Fusion detection method for intelligent suppression of interference under non-uniform background

Abstract

The application belongs to the technical field of wideband radar signal processing, and particularly relates to a fusion detection method for intelligent interference suppression under a non-uniform background. In view of the problem that an existing wideband radar range expansion target adaptive detector is difficult to consider algorithm calculation complexity, CFAR characteristics, detection performance and mismatch robustness, and considering that actual clutter non-uniformity leads to difficulty in obtaining pure clutter auxiliary data, a reasonable range expansion target adaptive detector form is designed, CFAR characteristics are ensured, calculation complexity of the range expansion target adaptive detection algorithm, intelligent anti-interference, mismatch robustness and detection performance and other aspects are considered, and the adaptive detection performance of a multi-channel wideband radar on a weak small target under a complex environment is improved.

Description

Technical Field

[0001] This invention belongs to the field of broadband radar signal processing technology, specifically relating to a fusion detection method for intelligent interference suppression under non-uniform background. Background Technology

[0002] With the increase in radar bandwidth, broadband radar is gradually covering modern military and civilian fields such as anti-jamming, counter-reconnaissance, precision detection and imaging, high-precision tracking, and target recognition. Adaptive detection of range-extended targets has become a hot topic in the radar field. Unlike narrowband radar, where target echo signals typically occupy only one range resolution cell, broadband radar target energy may diffuse to adjacent range cells, presenting a "one-dimensional range image" and forming range-extended targets. If point target detection methods are used to detect targets on echo signals of a single range cell and background clutter statistical characteristics are estimated using sampling from neighboring range cells, on the one hand, signal pollution in neighboring cells is easily generated due to energy leakage from strong scattering points of range-extended targets, further obscuring the target signal of the individual range cell to be detected, resulting in poor target detection performance. On the other hand, in practical applications, the target's natural environment is complex and variable, and there may be natural or man-made interference sources such as electronic countermeasures or civilian broadcasting systems, which enhances clutter non-uniformity, further making it difficult for existing range-extended target detection methods to achieve ideal detection results.

[0003] Range-extended target adaptive detection primarily relies on auxiliary data. This auxiliary data is typically taken from reference range cells spatially adjacent to the range cell to be detected, and is assumed to contain no target signals, only clutter components that are independent and identically distributed with the main clutter components of the range cell to be detected. Sufficient auxiliary data allows for accurate estimation of the unknown clutter covariance matrix. However, in real-world radar systems, clutter power variations, discrete clutter, and clutter edges are common anomalies, disrupting the uniformity of the clutter background. Obtaining globally uniform auxiliary data is sometimes difficult, severely impacting the performance of range-extended target adaptive detection. In fact, while the global uniformity of complex clutter backgrounds is disrupted, local clutter uniformity still exists within a certain radial range. In such cases, a partially uniform model can be used to model the clutter, where the clutter components in the range cell to be detected and the reference range cell have the same covariance matrix structure but different power levels. This model fully utilizes the local uniformity of the clutter, but the number of reference range cells available is limited by the actual degree of clutter non-uniformity.

[0004] Furthermore, in commonly used rank-one signal target detection models, the target's steering vector is usually assumed to be a known, fixed vector. However, in practical applications, due to beam pointing errors and multipath propagation, the target's steering vector may be mismatched. To address this issue, a subspace model can be used to model the target signal. In the subspace model, the signal is represented as the product of a known subspace matrix and an unknown coordinate matrix. If a subspace model is applied to the target and interference signals based on a holistic dataset consisting of master and auxiliary data from multiple range cells to be detected, and detection statistics are constructed using the GLRT criterion, a subspace GLRT detector (S-GLRT-PHE) for range-extended targets under partially uniform clutter and structured interference can be obtained. This detector achieves good detection performance, but the calculation process is complex and inconvenient to solve. If the Rao detection criterion is used, a subspace Rao detector (S-Rao-PHE) for range-extended targets under partially uniform clutter and structured interference can be obtained. Compared to the GLRT detector, the detection performance of this detector is improved in some scenarios, but the computational complexity of the detection statistics is high, making it inconvenient for engineering implementation.

