Phased array multi-target capturing method based on adaptive filter

Abstract

The invention discloses a phased array multi-target capturing method based on an adaptive filter, and belongs to the technical field of unmanned aerial vehicle phased array communication. The method comprises the following steps: setting an initial parameter combination of an adaptive filter; obtaining a mixed signal vector at the current moment, distinguishing far and near target signals, and preliminarily setting filter parameters; establishing a joint optimization function including phased array filter parameters, target capture probability and target capture speed; and a self-adaptive Bregman ADMM algorithm is adopted to carry out dynamic parameter updating on the joint optimization function, finally optimized filter parameters are loaded to the phased array system to form a synthetic beam pointing to multiple targets, differential gain distribution is carried out on far and near targets, and multi-target synchronous capture is realized. The method can distinguish and capture far and near targets, and is especially suitable for the measurement and control communication scene of the unmanned aerial vehicle in a complex electromagnetic environment.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of unmanned aerial vehicle phased array communication, and particularly relates to a phased array multi-target capturing method based on an adaptive filter, which is suitable for unmanned aerial vehicle cluster communication, low-altitude detection, dynamic target tracking and the like. BACKGROUND

[0002] The prior art has the following defects:

[0003] (1) Low efficiency of beam resource allocation: the traditional phased array adopts fixed period scanning, and cannot meet the tracking accuracy requirement in a large-scale target scene;

[0004] (2) Serious interference of near and far targets: the current phased array system cannot distinguish near and far targets, resulting in poor capturing accuracy;

[0005] (3) Slow convergence speed of filter: the conventional algorithm is difficult to converge in the unmanned aerial vehicle motion scene, and cannot adapt to the high-speed dynamic environment;

[0006] The existing multi-target capturing algorithm can achieve approximate capturing capability, but cannot solve the near and far target problem in dynamic tracking, and does not integrate a closed-loop feedback mechanism. SUMMARY

[0007] Therefore, the application provides a phased array multi-target capturing method based on an adaptive filter. The application can realize multi-target accurate capturing, combines near and far target distinguishing and joint optimization, and solves the problems of difficulty in distinguishing near and far targets, low capturing accuracy and high computational complexity of the traditional phased array system in multi-target capturing.

[0008] In order to achieve the above purpose, the technical scheme adopted by the application is as follows:

[0009] A phased array multi-target capturing method based on an adaptive filter is applied to an unmanned aerial vehicle phased array system, and includes the following steps:

[0010] Step a: before starting the measurement and control data transmission of the phased array system, an initial parameter combination of the adaptive filter is set;

[0011] Step b: a mixed signal vector at the current time is acquired through a phased array receiving channel, near and far target signals are distinguished based on the mixed signal vector, and filter parameters are preliminarily set according to the near and far degrees of the targets;

[0012] Step c: based on a signal model, a joint optimization function containing phased array filter parameters, target capturing probability and target capturing speed is established;

[0013] Step d involves using the adaptive Bregman ADMM algorithm to dynamically update the parameters of the joint optimization function, loading the final optimized filter parameters into the phased array system to form a synthetic beam pointing to multiple targets, and performing differentiated gain allocation for near and far targets to achieve synchronous acquisition of multiple targets.

[0014] Furthermore, the initial parameter combination of the adaptive filter in step a is:

[0015]

[0016] in, This indicates the total number of subarrays in the phased array antenna. For the first The initial weight coefficients of each submatrix, i=1,2,...,N, where the superscript T denotes the transpose of the matrix.

[0017] Furthermore, in step b, the mixed signal vector at the current moment is:

[0018]

[0019] in, For the target total number, For the first The reflectance coefficient of the target For the first The relative distance between the targets This is the path loss index. Let m be the guidance vector for the m-th target. This represents the direction of arrival (DOA) of the m-th target. The distance between adjacent array elements. Represents the total number of subarrays in a phased array antenna system. For the signal wavelength, It is additive white Gaussian noise.

