Extended target detection method and system under interference direction vector uncertainty

Abstract

The application relates to the technical field of radar signal processing, and provides an extended target detection method and system under uncertain interference direction vectors. N K A matrix is modeled into a rank-one interference and noise, thereby adapting to a multi-channel extended target matrix form echo.The rank-one interference is constructed into a combined form composed of a coordinate vector and an amplitude vector in a known subspace, so that performance loss caused by mismatch of a multi-rank interference model is avoided. Parameter estimation is respectively carried out under a zero hypothesis composed of only the rank-one interference and noise clutter and an alternative hypothesis containing target signals, algorithm performance is optimized under the condition that all parameters are completely unknown and there is no theoretical optimal detector, and the adaptability of a detector to uncertain working conditions of a flickering interference direction vector is improved. A test statistic is constructed by combining parameters estimated under the zero hypothesis and the alternative hypothesis, a detection threshold is determined in combination with a system false alarm probability, and detection capability and reliability are effectively improved.

Description

Technical Field

[0001] This application relates to the field of radar signal processing technology, and in particular to an extended target detection method and system under interference steering vector uncertainty. Background Technology

[0002] As the number of radar channels and signal bandwidth gradually increase, the echoes of large detection targets such as transport aircraft cover multiple range cells, exhibiting extended target characteristics. Under multi-channel radar, the echoes of such targets can be modeled as subspace matrix signals. In complex electromagnetic environments, radar echoes are susceptible to man-made electromagnetic interference in addition to clutter and thermal noise. Among them, scintillation interference, which is commonly used in actual electromagnetic countermeasures environments, can be modeled as rank-one interference with subspace constraints within the coherent processing interval. That is, the interference as a whole is confined to a known subspace, with only the specific coordinates being unknown. At the same time, the interference steering vector is uncertain due to factors such as array calibration deviation.

[0003] Currently, most detectors in the industry are based on GLRT, Rao, and Wald detection criteria, relying on a single-hypothesis architecture to solve for unknown parameters and construct detection statistics using only the null hypothesis that the target does not exist. This type of approach is only suitable for simple scenarios where the interference state is stable, and its practical application has significant drawbacks. Specifically, on the one hand, the distribution of rank interference parameters differs depending on whether the target is present or not, and the single-hypothesis model cannot adapt to the parameter shift problem, resulting in incomplete interference removal and a high risk of false alarms. On the other hand, this detector has low tolerance for array errors, sample fluctuations, and interference steering vector mismatch, and its detection performance is easily degraded in complex adversarial environments. Furthermore, the single-hypothesis architecture makes it difficult to explore the characteristic differences between signals and interference under different operating conditions, resulting in poor environmental adaptability. Summary of the Invention

[0004] Therefore, it is necessary to provide an extended target detection method and system under the uncertainty of interference steering vector to address the above-mentioned technical problems.

[0005] An extended target detection method under uncertain interference steering vector, the method comprising: Get The dimension of the data matrix to be detected and A binary hypothesis testing model is constructed using a training sample data vector. In the binary hypothesis testing model, the null hypothesis is composed of rank-one interference and noise clutter. Based on the data matrix to be detected and the training sample data vector, estimate the parameters under the null hypothesis, including the interference coordinate vector, the interference amplitude vector, and the clutter covariance matrix; Based on the training sample data vector, the known interference matrix, and the known signal matrix, estimate the parameters under the alternative hypothesis, including the interference coordinate vector and the interference amplitude vector; The test statistic is constructed based on the parameters estimated under the null and alternative hypotheses, and the detection threshold is determined by combining the false alarm probability of the system. Compare the test statistic with the detection threshold. If the test statistic is greater than the detection threshold, the target is determined to exist; otherwise, the target does not exist.