[0005] For multi-channel broadband radar range-extended target adaptive detection, which faces complex detection environments composed of non-uniform clutter and external structured interference, the key to improving the detection capability of broadband radar in complex interference environments is to rationally design the range-extended target adaptive detector form, effectively suppress interference signals while maintaining constant false alarm rate (CFAR) characteristics, and achieve an effective balance between mismatch robustness, algorithm computational complexity, and detection performance. This is also one of the challenges faced by multi-channel broadband radar range-extended target adaptive detection. Summary of the Invention

[0006] To overcome the problems in the prior art, this invention proposes a fusion detection method for intelligent suppression of interference under non-uniform background.

[0007] The technical solution of the present invention to solve the above-mentioned technical problems is as follows:

[0008] A fusion detection method for intelligent interference suppression under non-uniform background includes the following steps:

[0009] Step 1. Obtain master data Z from K range cells to be detected, and obtain R auxiliary data from R reference range cells adjacent to the range cells to be detected; use the complex Gaussian probability density function of the master data Z without the target assumption to find the partial derivatives with respect to the clutter covariance matrix M and the unknown coordinate matrix Q of the interference subspace, and solve for the maximum likelihood estimates of the clutter covariance matrix M and the unknown coordinate matrix Q of the interference subspace without the target assumption; under the target assumption, solve for the maximum likelihood estimate of the target parameter vector.

[0010] Step 2. Obtain the maximum likelihood estimate of the unknown clutter power factor under the no-target hypothesis by solving the unique positive solution of the eigenvalue equation, and construct the detection statistic λ of the fusion detector for intelligent interference suppression in the non-uniform background;

[0011] Step 3. Set the detection threshold T according to the preset false alarm probability; compare the detection statistic λ with the detection threshold T. If λ≥T, it is determined that there are range extended targets in the current K to-be-detected range cells; otherwise, if λ<T, it is determined that there are no range extended targets in the current K to-be-detected range cells.

[0012] Further, the specific content in the above Step 1 includes:

[0013] Take the partial derivative of the clutter covariance matrix M using the complex Gaussian probability density function of the main data Z under the no-target hypothesis, and solve the maximum likelihood estimate of the clutter covariance matrix M under the no-target hypothesis:

[0014]

[0015] where, (·) H denotes conjugate transpose; the sample covariance matrix S = YY H , the auxiliary data is represented as an N×R-dimensional complex matrix Y = [y1, y2,..., y R , and the N×1-dimensional complex vector y t (t = 1, 2,..., R) at the t-th reference range cell satisfies and it is also independently and identically distributed among different range cells; γ is the proportionality factor of the unknown clutter power between the main data and the auxiliary data; the main data is represented as an N×K-dimensional complex matrix Z = [z1, z2,..., z K , z t = s t + j t + c t (t = 1, 2,..., K), the N×1-dimensional target complex signal vector s t and the N×1-dimensional interference complex vector j t are both assumed to be deterministic, and are respectively represented as s t = Ηp t and j t = Jq t , Η and J are respectively the known full column rank N×p-dimensional target signal subspace complex matrix and N×q-dimensional interference signal subspace complex matrix, the p×1-dimensional complex vector p t and the q×1-dimensional complex vector q tLet H and J represent the unknown complex coordinate vectors of the target signal and the interference signal, respectively. The subspaces H and J are linearly independent. Construct an N×(p+q) dimensional column full-rank augmented matrix B = [HJ], satisfying p+q≤N; the N×1 dimensional clutter vector c in the t-th range cell to be detected. t It is a zero-mean complex circular Gaussian vector, denoted as:

[0016] Furthermore, step 1 specifically includes:

[0017] Using the complex Gaussian probability density function of the master data Z under the no-objective assumption, we solve for the maximum likelihood estimate of the unknown coordinate matrix Q of the disturbance subspace under the no-objective assumption:

[0018]

[0019] in, Let m represent the set of m×n dimensional complex matrices.

[0020] Furthermore, step 1 specifically includes:

[0021] Maximum likelihood estimation of the objective parameter vector under the objective assumption:

[0022]

[0023] in, I N Represents an N×N dimensional identity matrix; the vec function vectorizes the matrix.

[0024] Furthermore, step 2 specifically includes:

[0025] The detection statistic λ of the fusion detector for intelligent suppression of interference in a non-uniform background:

[0026]

[0027] in, The `tr` function represents taking the trace of a square matrix. K Describes a K×K dimensional identity matrix. This represents the maximum likelihood estimate of γ.

