Phase-only array zero setting method and device under multi-frequency point constraint

Abstract

The invention discloses a phase-only array zero setting method and device under multi-frequency point constraint, and belongs to the field of array signal processing, and the method comprises the steps: constraining main lobe gain, and combining factor information to construct a combined cost function; iterative solution is carried out through a genetic algorithm, and an array weight is optimized; and finally, applying the optimized weight to the array to meet the zero demand of the phase-only position under the condition of multiple frequency points. According to the method, the problems of multiple array zero setting angles and multiple frequency points faced by phase-only weighting of the transceiving array in engineering are solved, and the adaptability of the array in a complex environment is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of array signal processing, and more specifically, to a method and device for phase-only array nulling under multi-frequency constraints. Background Art

[0002] Array signal processing is a technology that uses spatially distributed multi-sensors to receive electromagnetic signals and enhances useful signals, suppresses useless signals and noise through signal processing. To improve the performance of phased arrays, it is usually achieved by using complex weighting. However, this method has two problems: one is that the engineering implementation is complex and requires both attenuators and phase shifters for control; the other is that the use of attenuators results in main beam energy loss in the transmitting array, leading to a decrease in the power of the jammer or radar. In contrast, the phase-only array weighting that only uses phase shifters has the advantages of simple engineering implementation and no energy loss.

[0003] Since only the phase of the array weighting coefficient is regulated, the phase-only weighting is a typical non-linear optimization problem. Existing conventional solution ideas mostly focus on various heuristic search optimization methods. Taking the genetic algorithm as an example, by simulating the process of natural selection and reproduction, the dominant individuals are retained generation by generation, and the inferior individuals are eliminated until the optimal solution is selected. However, the performance of the heuristic search method is strongly correlated with the cost function. Currently, the construction of the cost function mostly considers factors such as null depth and flatness, and generates weights with null suppression ability for a specific frequency, and cannot meet the requirements of array nulling at multiple different frequencies and different angles in the actual scenario. Summary of the Invention

[0004] The purpose of the present invention is to overcome the deficiencies of the prior art and provide a method and device for phase-only array nulling under multi-frequency constraints, to solve the problems of multiple array nulling angles and multiple frequencies faced by the phase-only weighting of transceiver arrays in engineering, and to improve the adaptability of the array in complex environments.

[0005] The purpose of the present invention is achieved through the following solutions:

[0006] A method for phase-only array nulling under multi-frequency constraints includes the following steps:

[0007] Constrain the main lobe gain, construct a joint cost function by combining factor information; then perform iterative solution through the genetic algorithm to optimize the array weights; finally, apply the optimized weights to the array to meet the phase-only nulling requirements under multi-frequency conditions.

[0008] Further, the constraining of the main lobe gain specifically includes the following sub-steps:

[0009] Considering that a one-dimensional linear array consists of N omnidirectional antennas with a spacing of d, define the array pattern as:

[0010] F(f,θ) = |w H a(f,θ)| (1);

[0011]

[0012] Wherein, f represents the frequency point, θ represents the included angle between the incident wave direction and the normal direction of the array, and a(f,θ) represents the far-field incident wave steering vector, λ = c / f represents the wavelength, w = [w0, w1,..., w N-1 T represents the array weighting value. When using the pure phase weighting method, |w i | = 1, i = 0, 1,..., N - 1;

[0013] When the frequency point of the main lobe received by the array is f0 and the angle is θ0, the array weighting value is:

[0014]

[0015] Based on the DBF weights, select the discrete phases in the interval Φ M to add perturbations to each element:

[0016]

[0017] Wherein, φ i ∈Φ M , i = 1, 2,..., N - 1; The interval Φ M represents the selection range of the perturbation amount.

[0018] Furthermore, the joint factor information constructs a joint cost function, specifically including the following sub-steps:

[0019] Jointly optimize the null depth, null flatness, and main lobe gain at multiple frequency points, and design the following cost function:

[0020] η = αM(|w H a z |) + βS(|w H a z |) - γM(|w H a b |) (5);

[0021] Wherein, α, β, and γ respectively represent the weighting parameters of the null depth, null flatness, and main lobe gain, and the focus of the cost function is controlled by adjusting the parameters; M(·) and S(·) respectively represent calculating the mean value and standard deviation, w represents the weight to be optimized, and a z represents the array steering vector of the far-field null direction at multiple frequency points, Φ f ​Denote the multi - frequency point set, \(Q\) represents the number of frequency points, Denote the zero - point direction set, \(P\) represents the number of zero - point directions; \(a\) b \(=a(\varPhi\) f ,\(\theta\) b )\(\in C\) N×Q ; where, \(\theta\) b represents the main - lobe pointing; \(a\) z and \(a\) b are generated according to formula (2).