[0006] In one embodiment, the binary hypothesis testing model includes a null hypothesis and alternative hypotheses, the null hypothesis being... for: ; Alternative Hypothesis for: ; in, for A 3D matrix of data to be detected; This is a noise component; for The interference matrix is ​​known in dimension 1; for 3D unknown interference coordinate vector; for An unknown disturbance amplitude vector; for dimensional training sample data vector; for A known signal matrix; for An unknown signal coordinate matrix; The number of training sample data vectors; Number of channels; Expand the dimension of the target; The training sample vector is pure noise.

[0007] In one embodiment, parameters under the null hypothesis are estimated based on the data matrix to be detected and the training sample data vector. These parameters include the interference coordinate vector, the interference amplitude vector, and the clutter covariance matrix, including: Based on the data matrix to be detected and the training sample data vector, estimate the parameters under the null hypothesis and estimate the interference coordinate vector. For matrix The normalized eigenvector corresponding to the largest eigenvalue; Estimated disturbance amplitude vector for: ; ; Estimated clutter covariance matrix for: ; in, The covariance matrix is ​​sampled for the training samples.

[0008] In one embodiment, parameters under the alternative hypothesis are estimated based on the training sample data vector, the known interference matrix, and the known signal matrix, including the interference coordinate vector and the interference amplitude vector, including: Construct the intermediate variable matrix based on the training sample data vector and the known signal matrix: ; in, This is an intermediate variable matrix; Construct a matrix based on the intermediate variable matrix and the known interference matrix. : ; in, for An identity matrix of order 1; With matrix The normalized eigenvector corresponding to the largest eigenvalue is the estimated value of the disturbance coordinate vector under the alternative hypothesis. ; Estimated value of disturbance amplitude vector under alternative assumptions for: .

[0009] In one embodiment, the test statistic is: ; in, It is the absolute value; Let be the trace of the matrix.

[0010] In one embodiment, a test statistic is constructed based on the estimated parameters, and a detection threshold is determined in conjunction with the system false alarm probability, including: Construct a test statistic based on the estimated parameters; The detection threshold is determined using standard Monte Carlo simulation based on the test statistic and the system false alarm probability. ; in, For detection threshold; This represents the probability of a false alarm in the system. To be The detection statistics obtained under the null hypothesis, arranged in descending order, are as follows: One value, Number of Monte Carlo simulations; This is a floor operation.

[0011] An extended target detection system under uncertain interference steering vectors, the system comprising: The model building module is used to obtain the data matrix to be detected and the training sample data vector, and to build a binary hypothesis testing model. In the binary hypothesis testing model, the null hypothesis is composed of rank-one interference and noise clutter. The null hypothesis parameter estimation module is used to estimate the parameters under the null hypothesis based on the data matrix to be detected and the training sample data vector, including the interference coordinate vector, the interference amplitude vector and the clutter covariance matrix. The alternative hypothesis parameter estimation module is used to estimate the parameters under the alternative hypothesis, including the interference coordinate vector and the interference amplitude vector, based on the training sample data vector, the known interference matrix, and the known signal matrix. The detection setting module is used to construct a test statistic based on the parameters estimated under the null and alternative hypotheses, and to determine the detection threshold in combination with the system false alarm probability. The detection execution module compares the test statistic with the detection threshold. If the test statistic is greater than the detection threshold, the target is determined to exist; otherwise, the target does not exist.

[0012] The aforementioned extended target detection method and system under uncertain interference steering vectors model the echo data containing extended targets, rank-one interference, and noise as follows: A multidimensional matrix is ​​used to adapt to multi-channel extended target matrix echoes, fully preserving the guidance subspace information of each range cell of the target. By constructing rank-one interference as a combination of coordinate vectors and amplitude vectors within a known subspace, the uncertainty of interference guidance vectors in radar countermeasures environments is accurately characterized, avoiding performance loss caused by mismatch in multi-rank interference models. Parameter estimation is performed under the null hypothesis consisting only of rank-one interference and noise clutter, and under the alternative hypothesis including the target signal. Under the given condition that multiple parameters are completely unknown and there is no theoretically optimal detector, the algorithm performance is optimized, improving the detector's adaptability to uncertain conditions of scintillation interference guidance vectors. By combining the parameters estimated under the null and alternative hypotheses to construct test statistics, the model perturbation caused by unknown parameters is weakened based on the measured parameter estimation results; and the detection threshold is determined by combining the system false alarm probability, fully exploring the detection potential while ensuring that false alarms are controllable, effectively improving detection capability and reliability. Attached Figure Description