[0028] Compared with the prior art, the present invention has the following technical effects:

[0029] 1) A fusion detector for intelligent suppression of interference under non-uniform background was constructed. The detector has a closed-form expression and does not require iterative calculation.

[0030] 2) For interference environments with subspace structure, the intelligent fusion detection method for radar targets under subspace interference of the present invention can effectively suppress interference signals of different intensities and has good intelligent anti-interference ability;

[0031] 3) For the target signal steering vector mismatch, the fusion detection method of intelligent interference suppression under non-uniform background of the present invention can effectively detect the mismatch signal and has strong detection robustness for the mismatch signal;

[0032] 4) The detection method of the present invention maintains the characteristics of CFAR while balancing the performance of algorithm computational complexity, detection performance and mismatch robustness, and improves the adaptive detection performance of multi-channel broadband radar for weak and mismatched targets in complex environments.

[0033] 5) The detector of the present invention can be applied to partially uniform and other non-uniform environments. By fully exploiting the local uniformity of clutter and using the maximum likelihood estimation of clutter power factor under the no-target assumption, the detector's intelligent adaptability to non-uniform clutter environments is improved.

[0034] 6) The method of the present invention is applicable to some non-wideband radar detection situations, such as using low / medium resolution radar to detect large targets or to detect groups of spatially adjacent point targets moving at the same speed (such as ship formations, aircraft formations, vehicle formations, etc.), and has good application prospects. Attached Figure Description

[0035] Figure 1 This is a functional block diagram of the fusion detection method for intelligent suppression of interference under non-uniform background according to the present invention;

[0036] Figure 2 This is a comparison chart of the detection performance of the method of the present invention and existing detection methods on the matched signal;

[0037] in, Figure 2 In the given conditions, N=12, K=15, R=12, p=2, q=2, the false alarm probability P0 fa =10 -3 Interference clutter power ratio (ICR) = 15 dB;

[0038] Figure 3 This is a comparison chart of the detection performance of the method of the present invention and existing detection methods for mismatch signals;

[0039] in, Figure 3 In the given information, N = 12, K = 15, R = 12, p = 2, q = 2, P fa =10 -3 ICR = 15dB, mismatch angle squared value cos 2 φ = 0.5. Detailed Implementation

[0040] The principles and features of the present invention are described below with reference to the accompanying drawings. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention.

[0041] To address the challenges of existing wideband radar range-extended target adaptive detectors (REEDs) in balancing computational complexity, CFAR characteristics, detection performance, and mismatch robustness, and considering the difficulty in obtaining pure clutter auxiliary data due to actual clutter non-uniformity, this paper explores how to design a reasonable REED form that, while ensuring CFAR characteristics, also meets the requirements of computational complexity, intelligent anti-interference, mismatch robustness, and detection performance of the range-extended target adaptive detection algorithm, thereby improving the adaptive detection performance of multi-channel wideband radars for weak targets in complex environments.

[0042] The fusion detection method for intelligent interference suppression under non-uniform background described in this invention includes the following steps:

[0043] Step 1. Obtain master data Z from K range cells to be detected, and obtain R auxiliary data from R reference range cells adjacent to the range cells to be detected; use the complex Gaussian probability density function of the master data Z under the no-target assumption to find the partial derivatives with respect to the clutter covariance matrix M and the unknown coordinate matrix Q of the interference subspace, and solve for the maximum likelihood estimates of the clutter covariance matrix M and the unknown coordinate matrix Q of the interference subspace under the no-target assumption; combine the maximum likelihood estimate of the unknown coordinate matrix P of the target subspace under the target assumption, and construct the intermediate statistics of the subspace Gradient test under the condition that the clutter power factor is known.

[0044] The specific steps include:

[0045] For a coherent radar system with N joint space-time channels, consider the binary hypothesis testing problem of H0 and H1, where under the H0 hypothesis, there is no target and only clutter and interference exist; under the H1 hypothesis, there are targets, clutter and interference.