[0022] By optimization, it is hoped to obtain the array weights when the cost function is minimized, that is:

[0023]

[0024] Furthermore, the iterative solution by the genetic algorithm specifically includes the following sub - steps:

[0025] Step 1, encoding: Adopt the binary - encoding method. Let the number of bits of the phase shifter be \(D\), then the minimum quantization interval is \(360 / 2\) D degrees, and encode the phase angles of \(N\) array weights in binary as a \(D\times N\times N\) p matrix, where \(N\) p represents the number of chromosomes;

[0026] Step 2, selection: Substitute all chromosomes into the cost function, evaluate the optimization effect, and arrange them in ascending order according to the value of \(\eta\), and select the chromosomes for crossover operation;

[0027] Step 3, crossover: Adopt the linear - weighted method to fuse the phase angles represented by the paired and selected chromosomes, that is:

[0028] \(\varphi\) c \(=\kappa\varphi\) p1 +(1 - \(\kappa\))\(\varphi\) p2 (7);

[0029] where, \(\kappa\) represents the fusion parameter, \(\varphi\) c represents the phase angle of the offspring, \(\varphi\) p1 and \(\varphi\) p2 represent the phase angles of the parents;

[0030] Step 4, mutation: With a certain probability, add perturbations to the phase angles to increase the diversity of the samples to be optimized.

[0031] Furthermore, the optimization of the array weights specifically includes the following sub - steps:

[0032] Take a certain number of iterations and the difference between two adjacent optimization results as the exit condition to complete the optimization of the array weights.

[0033] Furthermore, the interval \(\varPhi\)M The larger the selection range is, the worse the fidelity of the weight value to the main lobe is. However, the degree of freedom increases, and the optimization of the null points will be better. In actual situations, a compromise is made for selection.

[0034] Furthermore, for different weight values, the smaller the value of the cost function is, the better the optimization effect of the weight value is.

[0035] Furthermore, in step two, the chromosomes selected for crossover operation are specifically selected according to the classical roulette method.

[0036] Furthermore, in step three, it further includes a sub-step: the phase angle is blurred with a period of 2π. The phase angles near 0 degrees and across periods are linearly weighted incorrectly, resulting in pattern distortion and the crossover operation being meaningless. Then, the phase angles are adjusted to the same fuzzy interval before crossover for further processing.

[0037] A phase-only array nulling device under multi-frequency point constraints includes a processor and a memory. A computer program is stored in the memory, and when the computer program is loaded by the processor, it executes the method described in any one of the above.

[0038] The beneficial effects of the present invention include:

[0039] The present invention solves the problems of multiple array nulling angles and multiple frequency points faced by phase-only weighting of transceiver arrays in engineering. By optimizing the design of the cost function, the nulling angles, null depths, and flatness of multiple frequency points are weighted and synthesized. The genetic algorithm is used to iteratively optimize the array weight values to form the null generation ability at different frequency points, improving the adaptability of the array in complex environments.

[0040] Based on the constraint of the main lobe gain, the method of the present invention combines factors such as angle, frequency, null depth, and null flatness to construct a joint cost function, and performs iterative solution through the genetic algorithm. Through simulation verification, compared with the conventional phase-only angle nulling method for a single frequency point, the array weight values optimized by the present invention simultaneously have the suppression ability for multiple frequency points and can be used for wideband array nulling reception. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0042] Figure 1 It is a schematic diagram of a uniform linear array model;

[0043] Figure 2 Schematic diagram of binary coding;

[0044] Figure 3 Flowchart of the method of the present invention;

[0045] Figure 4 Array response pattern of DBF weights at 7 GHz and 8 GHz frequency points, where (a) is the 7 GHz response and (b) is the 8 GHz response;

[0046] Figure 5 Optimization diagram of the conventional phase-only optimization method at 7 GHz, (a) is the 7 GHz response and (b) is the 8 GHz response; optimization of the conventional phase-only optimization method at 8 GHz, (c) is the 7 GHz response and (b) is the 8 GHz response;

[0047] Figure 6 Method proposed by the present invention, through joint optimization of 7 GHz and 8 GHz, (a) is the 7 GHz response and (b) is the 8 GHz response. Specific implementation manner

[0048] All features disclosed in all embodiments in this specification, or all steps in all methods or processes implicitly disclosed, except for mutually exclusive features and / or steps, can be combined and / or extended and replaced in any manner.