[0013] Figure 1 This is a flowchart illustrating an extended target detection method under uncertain interference steering vector in one embodiment; Figure 2 This is a detection probability map under different signal-to-noise ratios (SNR) in one embodiment, wherein, Figure 2 (a) is a comparison chart of detection probabilities at a signal-to-noise ratio of 1. Figure 2(b) is a comparison chart of detection probabilities at a signal-to-noise ratio of 2; Figure 3 This is a block diagram of an extended target detection system under uncertain interference steering vector conditions in one embodiment. Detailed Implementation

[0014] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0015] In one embodiment, such as Figure 1 As shown, an extended target detection method under uncertain interference steering vectors is provided, the method comprising: Step 201, obtain The dimension of the data matrix to be detected and A binary hypothesis testing model is constructed using a training sample data vector. In the binary hypothesis testing model, the null hypothesis consists of rank-one interference and noise clutter.

[0016] It should be noted that the data to be detected is the steering vector of rank-one interference.

[0017] Step 202: Based on the data matrix to be detected and the training sample data vector, estimate the parameters under the null hypothesis, including the interference coordinate vector, the interference amplitude vector, and the clutter covariance matrix.

[0018] Step 203: Estimate the parameters under the alternative hypothesis, including the interference coordinate vector and the interference amplitude vector, based on the training sample data vector, the known interference matrix, and the known signal matrix.

[0019] Step 204: Construct a test statistic based on the parameters estimated under the null and alternative hypotheses, and determine the detection threshold by combining it with the system false alarm probability.

[0020] Step 205: Compare the test statistic with the detection threshold. If the test statistic is greater than the detection threshold, the target is determined to exist; otherwise, the target does not exist.

[0021] In the aforementioned extended target detection method under uncertain interference steering vectors, the echo data containing the extended target, rank-one interference, and noise is modeled as... A multidimensional matrix is ​​used to adapt to multi-channel extended target matrix echoes, fully preserving the guidance subspace information of each range cell of the target. By constructing rank-one interference as a combination of coordinate vectors and amplitude vectors within a known subspace, the uncertainty of interference guidance vectors in radar countermeasures environments is accurately characterized, avoiding performance loss caused by mismatch in multi-rank interference models. Parameter estimation is performed under the null hypothesis consisting only of rank-one interference and noise clutter, and under the alternative hypothesis including the target signal. Under the given condition that multiple parameters are completely unknown and there is no theoretically optimal detector, the algorithm performance is optimized, improving the detector's adaptability to uncertain conditions of scintillation interference guidance vectors. By combining the parameters estimated under the null and alternative hypotheses to construct test statistics, the model perturbation caused by unknown parameters is weakened based on the measured parameter estimation results; and the detection threshold is determined by combining the system false alarm probability, fully exploring the detection potential while ensuring that false alarms are controllable, effectively improving detection capability and reliability.

[0022] In one embodiment, the binary hypothesis testing model includes a null hypothesis and alternative hypotheses, the null hypothesis being... for: ; Alternative Hypothesis for: ; in, for A 3D matrix of data to be detected; This is a noise component; for The interference matrix is ​​known in dimension 1; for 3D unknown interference coordinate vector; for An unknown disturbance amplitude vector; for dimensional training sample data vector; for A known signal matrix; for An unknown signal coordinate matrix; The number of training sample data vectors; Number of channels; Expand the dimension of the target; The training sample vector is pure noise.