[0046] Assuming the target may occupy K consecutive range cells to be detected, under assumption H1, the N×1 dimensional received complex signal in the t-th range cell to be detected is represented as z. t =s t +j t +c t (t=1,2,...,K), where the N×1 dimensional target complex signal vector s t and N×1 dimensional interference complex vector j t All are assumed to be deterministic and can be represented as s respectively. t =Hp t and j t =Jq tH and J are the known full-rank N×p-dimensional target signal subspace complex matrix and N×q-dimensional interference signal subspace complex matrix, respectively, and p×1-dimensional complex vector p. t and q×1 dimensional complex vector q t The unknown complex coordinate vectors representing the target signal and the interference signal are respectively represented by the main data, which can be represented as an N×K dimensional complex matrix Z = [z1, z2, ..., z...]. K Furthermore, subspaces H and J are linearly independent. Construct an N×(p+q) dimensional column full-rank augmented matrix B = [HJ], satisfying p+q≤N. The N×1 dimensional clutter vector c in the t-th range cell to be detected... t It is a zero-mean complex circular Gaussian vector, denoted as: Furthermore, the clutter vectors between different distance cells are independent and identically distributed. The N×N clutter covariance matrix M is an unknown positive definite Hermitian complex matrix, and γ is the scaling factor of the unknown clutter power between the main data and the auxiliary data.

[0047] In addition, R observation data y are obtained from R reference range cells that are adjacent to the range cell to be detected. t (t = 1, 2, ..., R), assume y t If (t=1,2,...,R) contains only pure clutter components, then the auxiliary data can be represented as an N×R dimensional complex matrix Y=[y1,y2,...,y R ], where the N×1 dimensional complex vector y of the t-th reference distance unit. t (t=1,2,...,R) satisfies They are also independently and identically distributed across different distance units.

[0048] Under hypotheses H0 and H1, the complex Gaussian joint probability density function (PDF) of the main data Z and the auxiliary data Y can be expressed as follows:

[0049]

[0050]

[0051] Wherein, the sample covariance matrix S = YY H Unknown coordinate matrix of the target subspace Unknown coordinate matrix of the interference subspace Superscript (·) T and(·) H represents the transpose and conjugate transpose, respectively; |·| represents the determinant of the square matrix; and the tr function represents taking the trace of the square matrix.

[0052] A fusion detector for intelligent interference suppression under non-uniform backgrounds is constructed based on the Gradient detection criterion. The Gradient detection statistic for the range-extended target can be expressed as:

[0053]

[0054] in, Target parameter vector It is unknown, the interference parameter vector. It is unknown; This represents the maximum likelihood estimate of Θ under the H0 assumption; Θ r0 Represents Θ r The value under the H0 assumption, Represents Θ r Maximum likelihood estimation under the H1 assumption; the vec function performs matrix vectorization.

[0055] By taking the partial derivative of the joint complex Gaussian probability density function of the main data Z and auxiliary data Y with respect to M under the no-target assumption, i.e., taking the partial derivative of equation (1) with respect to M, and setting the derivative to zero, we can obtain the maximum likelihood estimate of M given Q under the H0 assumption as follows:

[0056]

[0057] Then, substituting equation (4) into equation (1), we can obtain:

[0058]

[0059] Among them, I m Let Q represent an m×m dimensional identity matrix. Using (5) to take the partial derivative with respect to Q and setting the result to zero, we can obtain the maximum likelihood estimate of Q under the H0 assumption as follows:

[0060]

[0061] in,

[0062] Then, by taking the partial derivative of the joint complex Gaussian probability density function of the main data Z and auxiliary data Y with respect to M under the objective assumption, i.e., taking the partial derivative of equation (2) with respect to M, and setting the derivative result to zero, we can obtain the maximum likelihood estimate of M given D under the H1 assumption as follows:

[0063]

[0064] Substituting equation (7) into equation (2), we can obtain:

[0065]

[0066] By taking the partial derivative of (8) with respect to D and setting the result to zero, we can obtain the maximum likelihood estimate of D under the H1 assumption as follows:

[0067]

[0068] in, Note the maximum likelihood estimate of the target coordinate matrix P under the H1 assumption (denoted as ). )yes From the first p columns, we can obtain the maximum likelihood estimate of the objective parameter vector under the H1 assumption. for:

[0069]