[0049] The present invention aims to improve the nulling ability of the phased array for multiple frequency points and multiple angles by optimizing the design cost function and using the genetic algorithm to solve the phase-only weights. The specific implementation process is as follows:

[0050] Without loss of generality, consider a one-dimensional receiving array, as Figure 1 shown. Consider a one-dimensional linear array composed of N omnidirectional antennas with a spacing of d, and define the array pattern as:

[0051] F(f,θ) = |w H a(f,θ)| (1);

[0052]

[0053] where f represents the frequency point, θ represents the angle between the incoming wave direction and the normal of the array, a(f,θ) represents the far-field incoming wave steering vector, λ = c / f represents the wavelength, w = [w0, w1,..., w N-1 T represents the array weighting value. When using the phase-only weighting method, |w i | = 1, i = 0, 1,..., N - 1.

[0054] When the frequency point of the main lobe received by the array is f0 and the angle is θ0, the conventional array weighting value (DBF weight) is:​

[0055]

[0056] To meet the main lobe gain constraint of only phase weighting, on the basis of the conventional DBF weights, discrete phases in the interval Φ M are selected to add perturbations to each array element:

[0057]

[0058] where φ i ∈Φ M , i = 1, 2,..., N - 1. The interval Φ M represents the selection range of the perturbation amount. The larger the range is selected, the worse the fidelity of the weights to the main lobe is. However, due to the increase in degrees of freedom, the optimization of the nulls will be better. In actual situations, a trade-off needs to be made.

[0059] To optimize the weights of only phase weighting, a cost function needs to be defined. The quality of the designed cost function significantly affects the optimization result of the weights. In the present invention, the cost function is designed by jointly optimizing the null depth, null flatness, and main lobe gain at multiple frequency points as follows

[0060] η = αM(|w H a z |) + βS(|w H a z |) - γM(|w H a b |)(5);

[0061] where α, β, and γ respectively represent the weighting parameters of the null depth, null flatness, and main lobe gain, and the emphasis of the cost function is controlled by adjusting the parameters. M(·) and S(·) respectively represent calculating the mean and standard deviation, w represents the weights to be optimized, a z represents the array steering vector of the far-field null direction at multiple frequency points, Φ f represents the set of multiple frequency points, Q represents the number of frequency points, represents the set of null directions, and P represents the number of null directions. a b = a(Φ f , θ b ) ∈ C N×Q , where θ b represents the main lobe pointing. a z and a bIt can be generated according to formula (2). It can be seen that the above cost function comprehensively considers factors such as frequency point information, angle information, null depth, null flatness, and main beam gain, which belongs to a joint optimization problem. It can be seen from formula (5) that for different weights, the smaller the value of its cost function, the better the optimization effect of the weight. Through optimization, it is hoped to obtain the array weight when the cost function is minimized, that is:

[0062]

[0063] Use the genetic algorithm to complete the iterative optimization of the array weight w. Its basic process includes: encoding, selection, crossover, and mutation. Compared with the genetic algorithm used in other fields, in the process of array optimization, considering the physical model of the discrete phase shifter and the 2π-periodic ambiguity of the weight phase angle, two improvements are made in the information encoding and crossover links to make the genetic algorithm fit the application requirements.

[0064] Step 1: Encoding. Since in the beamforming of the analog phased array, the phase shifter is mostly used to adjust the phase of the array elements, according to the number of bits of the phase shifter, the phase adjustment shows a discrete characteristic. Therefore, the binary encoding method is adopted. Assuming that the number of bits of the phase shifter is D, the minimum quantization interval is 360 / 2 D degrees, and the phase angles of N array weights are encoded in binary as a DN×N p matrix, where N p represents the number of chromosomes. Figure 2 The encoding schematic diagram when D = 6 is given.

[0065] Step 2: Selection. Substitute all chromosomes into the cost function, evaluate the optimization effect, and arrange them in ascending order according to the value of η, and select the chromosomes for crossover operation according to the classic roulette method.