[0020] Furthermore, in step b, the signals of distant and near targets are distinguished based on the mixed signal vector, and the filter parameters are initially set according to the distance of the targets. The specific method is as follows:

[0021] Step b01, generate using phased array antenna Each preformed beam points to a different angle. For mixed signals Spatial filtering is performed to separate the independent signal components of each target. ,in The superscript H indicates conjugate transpose;

[0022] Step b02, for each independent signal component Calculate the time-domain average power :

[0023]

[0024] in, Indicates the start time of the current signal processing moment. This indicates the length of the observation time window used for power estimation, in seconds;

[0025] Then calculate the relative distance to the target. :

[0026]

[0027] in, The nominal received power at the reference distance;

[0028] Step b03, according to Calculate dynamic distance threshold , will satisfy The target is classified as a short-range set. The rest are grouped into a remote set, and classification label vectors are generated. ,in, For the first The classification identifier of each target, if Then it means the first The first goal is a short-term goal, if... Then it means the first The target is a remote target;

[0029] Step b04: Differentiate the initial filter weights based on the classification identifier:

[0030]

[0031] in, Let be the normalization coefficient, so that , It is the first Each sub-array is paired with the first One target direction The magnitude of the guiding vector component. It characterizes the spatial response intensity of the subarray to the signal in that direction.

[0032] Furthermore, the joint optimization function in step c is:

[0033]

[0034] st

[0035]

[0036] in, For the first The probability of capturing a target. The variance of additive white Gaussian noise characterizes the noise power of the receiving channel. The capture time for the slowest target. For the first Detection range of each target For the first The speed of movement of the target These are weighting coefficients. This is the minimum overall capture probability threshold required by the system. Where W is the longest capture time allowed for the task, W is the parameter combination of the adaptive filter, and Z is an auxiliary variable.

[0037] Furthermore, in step d, the adaptive Bregman ADMM algorithm is used to dynamically update the parameters of the joint optimization function. Specifically, the method is as follows:

[0038] Step d01: Decompose Z into weighted modulus constraints. Capture probability constraints Capture time constraints Three auxiliary variables are used to decompose the constraints on the joint optimization function:

[0039]

[0040] Step d02: Select KL divergence as the Bregman distance function to match the exponential structure in the capture probability.

[0041]

[0042] Setting the augmented Lagrange function:

[0043]

[0044] definition For dual variables, corresponding to the Lagrange multipliers of the three constraints;

[0045] in, For adaptive penalty parameters;

[0046] Step d03, for the non-convex terms in the objective function Perform a first-order Taylor expansion:

[0047]

[0048] Where, gradient Calculated using the chain rule:

[0049]

[0050] The linearized objective function is obtained as

[0051]

[0052] Step d04, initialize iteration counter , set maximum iteration number and convergence threshold , perform alternating direction update:

[0053] (1) update the primary variable :

[0054]

[0055] Solve using complex field conjugate gradient method;

[0056] (2) update the auxiliary variable :

[0057]

[0058]

[0059]

[0060] where is a projection operator that truncates matrix elements to the unit circle;

[0061] Use the LogSumExp function to smooth the maximum value term:

[0062]

[0063] where is used to control the degree of smoothing;

[0064] (3) update the dual variable :

[0065]

[0066] (4) according to the original residual and the dual residual , adjust the adaptive step size according to the rules:

[0067]

[0068] where is the adjustment factor, is the balance threshold;

[0069] (5) judge whether the following termination conditions are met:

[0070] Original feasibility:

[0071] Dual feasibility:

[0072] Relative change of objective function:

[0073] If all the above termination conditions are met, the iteration is ended, otherwise the alternating direction update is continued.