[0070] in,

[0071] Furthermore, by taking the partial derivative of the complex Gaussian joint probability density function of the main data Z and auxiliary data Y with respect to the objective parameter vector under the objective assumption, we can obtain:

[0072]

[0073] in, and Let M, P, and Q represent the maximum likelihood estimates of the clutter covariance matrix, target coordinate matrix, and interference coordinate matrix respectively, without the target assumption; it is obvious that... According to equation (4), and using the matrix inversion lemma, we can obtain:

[0074]

[0075] Substituting equations (6) and (12) into equation (11), we get:

[0076]

[0077] Note that the target does not exist under the H0 assumption, therefore Θ r0 =0. Substituting equations (10) and (13) into equation (3), and performing multiple algebraic operations, we can obtain the intermediate statistic of the subspace Gradient test under non-uniform clutter and structured interference:

[0078]

[0079] in,

[0080] Step 2. By solving the unique positive solution of the eigenvalue equation under the no-target assumption, the maximum likelihood estimate of the unknown clutter power factor under the no-target assumption is obtained. The maximum likelihood estimate of the unknown clutter power factor replaces the unknown clutter power factor of the subspace Gradient test intermediate statistic obtained in Step 1, and the detection statistic λ of the fusion detector for intelligent suppression of interference in non-uniform background is constructed.

[0081] The specific steps include:

[0082] By constructing the eigenvalue equation (15) and solving the unique positive solution of equation (15), the maximum likelihood estimate of the unknown clutter power factor γ under the no-target assumption can be obtained.

[0083] Maximum likelihood estimation of γ under assumption H0 It is the only positive solution that satisfies the eigenvalue equation of equation (15):

[0084]

[0085] Where υ represents the unknown parameter, λ k,0 Representation matrix The k-th non-zero eigenvalue.

[0086] Maximum likelihood estimation of unknown clutter power factor under assumption H0 Substituting into equation (14), replacing the unknown clutter power factor under the no-target assumption, i.e., using By substituting γ in equation (14) and simplifying the calculation, the detection statistic λ of the fusion detector for intelligent suppression of interference under non-uniform background can be expressed as:

[0087]

[0088] The present invention constructs a fusion detector for intelligent interference suppression under non-uniform backgrounds. As can be seen from Equation (16), the proposed fusion detector for intelligent interference suppression under non-uniform backgrounds has a closed-form expression for the detection statistics, requiring no iterative computation. Furthermore, it is worth noting that compared to the S-GLRT-PHE detector for range-extended targets, the proposed fusion detector for intelligent interference suppression under non-uniform backgrounds has lower computational complexity and stronger robustness against steering vector mismatch signals. In summary, the fusion detector for intelligent interference suppression under non-uniform backgrounds of the present invention maintains CFAR characteristics while effectively balancing computational complexity, mismatch robustness, and detection performance.

[0089] Step 3. To maintain the CFAR characteristic of the detection method, set the detection threshold T according to the preset false alarm probability; compare the detection statistic λ with the detection threshold T. If λ≥T, it is determined that there are range extension targets in the current K to-be-detected range cells; otherwise, if λ<T, it is determined that there are no range extension targets in the current K to-be-detected range cells.

[0090] To verify the effectiveness of the method described in this invention, two embodiments are given in this specific implementation manner. The first embodiment is for the sea detection environment, and the second embodiment is for the land detection environment.

[0091] Embodiment 1:

[0092] Refer to the attached Figure 1 of the specification. The specific implementation manner of Embodiment 1 is divided into the following steps:

[0093] Step A1. Use a sea detection radar to irradiate the to-be-detected sea area to obtain the main data Z of K to-be-detected range cells; irradiate the target-free area around the to-be-detected sea area to obtain the auxiliary data Y containing only pure sea clutter of R reference range cells. Send the main data Z and the auxiliary data Y to the maximum likelihood estimation solving module under the H0 hypothesis, the derivative module of the probability density function under the H1 hypothesis, and the maximum likelihood estimation solving module under the H1 hypothesis; in the maximum likelihood estimation solving module under the H0 hypothesis, obtain the maximum likelihood estimations of M, Q, and γ under the H0 hypothesis according to Equations (4), (6), and (15) respectively and In the derivative module of the probability density function under the H1 hypothesis, obtain the derivative result of the complex Gaussian joint probability density function of the main data Z and the auxiliary data Y with respect to the target parameter vector Θ r under the H1 hypothesis according to Equation (11); in the maximum likelihood estimation solving module under the H1 hypothesis, obtain the maximum likelihood estimation of Θ r under the H1 hypothesis according to Equation (10)