[0066] Step 3: Crossover. Adopt the linear weighting method to fuse the phase angles represented by the paired and selected chromosomes, that is

[0067] φ c = κφ p1 +(1 - κ)φ p2 (7)

[0068] where κ represents the fusion parameter, φ c represents the phase angle of the offspring, φ p1 and φ p2 represent the phase angles of the parents. Since the phase angle is ambiguous with a period of 2π, the linear weighting of the phase angles near 0 degrees and across periods is incorrect, resulting in the distortion of the radiation pattern and the meaninglessness of the crossover operation. Therefore, it is necessary to adjust them to the same ambiguity interval before crossover for processing.

[0069] Step 4: Mutation. With a certain probability, a perturbation is added to the phase angle to increase the diversity of the samples to be optimized.

[0070] Taking a certain number of iterations and the difference between two adjacent optimization results as the termination condition, the optimization of the array weights is completed. Figure 3 The algorithm flow chart is given. Applying the optimized weights to the array can meet the zero-phase-only position requirement under multi-frequency conditions.

[0071] In other embodiments, consider the following simulation scenario: the element spacing is 0.018 m, the number of elements is 8, and it is required to suppress the signal with a direction of arrival of 60 degrees using the phase-only method when the frequencies are 7 GHz and 8 GHz and the main beam points to the normal direction of 0 degrees. The quantization bit number of the phase shifter is 6 bits. When using the genetic algorithm for iterative optimization, the number of chromosomes is set to 500, and the termination condition is that the number of iterations exceeds 100 times. In terms of the cost function, the weights are set as α = 20, β = 1, and γ = 50. In the embodiment, first, the array response results of directly using the DBF weights without any phase-only optimization are given, as shown in Figure 4 (a) and (b). It can be seen that the unoptimized array weights cannot form a null at the direction of arrival of 60 degrees (the position of the virtual vertical line in the figure).

[0072] To compare with the conventional phase-only array nulling method, Figure 5 the array response results optimized by the conventional method at 7 GHz and 8 GHz are given. It can be seen that the weights optimized by the conventional method cannot be shared among multiple frequencies. If a null is formed at 7 GHz, then a null cannot be formed at 8 GHz, and vice versa for 8 GHz.

[0073] Using the method proposed in the present invention, the results are as shown in Figure 6 As shown. By setting the multi-frequency joint cost function and jointly optimizing the phase-only weights, the obtained weights have the suppression ability at both 7 GHz and 8 GHz and accurately form a null at the 60-degree direction. Therefore, the array weights calculated by the method of the present invention have the multi-frequency joint suppression ability.

[0074] It should be noted that within the protection scope defined in the claims of the present invention, the following embodiments can be combined and / or extended, replaced in any logical manner from the above specific implementation manners, such as the disclosed technical principles, disclosed technical features, or implicitly disclosed technical features.

[0075] Embodiment 1

[0076] A phase-only array nulling method under multi-frequency constraints, characterized by comprising the following steps:

[0077] Constrain the main lobe gain, construct a joint cost function by combining factor information; then perform iterative solution through the genetic algorithm to optimize the array weights; finally, apply the optimized weights to the array to meet the zero-phase position requirement under multi-frequency conditions.

[0078] Embodiment 2

[0079] Based on Embodiment 1, the constraint of the main lobe gain specifically includes the following sub-steps:

[0080] Considering that a one-dimensional linear array consists of N omnidirectional antennas with a spacing of d, the array pattern is defined as:

[0081] F(f,θ)=|w H a(f,θ)|(1);

[0082]

[0083] where f represents the frequency point, θ represents the angle between the incoming wave direction and the array normal, a(f,θ) represents the far-field incoming wave steering vector, λ=c / f represents the wavelength, w=[w0,w1,...,w N-1 T represents the array weighting value. When using the phase-only weighting method, |w i |=1, i=0,1,...,N - 1;

[0084] When the frequency point of the main lobe received by the array is f0 and the angle is θ0, the array weighting value is:

[0085]

[0086] Based on the DBF weights, select the discrete phases in the interval Φ M to add perturbations to each element:

[0087]

[0088] where φ i ∈Φ M , i=1,2,...,N - 1; the interval Φ M represents the selection range of the perturbation amount.