[0074] Further, in step d, the final optimized filter parameters are loaded into the phased array system to form a synthetic beam pointing to multiple targets, in a specific manner:

[0075] (1) The optimized filter weight coefficients are distributed to each subarray of the phased array antenna, each subarray containing adjacent elements, and the direction of arrival angle of each subarray is corrected through coordinate transformation ;

[0076] (2) Based on the adaptive LMS algorithm, the signals between subarrays are phase-aligned to eliminate the position deviation of the elements caused by the change in the attitude of the unmanned aerial vehicle, and the coherent superposition condition of the synthetic beam is met ; wherein, represents the maximum phase difference between adjacent subarrays, corresponding to a path difference constraint of 0.125λ;

[0077] (3) According to the dynamic position of the target, the beam pointing parameters are updated, and a three-level beam forming architecture is adopted: the first level uses an analog phase shifter to realize ±45° coarse scanning within the subarray, the second level uses a digital phase shifter for ±5° fine adjustment, and the third level uses a digital beam former to complete 0.1° level micro pointing;

[0078] (4) The beam state feedback information of each subarray is fused through federated filtering to dynamically adjust the 3dB bandwidth of the synthetic beam :

[0079]

[0080] wherein, is the total number of all elements participating in the synthesis.

[0081] Further, in step d, differentiated gain distribution is performed on the near and far targets, in a specific manner:

[0082] (1) According to the relative distance of the target, a nonlinear gain function is set:

[0083] When , an attenuation gain of is assigned;

[0084] allocating enhanced gain to;

[0085] (2) Calculate the signal-to-noise ratio :

[0086]

[0087] Based on the signal-to-noise ratio Dynamic adjustment of beam residence time, when , the residence time is extended to:

[0088]

[0089] Where min represents the minimum number; at the same time, the beam width is compressed to:

[0090]

[0091] Where, is the current active element number;

[0092] (3) Adopt a double priority arbitration mechanism, the first level allocates 60% of the gain resources according to the threat degree, and the second level allocates the remaining resources according to the motion speed , meet ; Where is the second level resource amount allocated to the entity, is the total amount of remaining resources after the first level allocation, is the motion speed of the entity, is the total number of entities, is the sum of the motion speeds of all entities;

[0093] (4) Generate a set of gain allocation schemes through Monte Carlo sampling, and select the scheme that meets the following conditions as the optimal scheme for loading and execution:

[0094] And .

[0095] The beneficial effects of the present application are:

[0096] 1. Near and far target differentiation processing: through dynamic distance classification and weight gradient initialization, high gain is allocated to short-range targets to improve accuracy, and long-range targets reduce noise interference and optimize signal-to-noise ratio.

[0097] 2. Multi-target cooperative optimization: joint optimization function balances capture probability and speed, ensures minimum performance threshold, preferentially processes time-consuming targets, and avoids task failure.

[0098] 3. Efficient parameter update: Bregman ADMM algorithm improves the convergence speed of non-convex problems, supports real-time processing, and meets the low delay requirements of unmanned aerial vehicles. DETAILED DESCRIPTION

[0099] The technical solutions of the present application will be further described below. Obviously, these contents are only a part of the embodiments of the present application, not all the embodiments. Based on the following embodiments, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the present application.

[0100] A phased array multi-target capture method based on an adaptive filter is applied to a phased array system of an unmanned aerial vehicle, comprising the following steps:

[0101] Step a, initializing filter parameters. Before the phased array system starts the transmission of measurement and control data, set the initial parameter combination of the adaptive filter.

[0102] Step b, multi-target signal acquisition and calculation. Obtain the mixed signal vector at the current time through the phased array receiving channel, distinguish the far and near target signals based on the mixed signal vector, and preliminarily set the filter parameters according to the far and near degree of the target.

[0103] Step c, constructing a joint optimization function. Based on the signal model, a joint optimization function containing phased array filter parameters, target capture probability and target capture speed is established.

[0104] Step d, dynamic parameter update and target capture. The adaptive Bregman ADMM algorithm is used to dynamically update the joint optimization function, and the finally optimized filter parameters are loaded into the phased array system to form a synthesized beam pointing to multiple targets. The differentiated gain allocation of far and near targets is realized, and the multi-target synchronous capture is completed.

