization is ( + ) / 2; If the main lobe width in the first optimization meets the requirement, then If the main lobe width in the first optimization does not meet the requirement, let and iterate according to the main lobe width bisection method until - ζ According to the obtained optimal main lobe width, the optimal filter is obtained.

2. The particle swarm optimization based MTD filter design method of claim 1, wherein, In step S1, the filter information is obtained according to the constraint condition and the global pilot vector matrix, and the method further comprises: S11: constructing the global pilot vector matrix according to a filter order and normalized Doppler frequency full sampling points; S12: calculating a filter frequency response amplitude according to the constraint condition and the global pilot vector matrix, obtaining a frequency response gain according to the filter frequency response amplitude, and obtaining the filter information according to the frequency response gain.

3. The particle swarm optimization based MTD filter design method of claim 2, wherein, In step S11, the global pilot vector matrix is constructed according to a filter order and normalized Doppler frequency full sampling points, and the method further comprises: A calculation formula of a row vector of the global pilot vector matrix is as follows: wherein, is the normalized Doppler frequency full sample point, denotes a row vector whose elements are the indices of the filter order, i.e. , is the filter order, the global pilot vector matrix has its row representing the complex response vector of the filter at the th frequency point; The global pilot vector matrix is obtained according to row vectors of the global pilot vector matrix, and the global pilot vector matrix The calculation formula is as follows: 。 4. The particle swarm optimization based MTD filter design method of claim 3, wherein, In step S12, the filter frequency response amplitude is calculated according to the constraint condition and the global pilot vector matrix, and the method further comprises: According to the normalized Doppler center frequency An initial pilot vector is calculated, the initial pilot vector The calculation formula of the initial pilot vector is as follows: wherein is the imaginary unit; A complex weight vector is obtained according to the initial pilot vector, and the complex weight vector The calculation formula is as follows: wherein the complex weight vector is an optimization variable is an element-wise product of the initial pilot vector and the complex weight vector The initial pilot vector information including filter amplitude, filter phase and filter center frequency directly affects the frequency response characteristics of the filter. A filter frequency response amplitude is obtained according to the constraint condition, the global pilot vector matrix and the complex weight vector, and the filter frequency response amplitude is calculated according to a formula as follows: 。 5. The particle swarm optimization based MTD filter design method of claim 4, wherein, In step S12, the frequency response gain is obtained according to the filter frequency response amplitude, and the method further comprises: The filter frequency response amplitude is normalized to a normal frequency response gain, the normal frequency response gain is calculated by the following formula: According to the normal frequency response gain The frequency response gain is obtained The calculation formula is as follows: 。 6. The particle swarm optimization based MTD filter design method of claim 5, wherein, In step S12, the filter information is obtained according to the frequency response gain, and the method further comprises: A maximum sidelobe level, a maximum null level and a maximum main lobe level are obtained according to the frequency response gain; wherein the side lobe maximum level of the side lobe region is calculated according to the frequency response gain, and the side lobe maximum level of the side lobe region is calculated according to the following formula: wherein is a zero region; The null maximum level of the null region is calculated according to the frequency response gain, and the null maximum level is calculated by a formula as follows: wherein is a null region; The main lobe maximum level of the main lobe region is calculated according to the frequency response gain, and the main lobe maximum level The calculation formula is as follows: 。 7. A computer device, characterized by The computer readable instructions are executed by one or more processors, so that the one or more processors perform the steps of the particle swarm algorithm-based MTD filter design method according to any one of claims 1-6.

8. A storage medium storing computer readable instructions, wherein, The computer readable instructions are executed by one or more processors, so that the one or more processors perform the steps of the particle swarm algorithm-based MTD filter design method according to any one of claims 1-6.

Images