[0089] Embodiment 3

[0090] Based on Embodiment 2, the construction of the joint cost function by combining factor information specifically includes the following sub-steps:

[0091] Combine multi-frequency point to optimize the null depth, null flatness, and main lobe gain, and design the following cost function:

[0092] η=αM(|w H ​a z |)+βS(|w H a z |)-γM(|w H a b |)(5);

[0093] Among them, α, β, and γ respectively represent the weighted parameters of the zero - depth, zero - flatness, and main - lobe gain. The focus of the cost function is controlled by adjusting the parameters; M(·) and S(·) respectively represent taking the mean value and standard deviation, w represents the weight value to be optimized, a z represents the array steering vector of the multi - frequency far - field zero - direction, Φ f represents the multi - frequency point set, Q represents the number of frequency points, represents the zero - direction set, P represents the number of zero - directions; a b =a(Φ f ,θ b )∈C N×Q ; among them, θ b represents the main - lobe pointing; a z and a b are generated according to formula (2).

[0094] By optimization, it is hoped to obtain the array weight when the cost function is minimized, that is:

[0095]

[0096] Example 4

[0097] Based on Example 3, the iterative solution by the genetic algorithm specifically includes the following sub - steps:

[0098] Step 1, encoding: Using binary encoding, assuming the number of bits of the phase shifter is D, then the minimum quantization interval is 360 / 2 D degrees, and the phase angles of N array weights are encoded in binary as a DN×N p matrix, where N p represents the number of chromosomes;

[0099] Step 2, selection: Substitute all chromosomes into the cost function, evaluate the optimization effect, and arrange them in ascending order according to the value of η, and select the chromosomes for crossover operation;

[0100] Step 3, crossover: Using linear weighting, fuse the phase angles represented by the paired and selected chromosomes, that is:

[0101] φ c =κφ p1 +(1 - κ)φ p2 (7);

[0102] Among them, κ represents the fusion parameter, and φ c represents the phase angle of the offspring, and φ p1 and φ p2 represent the phase angles of the parents;

[0103] Step 4, Mutation: With a certain probability, add perturbations to the phase angles to increase the diversity of the samples to be optimized.

[0104] Example 5

[0105] Based on Example 1, the optimization of the array weights specifically includes the following sub-steps:

[0106] With a certain number of iterations and the difference between two adjacent optimization results as the exit condition, complete the optimization of the array weights.

[0107] Example 6

[0108] Based on Example 2, the larger the selection range of the interval Φ M , the worse the fidelity of the weights to the main lobe, but the degree of freedom increases, and the optimization of the null points will be better. A compromise is made in actual situations.

[0109] Example 7

[0110] Based on Example 3, for different weights, the smaller the value of the cost function, the better the optimization effect of the weights.

[0111] Example 8

[0112] Based on Example 4, in Step 2, the chromosomes selected for the crossover operation are specifically selected according to the classical roulette method.

[0113] Example 9

[0114] Based on Example 4, in Step 3, it further includes a sub-step: the phase angles are blurred with a period of 2π. The phase angles near 0 degrees and across periods are linearly weighted incorrectly, resulting in pattern distortion, and the crossover operation is meaningless. Therefore, the phase angles are adjusted to the same blurred interval before crossover for further processing.

[0115] Example 10

[0116] A multi-frequency point constrained phase-only array nulling device includes a processor and a memory. The memory stores a computer program, and when the computer program is loaded by the processor, it executes the method described in any one of Examples 1 to 9.

[0117] The units involved in the embodiments of the present invention can be implemented in software or in hardware, and the described units can also be provided in a processor. Among them, the names of these units do not, in some cases, constitute a limitation on the units themselves.

[0118] According to one aspect of the embodiments of the present invention, there is provided a computer program product or a computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the methods provided in the above various alternative implementations.

[0119] As another aspect, the embodiments of the present invention further provide a computer-readable medium, which may be included in the electronic device described in the above embodiments; or may exist separately without being assembled into the electronic device. The above computer-readable medium carries one or more programs, and when the one or more programs are executed by an electronic device, the electronic device implements the methods described in the above embodiments.

Claims

1. A method for phase-only array nulling under multi-frequency point constraints, characterized in that It includes the following steps: Constraint the main lobe gain, construct a joint cost function by combining factor information; then perform iterative solution through a genetic algorithm to optimize the array weights; finally, apply the optimized weights to the array to meet the zero-phase position requirement under multi-frequency conditions.