[0023] It should be noted that the noise component Each column (containing clutter and thermal noise) is independently and identically distributed, following a mean of zero and a covariance matrix of... The complex Gaussian distribution; pure noise training sample vector and noise components Each column is independent and identically distributed. (Superscript) This indicates the conjugate transpose.

[0024] Understandable. Represents rank-one interference, noise component It can represent noise and clutter. By employing The data matrix to be detected is constructed in dimensional matrix form to adapt to the extended target across resolution units in the distance dimension; , The rank-one interference component is uniformly set in the middle. Using a known matrix To achieve interference subspace constraints, and at the same time , By setting it as an unknown vector, the inherent uncertainty of the interference steering vector can be accurately characterized. Furthermore, the binary hypothesis testing model constructed in this invention strictly distinguishes between "interference only + clutter" (…). "Target + Interference + Clutter" The two types of real electromagnetic conditions enable the parameters under the two types of assumptions to have different characteristics, providing model support for the subsequent estimation of the interference parameters of the null hypothesis and the alternative hypothesis.

[0025] In this embodiment, the data to be detected by rank-one interference is modeled as an unknown vector in a known subspace. The interference coordinate vector and interference amplitude vector are set as unknown variables and are not pre-assigned, thus avoiding the uncertainty of the steering vector caused by array calibration error and interference identification error.

[0026] In one embodiment, parameters under the null hypothesis are estimated based on the data matrix to be detected and the training sample data vector. These parameters include the interference coordinate vector, the interference amplitude vector, and the clutter covariance matrix, including: Based on the data matrix to be detected and the training sample data vector, estimate the parameters under the null hypothesis and estimate the interference coordinate vector. For matrix The normalized eigenvector corresponding to the largest eigenvalue; Estimated disturbance amplitude vector for: ; ; Estimated clutter covariance matrix for: ; in, The covariance matrix is ​​sampled for the training samples.

[0027] In one embodiment, parameters under the alternative hypothesis are estimated based on the training sample data vector, the known interference matrix, and the known signal matrix, including the interference coordinate vector and the interference amplitude vector, including: Construct the intermediate variable matrix based on the training sample data vector and the known signal matrix: ; in, This is an intermediate variable matrix; Construct a matrix based on the intermediate variable matrix and the known interference matrix. : ; in, for An identity matrix of order 1; With matrix The normalized eigenvector corresponding to the largest eigenvalue is the estimated value of the disturbance coordinate vector under the alternative hypothesis. ; Estimated value of disturbance amplitude vector under alternative assumptions for: .

[0028] In one embodiment, the test statistic is: ; in, It is the absolute value; Let be the trace of the matrix.

[0029] In this embodiment, the residual echo matrix after removing flicker interference is used. and A test statistic is constructed to eliminate the interference of flickering interference on detection at the source. Simultaneously, the extended target energy is accurately extracted by utilizing the target subspace coupling characteristics, adapting to the modeling features of the extended target matrix.

[0030] In one embodiment, a test statistic is constructed based on the estimated parameters, and a detection threshold is determined in conjunction with the system false alarm probability, including: Construct a test statistic based on the estimated parameters; The detection threshold is determined using standard Monte Carlo simulation based on the test statistic and the system false alarm probability. ; in, For detection threshold; This represents the probability of a false alarm in the system. To be The detection statistics obtained under the null hypothesis, arranged in descending order, are as follows: One value, Number of Monte Carlo simulations; This is a floor operation.

[0031] Specifically It is a sequence Arranged from largest to smallest There are 1 maximum value; among them, Represents the data matrix to be detected that does not contain signal components. The This experiment, Indicates the null hypothesis Lower interference coordinate vector The estimated value The This experiment, Indicates the null hypothesis Lower interference amplitude vector The estimated value The This experiment, Indicates the null hypothesis Lower clutter covariance matrix The estimated value The This experiment, Represents the sampling covariance matrix The This experiment, Indicate alternative hypothesis Lower interference coordinate vector The estimated value The This experiment, Indicate alternative hypothesis Lower interference amplitude vector The estimated value The This experiment.