[0094] It is worth noting that in step A1, the subspace range-extended target signal model constructed by the method of the present invention can effectively address the difficulty of handling target guidance vector mismatch in rank-one signal models, thus improving the robustness of broadband radar to target guidance vector mismatch at sea. Simultaneously, considering that external interference in the actual marine environment may adversely affect the adaptive detection of range-extended targets, external interference is also considered in the detector design process, and subspace signals are used to model the interference to reduce the potential mismatch impact of interference signals. For interference environments with subspace structure, the range-extended target Gradient intelligent fusion detection method of the present invention can effectively suppress interference signals of different intensities, exhibiting good intelligent anti-interference capabilities. Furthermore, the detector of the method of the present invention is applicable to some uniform and non-uniform sea clutter environments. By fully exploiting the local uniformity of sea clutter and utilizing the maximum likelihood estimation of the clutter power factor under the no-target assumption, the intelligent adaptability of the detector to non-uniform sea clutter environments is improved.

[0095] Step A2. Send the results obtained from the maximum likelihood estimation solution module under the H0 assumption, the probability density function derivative module under the H1 assumption, and the maximum likelihood estimation solution module under the H1 assumption to the distance extended target Gradient detection statistic construction module. Construct the detection statistic λ of the fusion detection method for intelligent suppression of interference in non-uniform background according to Equation (14), and send λ to the detection decision module.

[0096] It is worth noting that in step A2, compared with the S-GLRT-PHE detector for range-extended targets, the method of this invention has lower computational complexity and stronger robustness against steering vector mismatch signals. Furthermore, the constructed fusion detection method for intelligent interference suppression under non-uniform backgrounds has a closed-form expression. Compared with existing adaptive detection methods for range-extended targets, it maintains CFAR characteristics while balancing computational complexity, detection performance, and mismatch robustness, thus improving the adaptive detection capability of multi-channel broadband radar for weak and mismatched targets on the sea surface in complex electromagnetic environments.

[0097] Step A3. Set the detection threshold T according to the preset false alarm probability: Specifically, set the false alarm probability to P. fa According to the Monte Carlo method, based on the previously accumulated 100 / P fa The detection threshold T is calculated based on measured sea clutter data. Considering the difficulty in obtaining sea clutter data, if the actual amount of pure sea clutter measured data R is less than 100 / P... fa Then the missing 100 / P fa-R clutter data can be obtained by simulating with a sea clutter simulation model, and the model parameters are reasonably estimated and set according to the measured data of pure sea clutter that has been obtained. Further, the detection statistic λ is compared with the detection threshold T. If λ≥T, it is determined that there is a range extended target in the current K range cells to be detected, and the main data is not used as the auxiliary data for other subsequent range cells to be detected; conversely, if λ<T, it is determined that there is no range extended target in the current K range cells to be detected, and the main data is used as the auxiliary data for other subsequent range cells to be detected.

[0098] The comparison results of the detector performance under the target-oriented vector matching environment are shown in the appendix Figure 2 . The results show that, compared with the existing detectors for range extended targets such as S-GLRT-PHE and S-Rao-PHE, the detector of the method of the present invention has better detection performance in the matching environment.

[0099] Embodiment 2:

[0100] Refer to the appendix of the specification Figure 1 , and the specific implementation of Embodiment 2 is divided into the following steps:

[0101] Step B1. Use a ground detection radar to irradiate the area to be detected to obtain the main data Z of K range cells to be detected; irradiate the target-free area around the area to be detected to obtain the auxiliary data Y containing only pure ground clutter of R reference range cells. Send the main data Z and the auxiliary data Y to the maximum likelihood estimation solving module under the H0 hypothesis, the derivative module of the probability density function under the H1 hypothesis, and the maximum likelihood estimation solving module under the H1 hypothesis; in the maximum likelihood estimation solving module under the H0 hypothesis, obtain the maximum likelihood estimates of M, Q, and γ under the H0 hypothesis according to Equations (4), (6), and (15) respectively and In the derivative module of the probability density function under the H1 hypothesis, obtain the derivative result of the complex Gaussian joint probability density function of the main data Z and the auxiliary data Y with respect to the target parameter vector Θ r under the H1 hypothesis according to Equation (11); in the maximum likelihood estimation solving module under the H1 hypothesis, obtain the maximum likelihood estimate of Θ r under the H1 hypothesis according to Equation (10)