[0105] This embodiment verifies the robustness of the multi-target capture method in the far and near flight for the communication scene of the medium and low altitude unmanned aerial vehicle. The system configuration environment is as follows:

[0106] Table 1 System configuration environment

[0107]

[0108] In step a, the initial parameter combination of the adaptive filter is:

[0109]

[0110] Wherein: N represents the number of phased array elements, is the initial weight coefficient of the th element.

[0111] Meanwhile, initialize iteration counter , set maximum iteration number and convergence threshold .

[0112] In step b, the mixed signal vector at current time is:

[0113]

[0114] Wherein: is the total number of targets, is the reflection coefficient of the th target, is the relative distance of the th target, is the path loss exponent (typical value 2-4), is the steering vector of the mth target, is the adjacent element spacing, represents the total number of phased array antenna units, is the signal wavelength, is the complex additive white Gaussian noise.

[0115] In step b, based on the mixed signal vector , the near and far target signals are distinguished, and the filter parameters are preliminarily set according to the near and far degree of the target. The specific way is:

[0116] (1) Multi-beam signal separation: generate pre-beams through phased array antenna, respectively pointing to different angles , perform spatial filtering on the mixed signal , and separate out the independent signal components of each target , wherein ;

[0117] (2) Time delay power mapping: calculate the time domain average power of each separated signal , combine the path loss model , and back-propagate the target relative distance , wherein is the nominal received power at the reference distance;

[0118] (3) Dynamic threshold classification: calculate the dynamic distance threshold according to , classify the targets meeting as the near range set , and the rest as the far range set, generate a classification identification vector ( represents near range, represents far range);

[0119] (4) Weight gradient initialization: Differentiate the initial filter weights according to the classification results:

[0120]

[0121] where γ is a normalization coefficient, so that .

[0122] The joint optimization objective function in step c is:

[0123]

[0124] s.t.

[0125]

[0126] where: is the capture probability of the i-th target, is the variance of additive white Gaussian noise, representing the noise power of the receiving channel, is the capture time of the slowest target, is the detection distance of the i-th target, is its motion speed, is the weighting coefficient, is the minimum comprehensive capture probability threshold required by the system, is the maximum capture time allowed by the task, both of which are pre-set according to the requirements of the unmanned aerial vehicle task. In step d, the adaptive Bregman ADMM algorithm is used to update the parameters of the joint optimization function dynamically, including:

[0127] (1) Introduce auxiliary variables ,

[0128] , , , , ,

[0129] , , ,

[0130] (2) Choose KL divergence (Kullback-Leibler divergence) as the Bregman distance function, which matches the exponential structure in the capture probability:

[0131]

[0132] Augmented Lagrangian function:

[0133]

[0134] Parameter definition: For the dual variable, the Lagrange multiplier corresponding to the three constraints; For the adaptive penalty parameter.

[0135] (3) First-order Taylor expansion (linearization) is performed on the non-convex term in the objective function:

[0136]

[0137] Where the gradient is calculated by the chain rule:

[0138]

[0139] The linearized objective function:

[0140]

[0141] (4) Alternating direction update steps

[0142] Step 1: Main variable update:

[0143]

[0144] Step 2: Auxiliary variable update:

[0145] Amplitude constraint term :

[0146]

[0147] Where is the projection operator, which truncates the matrix elements to within the unit circle.

[0148] Capture probability term :

[0149]

[0150] Time constraint term :

[0151]

[0152] Use the LogSumExp function to smooth the maximum value term:

[0153]

[0154] Where, is the control of the smoothing degree.

[0155] Step 3: Dual variable update:​

[0156]

[0157] Step 4: Adaptive step size adjustment:

[0158] According to the original residual error and the dual residual error , adjust the step size according to the rule:

[0159]

[0160] where is the adjustment factor, is the balance threshold.