[0032] In this embodiment, the detection threshold is determined by standard Monte Carlo simulation and the system false alarm probability, which fully taps the detection potential while ensuring that false alarms are controllable, thereby effectively improving detection capability and reliability.

[0033] Understandably, the greatest complexity of this invention stems from matrix inversion. Compared with existing methods, the computational complexity of this invention is comparable, making it easy to implement in engineering.

[0034] To verify the effectiveness of this invention, Monte Carlo simulations were used to compare the detection probabilities of this invention with those of a GLRT detector, thereby verifying the performance of this invention. In the simulation, the clutter covariance matrix adopted an exponential correlation model, where the... The elements are , ,symbol Represents absolute value. Signal matrix. With interference matrix Once randomly generated, the signal-to-noise ratio remains constant. The signal-to-noise ratio is defined as... The noise-to-interference ratio is defined as follows: The simulation parameters are set as follows: INR = 10 dB, false alarm probability In a signal-to-noise ratio of 1, , , , , ( Figure 2 (a)); In signal-to-noise ratio 2, , , , , ( Figure 2 (b)). Figure 2 The results show that from Figure 2 (a) It can be seen that, under the condition of a signal-to-noise ratio of 1, the detection probability of the present invention is superior to that of existing GLRT detectors across all SNR ranges; Figure 2 (b) It can be seen that when SNR>20 dB, the detection performance of the present invention is also superior to that of existing detectors. It should be noted that for early warning radar, the focus is usually on the region with a detection probability greater than 0.5; while for tracking radar, a detection probability greater than 0.8 or even higher is required. The present invention shows significant advantages in all of the above ranges.

[0035] It should be understood that, although Figure 1 The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figure 1 At least some of the steps in the process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.

[0036] In one embodiment, such as Figure 3 As shown, an extended target detection system under uncertain interference steering vector conditions is provided. The system includes: The model building module 901 is used to obtain the data matrix to be detected and the training sample data vector, and to build a binary hypothesis testing model. In the binary hypothesis testing model, the null hypothesis is composed of rank-one interference and noise clutter.

[0037] The null hypothesis parameter estimation module 902 is used to estimate the parameters under the null hypothesis based on the data matrix to be detected and the training sample data vector, including the interference coordinate vector, the interference amplitude vector and the clutter covariance matrix.

[0038] The alternative hypothesis parameter estimation module 903 is used to estimate the parameters under the alternative hypothesis, including the interference coordinate vector and the interference amplitude vector, based on the training sample data vector, the known interference matrix, and the known signal matrix.

[0039] The detection setting module 904 is used to construct a test statistic based on the parameters estimated under the null hypothesis and alternative hypothesis, and to determine the detection threshold in combination with the system false alarm probability.

[0040] The detection execution module 905 is used to compare the test statistic with the detection threshold. If the test statistic is greater than the detection threshold, the target is determined to exist; otherwise, the target does not exist.

[0041] Specific limitations regarding the extended target detection system under uncertain interference steering vectors can be found in the limitations of the extended target detection method under uncertain interference steering vectors described above, and will not be repeated here. Each module in the aforementioned extended target detection system under uncertain interference steering vectors can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0042] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0043] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for extended target detection under uncertain interference steering vectors, characterized in that, The method includes: Get The dimension of the data matrix to be detected and A binary hypothesis testing model is constructed using a training sample data vector. In the binary hypothesis testing model, the null hypothesis is composed of rank-one interference and noise clutter. Based on the data matrix to be detected and the training sample data vector, estimate the parameters under the null hypothesis, including the interference coordinate vector, the interference amplitude vector, and the clutter covariance matrix; Based on the training sample data vector, the known interference matrix, and the known signal matrix, estimate the parameters under the alternative hypothesis, including the interference coordinate vector and the interference amplitude vector; The test statistic is constructed based on the parameters estimated under the null and alternative hypotheses, and the detection threshold is determined by combining the false alarm probability of the system. Compare the test statistic with the detection threshold. If the test statistic is greater than the detection threshold, the target is determined to exist; otherwise, the target does not exist.