[0102] It should be noted that in step B1, the subspace distance extended target signal model constructed by the method of the present invention can effectively address the problem that the rank-one signal model is difficult to handle the mismatch of the target steering vector, improving the robustness of the broadband radar to the mismatch of the ground target steering vector. At the same time, considering that in the actual ground environment, external interference may have an adverse impact on the adaptive detection of the distance extended target, external interference is also considered in the detector design process, and the subspace signal is used to model the interference to reduce the possible mismatch impact of the interference signal. For the interference environment with subspace structure, the fusion detection method for intelligent interference suppression under non-uniform background of the present invention can effectively suppress interference signals with different intensities, having good intelligent anti-interference performance. At the same time, the detector of the method of the present invention can be applied to non-uniform clutter environments such as partially uniform clutter. By fully exploiting the local uniformity of the ground clutter and using the maximum likelihood estimation of the clutter power factor under the no-target hypothesis, the intelligent adaptation ability of the detector to the non-uniform ground clutter environment is improved.

[0103] Step B2. Send the results obtained from the maximum likelihood estimation solving module under the H0 hypothesis, the derivative module of the probability density function under the H1 hypothesis, and the maximum likelihood estimation solving module under the H1 hypothesis to the distance extended target Gradient detection statistic construction module. According to Equation (14), construct the detection statistic λ of the fusion detection method for intelligent interference suppression under non-uniform background, and send λ to the detection decision module.

[0104] It should be noted that in step B2, compared with the S-GLRT-PHE detector of the distance extended target, the algorithm of the method of the present invention has lower computational complexity and stronger detection robustness to the steering vector mismatch signal. In addition, the constructed fusion detection method for intelligent interference suppression under non-uniform background has a closed-form expression. Compared with the existing adaptive detection methods for distance extended targets, while maintaining the CFAR characteristics, it takes into account the performance balance of algorithm computational complexity, detection performance, and mismatch robustness, improving the adaptive detection ability of multi-channel broadband radar to weak ground targets and mismatch targets in complex electromagnetic environments.

[0105] Step B3. Set the detection threshold T according to the preset false alarm probability: Specifically, set the false alarm probability as P fa , and according to the Monte Carlo method, calculate the detection threshold T based on 100 / P fa measured ground clutter data accumulated previously. Further, compare the detection statistic λ with the detection threshold T. If λ≥T, it is determined that there is a distance extended target in the current K to-be-detected range cells, and the main data is not used as the auxiliary data for other subsequent to-be-detected range cells; otherwise, if λ<T, it is determined that there is no distance extended target in the current K to-be-detected range cells, and the main data is used as the auxiliary data for other subsequent to-be-detected range cells.

[0106] The results of the detector performance comparison under target-guided vector mismatch conditions are attached. Figure 3 The results show that, compared with existing range-extended target detectors such as S-GLRT-PHE and S-Rao-PHE, the detector of the present invention has better detection robustness in mismatch environments.