2. The method for nulling a pure-phase array under multi-frequency point constraints according to claim 1, characterized in that The constraint of the main lobe gain specifically includes the following sub-steps: Considering that a one-dimensional linear array consists of N omnidirectional antennas with a spacing of d, the array pattern is defined as: F(f,θ) = |w H a(f,θ)| (1); where \(f\) represents the frequency point, \(\theta\) represents the angle between the direction of the incoming wave and the normal direction of the array, and \(a(f,\theta)\) represents the far-field incoming wave steering vector. \(\lambda = c / f\) represents the wavelength, \(w=[w_0,w_1,\cdots,w\) N-1 T represents the array weighting value. When using the only phase weighting method, \(|w\) i | = 1, \(i = 0,1,\cdots,N - 1\);​ When the frequency of the main lobe received by the array is f0 and the angle is θ0, the array weighting value is: Based on the DBF weights, select the discrete phases in the interval Φ M to add perturbations to each array element: where φ i ∈Φ M , i = 1, 2, ..., N - 1; the interval Φ M represents the selection range of the disturbance quantity.

3. The method for nulling an all-phase array under multi-frequency point constraints according to claim 2, wherein The construction of the joint cost function by combining factor information specifically includes the following sub-steps: Jointly optimize the null depth, null flatness, and main lobe gain at multiple frequencies, and design the following cost function: η = αM(|w H a z |) + βS(|w H a z |) - γM(|w H a b |) (5); Among them, α, β, and γ represent the weighted parameters of the zero-depth, zero-flatness, and main-lobe gain respectively, and the focus of the cost function is controlled by adjusting the parameters; M(·) and S(·) represent calculating the mean and standard deviation respectively, w represents the weight to be optimized, and a z represents the array steering vector of the multi-frequency far-field zero direction, Φ f represents the multi-frequency point set, Q represents the number of frequency points, represents the zero direction set, and P represents the number of zero directions; a b = a(Φ f , θ b ) ∈ C N×Q ; among them, θ b represents the main-lobe pointing; a z and a b are generated according to formula (2); By optimization, the array weights when the cost function is minimized are expected to be obtained, that is:

4. The method for nulling an all-phase array under multi-frequency point constraints according to claim 3, wherein The iterative solution through the genetic algorithm specifically includes the following sub-steps: Step 1, Encoding: Using binary encoding, assuming the number of bits of the phase shifter is D, the minimum quantization interval is 360 / 2 D degrees, and encoding the phase angles of N array weights in binary form into a DN×N p matrix, where N p represents the number of chromosomes; Step 2, Selection: Substitute all chromosomes into the cost function, evaluate the optimization effect, and arrange them in ascending order according to the value of η, and select the chromosomes for crossover operation; Step 3, Crossover: Adopt the method of linear weighting to fuse the phase angles represented by the paired and selected chromosomes, that is: φ c = κφ p1 + (1 - κ)φ p2 (7); where κ represents the fusion parameter, and φ c represents the phase angle of the offspring, and φ p1 and φ p2 represent the phase angles of the parents; Step 4, Mutation: With a certain probability, add perturbations to the phase angles to increase the diversity of the samples to be optimized.

5. The method for nulling a phased-only array under multi-frequency point constraints according to claim 1, wherein The optimization of the array weights specifically includes the following sub-steps: Take a certain number of iterations and the difference between two adjacent optimization results as the exit condition to complete the optimization of the array weights.

6. The method for nulling of the phase-only array under multi-frequency point constraints according to claim 2, wherein Interval Φ M The larger the selection range of M is, the worse the fidelity of the weight value to the main lobe is, but the degree of freedom increases, and the optimization of the null points will be better. A compromise is chosen in actual situations.

7. The method for nulling of the only-phase array under multi-frequency point constraints according to claim 3, wherein For different weights, the smaller the value of the cost function, the better the optimization effect of the weights.

8. The method for nulling of the phase-only array under multi-frequency point constraints according to claim 4, characterized in that, In Step 2, the chromosomes selected for the crossover operation are specifically selected according to the classical roulette method.

9. The method for nulling an all-phase array under multi-frequency point constraints according to claim 4, characterized in that In Step 3, it also includes a sub-step: The phase angles are ambiguous with a period of 2π. The linear weighting of the phase angles near 0 degrees and across periods is incorrect, resulting in pattern distortion and the crossover operation being meaningless. Then, the phase angles are adjusted to the same ambiguity interval before crossover for processing.

10. A phase-only array nulling device under multi-frequency point constraints, characterized in that It includes a processor and a memory, and a computer program is stored in the memory. When the computer program is loaded and executed by the processor, the method described in any one of claims 1 to 9 is performed.

Images