2. The extended target detection method under uncertain interference steering vector according to claim 1, characterized in that, The binary hypothesis testing model includes a null hypothesis and alternative hypotheses, the null hypothesis being... for: Alternative Hypothesis for: in, for A 3D matrix of data to be detected; This is a noise component; for The interference matrix is ​​known in dimension; for Unknown interference coordinate vector; for An unknown disturbance amplitude vector; for dimensional training sample data vector; for A known signal matrix; for An unknown signal coordinate matrix; The number of training sample data vectors; Number of channels; Expand the dimension of the target; The training sample vector is pure noise.

3. The extended target detection method under uncertain interference steering vector according to claim 2, characterized in that, Based on the data matrix to be detected and the training sample data vector, the parameters under the null hypothesis are estimated. These parameters include the interference coordinate vector, the interference amplitude vector, and the clutter covariance matrix, including: Based on the data matrix to be detected and the training sample data vector, estimate the parameters under the null hypothesis and estimate the interference coordinate vector. For matrix The normalized eigenvector corresponding to the largest eigenvalue; Estimated disturbance amplitude vector for: Estimated clutter covariance matrix for: in, The covariance matrix is ​​sampled for the training samples.

4. The extended target detection method under uncertain interference steering vector according to claim 3, characterized in that, Based on the training sample data vector, the known interference matrix, and the known signal matrix, estimate the parameters under the alternative hypothesis, including the interference coordinate vector and the interference amplitude vector, including: Construct the intermediate variable matrix based on the training sample data vector and the known signal matrix: in, This is an intermediate variable matrix; Construct a matrix based on the intermediate variable matrix and the known interference matrix. : in, for An identity matrix of order 1; With matrix The normalized eigenvector corresponding to the largest eigenvalue is the estimated value of the disturbance coordinate vector under the alternative hypothesis. ; Estimated value of disturbance amplitude vector under alternative assumptions for: 。 5. The extended target detection method under uncertain interference steering vector according to claim 4, characterized in that, The test statistic is: in, It is the absolute value; Let be the trace of the matrix.

6. The extended target detection method under uncertain interference steering vector according to claim 5, characterized in that, Based on the estimated parameters, a test statistic is constructed, and the detection threshold is determined in conjunction with the system's false alarm probability, including: Construct a test statistic based on the estimated parameters; The detection threshold is determined using standard Monte Carlo simulation based on the test statistic and the system false alarm probability. in, For detection threshold; This represents the probability of a false alarm in the system. To be The detection statistics obtained under the null hypothesis, arranged in descending order, are as follows: One value, Number of Monte Carlo simulations; This is a floor operation.

7. An extended target detection system under uncertain interference steering vector conditions, characterized in that, The system includes: The model building module is used to obtain the data matrix to be detected and the training sample data vector, and to build a binary hypothesis testing model. In the binary hypothesis testing model, the null hypothesis is composed of rank-one interference and noise clutter. The null hypothesis parameter estimation module is used to estimate the parameters under the null hypothesis based on the data matrix to be detected and the training sample data vector, including the interference coordinate vector, the interference amplitude vector and the clutter covariance matrix. The alternative hypothesis parameter estimation module is used to estimate the parameters under the alternative hypothesis, including the interference coordinate vector and the interference amplitude vector, based on the training sample data vector, the known interference matrix, and the known signal matrix. The detection setting module is used to construct a test statistic based on the parameters estimated under the null and alternative hypotheses, and to determine the detection threshold in combination with the system false alarm probability. The detection execution module compares the test statistic with the detection threshold. If the test statistic is greater than the detection threshold, the target is determined to exist; otherwise, the target does not exist.

Images