[0107] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A fusion detection method for intelligent suppression of interference under non-uniform background, characterized in that, Includes the following steps: Step 1. From K Each distance unit to be detected acquires master data. Z From the distance unit that is close to the one to be detected R Each reference range cell acquires Y auxiliary data points containing only pure sea clutter; Under the no-target assumption H0, the target does not exist; only clutter and interference exist. Under the target-oriented assumption H1, the target, clutter, and interference all exist. Using master data without the objective assumption Z Complex Gaussian joint with auxiliary data Y probability density function versus clutter covariance matrix Taking the partial derivative and setting the result to zero yields the unknown coordinate matrix of the given disturbance subspace under the objective assumption. Q Time-clutter covariance matrix M Maximum likelihood estimation; clutter covariance matrix M Maximum likelihood estimation is substituted into the master data under the no-objective assumption. Z The maximum likelihood estimate of Q under the no-target assumption is obtained by taking the partial derivative of the joint probability density function of the complex Gaussian matrix with respect to the unknown coordinate matrix Q of the interference subspace and setting the derivative to zero. Under the objective assumption, using master data under the objective assumption The complex Gaussian joint probability density function of the auxiliary data Y versus the clutter covariance matrix Taking the partial derivative and setting the result to zero, we obtain the clutter covariance matrix given the unknown coordinate matrix D of the clutter subspace under the objective assumption. M Maximum likelihood estimation; given the unknown coordinate matrix D of the disturbance subspace under the objective assumption, the clutter covariance matrix... M Substituting maximum likelihood estimation into the master data under the objective assumption Z In the complex Gaussian joint probability density function of the auxiliary data Y, the partial derivative with respect to the unknown coordinate matrix D of the interference subspace is obtained, and the derivative is set to zero to obtain the maximum likelihood estimate of the unknown coordinate matrix D of the interference subspace under the objective assumption. By taking the partial derivative of the joint complex Gaussian probability density function of the master and auxiliary data with respect to the target parameter vector under the assumption of a target, and based on the maximum likelihood estimation of the target parameter vector, the intermediate statistic of the subspace Gradient test under non-uniform clutter and structured interference is calculated. Step 2. Construct the eigenvalue equation under the no-target assumption. By solving the unique positive solution of the eigenvalue equation under the no-target assumption, the maximum likelihood estimate of the unknown clutter power factor under the no-target assumption is obtained, and the detection statistic of the fusion detector for intelligent interference suppression in non-uniform background is constructed. ; Step 3. Set the detection threshold according to the preset false alarm probability. T ; Detection statistics With detection threshold T If a comparison is made, Then determine the current K Each distance cell to be detected contains a distance extension target; conversely, if... Then determine the current K The distance cell to be detected does not have a distance extension target.

2. The fusion detection method for intelligent suppression of interference under non-uniform background according to claim 1, characterized in that, Step 1 specifically includes: Using master data under the no-target assumption Z The complex Gaussian probability density function with respect to the clutter covariance matrix Find the partial derivatives and solve for the clutter covariance matrix. M Maximum likelihood estimation without objective assumptions: in, Represents the conjugate transpose; sample covariance matrix S = YY H Auxiliary data is represented as 3D complex matrix In the formula, the first t Reference distance cells 3D complex vector satisfy It is also independently and identically distributed across different distance cells; γ is the scaling factor for the unknown clutter power between the main data and auxiliary data; the main data is represented as... 3D complex matrix , , 3D target complex signal vector and 3D interference complex vector All are assumed to be deterministic, and are respectively expressed as and , and Each column has a known full rank. The complex matrix of the target signal subspace and the sum of the dimensions 3D interference signal subspace complex matrix, 3D complex vector p t and 3D complex vector q t Let the unknown complex coordinate vectors of the target signal and the interference signal be represented respectively, and the subspace be represented by the subspace. and They are linearly independent, construct dimensional full-rank augmented matrix And satisfy ;No. t In each distance unit to be detected dimensional clutter vector It is a zero-mean complex circular Gaussian vector, denoted as: , N represents the number of space-time joint channels.

3. The fusion detection method for intelligent suppression of interference under non-uniform background according to claim 2, characterized in that, Step 1 specifically also includes: Using master data under the no-target assumption Z The complex Gaussian probability density function for the unknown coordinate matrix of the disturbance subspace Q， Solve for the unknown coordinate matrix of the disturbance subspace Q Maximum likelihood estimation without objective assumptions: in, express m × n A set of complex matrices of dimension 1.

4. The fusion detection method for intelligent suppression of interference under non-uniform background according to claim 3, characterized in that, Step 1 specifically also includes: Maximum likelihood estimation of the objective parameter vector under the objective assumption: in, , , express A 3D identity matrix; the `vec` function vectorizes the matrix.

5. The fusion detection method for intelligent suppression of interference under non-uniform background according to claim 4, characterized in that, Step 2 specifically includes: Detection statistics of a fusion detector with intelligent interference suppression against a non-uniform background : in, The `tr` function represents taking the trace of a square matrix. express An identity matrix of dimension 1 express Maximum likelihood estimation.

Images