[0161] (5) Convergence guarantee and termination condition

[0162] The objective function needs to satisfy the Kurdyka-Łojasiewicz (KL) property to ensure global convergence to a stable point in non-convex scenarios, and the termination criterion is:

[0163] ① Original feasibility:

[0164] ② Dual feasibility:

[0165] ③ Relative change of objective function:

[0166] In step d, the final optimized filter parameters are loaded into the phased array system to form a synthesized beam pointing to multiple targets, including:

[0167] (1) Assign the optimized filter weight coefficients to each subarray module of the phased array, each subarray contains adjacent elements , where is the total number of elements in the phased array and is the total number of targets, and correct the direction of arrival angle of each subarray through coordinate transformation;

[0168] (2) Based on the adaptive LMS algorithm, align the phases of signals between subarrays to eliminate the position deviation of array elements caused by changes in the attitude of the unmanned aerial vehicle, and meet the coherent superposition condition of the synthesized beam , where represents the maximum phase difference between adjacent subarrays, corresponding to a path difference constraint of 0.125λ;

[0169] (3) Update the beam pointing parameters according to the target dynamic position and adopt a three-level beamforming architecture: the first level achieves ±45° coarse scanning in the subarray (using analog phase shifters), the second level performs ±5° fine adjustment through digital phase shifters, and the third level completes 0.1° level micro pointing using digital beamformers;

[0170] (4) The beam state feedback information of each subarray is fused by federated filtering (federated filtering is defined as a data fusion structure of distributed Kalman filtering and main filter) to dynamically adjust the 3dB bandwidth of the synthesized beam. ,in To activate the total number of array elements ( (This represents the number of active elements in the s-th subarray).

[0171] In step d, differentiated gain allocation is applied to near and far targets to complete multi-target synchronous acquisition, including:

[0172] (1) Based on the relative distance to the target Set the nonlinear gain function:

[0173] Close-range targets ( )distribute attenuation gain

[0174] distant targets ( )distribute Enhanced gain ( Indicates the first (Target gain coefficient)

[0175] (2) Calculate the signal-to-noise ratio:

[0176]

[0177] Based on signal-to-noise ratio ( , Dynamically adjust beam dwell time (for noise power): when At that time, the stay period will be extended to ( (to meet the maximum acquisition time constraint), while compressing the beamwidth to ( (Number of currently active array elements);

[0178] (3) A dual-priority arbitration mechanism is adopted: the first level allocates 60% of the gain resources according to the threat level (the threat level is input by the external tactical system), and the second level allocates resources according to the movement speed. Allocate remaining resources;

[0179] (4) Generate a set of gain allocation schemes through Monte Carlo sampling, and select the schemes that satisfy the requirements. and Optimal scheme loading execution, For small capture probability.

[0180] Through the verification method of combining the microwave darkroom with the field measurement, the following quantitative results are obtained:

[0181] Table 2 Performance test results

[0182]

[0183] The embodiment verifies the engineering applicability of the method in a complex electromagnetic environment, and the core innovative advantages are embodied in:

[0184] (1) The dynamic weight gradient mechanism compresses the capture probability difference of far and near targets to 8.7%;

[0185] (2) The Bregman ADMM optimization algorithm realizes fast iterative convergence, and the speed is increased by 40% compared with the standard ADMM;

[0186] (3) The joint optimization function design maintains a high capture probability under low signal-to-noise ratio;

[0187] (4) The differentiated allocation of beam gain makes the gain of short-range targets increase by 4.3dB, effectively suppressing the blocking effect of strong targets.

[0188] The application can be used for a vehicle-mounted phased array platform to realize multi-target capture. Combined with the joint optimization function and the improved ADMM algorithm, the problems of channel inconsistency, low angle resolution and high computational complexity of the traditional phased array system in multi-target capture are solved. The application adopts time delay compensation based on initial angle estimation, signal weighting processing by introducing an adaptive weighting factor, beam forming optimized by the MVDR criterion, and a closed-loop iterative updating mechanism.

[0189] In summary, the technical scheme of the application significantly improves multi-target angle tracking, signal stability and anti-interference ability, solves the problems of far and near target interference, slow filter convergence and the like of the conventional phased array system, and is suitable for the fields of unmanned aerial vehicle communication, radar detection and satellite communication.

[0190] The above describes only the specific implementation of the application, but the protection scope of the application is not limited thereto, and any skilled person in the art can easily think of changes or replacements within the technical range disclosed by the application, which should be covered within the protection scope of the application. Therefore, the protection scope of the application should be subject to the protection scope of the claims.

Claims

1. An adaptive filter based phased array multi-target acquisition method applied to a UAV phased array system, characterized in that, The method comprises the following steps: Step a, before the phased array system starts the transmission of measurement and control data, set the initial parameter combination of the adaptive filter; Step b, obtain the mixed signal vector at the current time through the phased array receiving channel, distinguish the far and near target signals based on the mixed signal vector, and preliminarily set the filter parameters according to the far and near degrees of the targets; Step c, based on the signal model, establish a joint optimization function containing phased array filter parameters, target capture probability and target capture speed; Step d, use the adaptive Bregman ADMM algorithm to dynamically update the joint optimization function, load the finally optimized filter parameters to the phased array system, form a synthetic beam pointing to multiple targets, and perform differential gain allocation for far and near targets to realize synchronous capture of multiple targets.

2. The adaptive filter based phased array multi-target acquisition method of claim 1, wherein, The initial parameter combination of the adaptive filter in step a is: in, This indicates the total number of subarrays in the phased array antenna. For the first The initial weight coefficients of each submatrix, i=1,2,...,N, where the superscript T denotes the transpose of the matrix.

3. The adaptive filter based phased array multi-target acquisition method of claim 2, wherein, In step b, the mixed signal vector at the current time is: wherein, is the total number of targets, is the reflection coefficient of the th target, is the relative distance of the th target, is the path loss exponent, is the steering vector of the mth target, denotes the direction of arrival angle of the mth target, is the adjacent element spacing, represents the total number of subarrays of the phased array antenna, is the signal wavelength, is the complex additive white Gaussian noise.

4. The adaptive filter based phased array multi-target acquisition method of claim 3, wherein, In step b, the far and near target signals are distinguished based on the mixed signal vector, and the filter parameters are preliminarily set according to the far and near degrees of the targets, and the specific manner is: Step b01, generating a pre-beam by a phased array antenna, respectively pointing to different angles , spatially filtering the mixed signal to separate independent signal components of each target , where the superscript H represents conjugate transpose; Step b02, for each individual signal component , calculate the time domain average power : wherein denotes the start time of the current signal processing instant, denotes the observation time window length for power estimation in seconds; Then the target relative distance is calculated : wherein, is the nominal received power at the reference distance; Step b03, according to Computing dynamic distance threshold , the target meeting is classified as a near-range set The rest is classified as a far-range set, generating a classification identification vector , wherein is the classification identification of the th target, if , it indicates that the th target is a near-range target, if , it indicates that the th target is a far-range target; Step b04, differentially assign the initial filter weights according to the classification identifier: wherein are normalization coefficients such that , is the th subarray's steering vector component for the th target direction , whose modulus characterizes the subarray's spatial response strength for that direction signal.

5. The adaptive filter based phased array multi-target acquisition method of claim 1, wherein, The joint optimization function in step c is: s.t. wherein, is the capture probability of the first target, is the capture probability of the nth target, is the variance of the additive white Gaussian noise, representing the noise power of the receiving channel, is the capture time of the slowest target, is the detection distance of the first target, is the detection distance of the nth target, is the motion speed of the first target, is the motion speed of the nth target, is the weighting coefficient, is the minimum overall capture probability threshold required by the system, is the maximum capture time allowed by the task, W is the parameter combination of the adaptive filter, and Z is the auxiliary variable.

6. The adaptive filter based phased array multi-target acquisition method of claim 1, wherein, In step d, the adaptive Bregman ADMM algorithm is used to dynamically update the joint optimization function, and the specific manner is: Step d01, decompose Z into weight modulus constraints , capture probability constraints , capture time constraints three auxiliary variables, decompose constraints on joint optimization function: Step d02, select the KL divergence as the Bregman distance function to match the exponential structure in the capture probability: Set the augmented Lagrangian function: Definitions Lagrange multipliers corresponding to the three constraints for the dual variables; wherein is an adaptive penalty parameter; Step d03, first order Taylor expansion of non-convex terms in the objective function Perform a first order Taylor expansion: where the gradient is calculated from the chain rule The linearized target function is obtained as: Step d04, initialize iteration counter , set maximum number of iterations and convergence threshold , perform alternating direction update: (1) Update master variables : Solve using the complex field conjugate gradient method; (2) update the auxiliary variables : wherein is a projection operator that truncates the matrix elements to lie within the unit circle; Use the LogSumExp function to smooth the maximum value term: wherein , for controlling the degree of smoothing; (3) updating dual variables : (4) Adjusting the adaptive step size according to the original residual error and the dual residual error , according to the rule : wherein, is a tuning factor, is a balancing threshold; (5) Determine whether the following termination conditions are met: Original feasibility: dual feasibility: Relative change in objective function: If all the above termination conditions are met, the iteration is ended, otherwise the alternating direction update is continued.

7. The adaptive filter based phased array multi-target acquisition method of claim 6, wherein, In step d, the finally optimized filter parameters are loaded to the phased array system to form a synthetic beam pointing to multiple targets, and the specific manner is: (1) The optimized filter weight coefficients are assigned to each subarray of the phased array antenna, each subarray containing adjacent elements, and the direction of arrival angle of each subarray is corrected by coordinate transformation ; (2) The adaptive LMS algorithm is used for phase alignment of signals between sub-arrays, eliminating the position deviation of array elements caused by the change of the attitude of the unmanned aerial vehicle, and meeting the coherent superposition condition of the synthesized beam ; wherein, represents the maximum phase difference between adjacent sub-arrays, corresponding to the path difference constraint of 0.125λ. (3) Update the beam pointing parameters according to the dynamic position of the target, and use a three-level beam forming architecture: the first level uses an analog phase shifter to realize ±45° coarse scanning in the subarray, the second level uses a digital phase shifter for ±5° fine adjustment, and the third level uses a digital beam former to complete 0.1° level micro pointing; (4) The beam state feedback information of each subarray is fused by federal filtering to dynamically adjust the 3dB bandwidth of the synthesized beam : wherein, is the total number of all participating elements.

8. The adaptive filter based phased array multi-target acquisition method of claim 7, wherein, In step d, differential gain allocation is performed for far and near targets, and the specific manner is: (1) According to target relative distance Setting a non-linear gain function: at the time of allocation of the attenuation gain; at the time of allocation of the enhanced gain; (2) Calculate the signal-to-noise ratio : Based on signal-to-noise ratio Dynamic adjustment of beam dwell time, when The dwell time is extended to: Where min represents the minimum number; at the same time, the beam width is compressed to: wherein is the current number of active elements; (3) A dual-priority arbitration mechanism is adopted. The first level allocates 60% of the gain resources according to the threat level, and the second level allocates them according to the movement speed. Allocate remaining resources to satisfy ;in To be assigned to the The second-level resource quantity of an entity, This refers to the total amount of remaining resources after the first level of allocation. For the first The speed of movement of each entity For the total number of entities, It is the sum of the velocities of all moving entities; (4) Generate a gain allocation scheme set through Monte Carlo sampling, and select the scheme that meets the following conditions as the optimal scheme for loading and execution: and .

Images