Signal filtering method based on array signal processing technology

Abstract

The invention discloses a signal filtering method based on an array signal processing technology, and the method comprises the steps: carrying out the sampling of a space signal field through constructing a multi-sensor array, and achieving the spatial filtering through a beam forming technology. The method comprises the following steps: firstly, designing an adaptive weighting coefficient according to a signal environment, enabling a main lobe of an array directional diagram to point to a target signal direction by adjusting the phase and amplitude of each array element receiving signal, and meanwhile, forming null in an interference direction by using a zero point forming technology to suppress an unexpected direction signal; aiming at a color noise background, a Wiener filtering criterion is adopted to optimize a weighting coefficient, and the filtering performance is improved; in combination with a matrix filtering technology, through setting passband and stopband response constraints, array element domain data preprocessing is realized, target signals are further enhanced, and interference is suppressed. The method breaks through the limitation of traditional time domain filtering, improves the signal resolution and the anti-interference capability by using the spatial domain sampling characteristic, is suitable for high-precision signal filtering requirements in the fields of radar, communication, sonar and the like, and effectively improves the signal-to-noise ratio of a system and the target detection accuracy.

Description

Technical Field

[0001] This invention relates to the field of array signal processing technology, specifically to a signal filtering method based on array signal processing technology, which can be applied to various scenarios such as anti-interference of communication systems, radar target detection, sonar signal processing, or wireless sensor network positioning. Background Art

[0002] In the field of array signal processing, signal filtering is a crucial step, and its performance directly affects the performance of the entire system. Traditional signal filtering methods have many problems when processing array signals in complex environments.

[0003] On the one hand, the computational complexity is high. In practical applications, the covariance matrix of the received signal has a large dimension, and directly processing it requires a lot of computational resources and time, making it difficult to meet the application scenarios with high real-time requirements.

[0004] On the other hand, non-ideal factors have a significant impact. In actual operation, the array is affected by non-ideal factors such as amplitude and phase errors and mutual coupling effects. These factors can cause deviations in the array's steering vector, thereby reducing filtering performance and making it difficult to extract the target signal.

[0005] Furthermore, traditional methods lack the ability to adapt to real-time environmental changes. In real-world environments, the statistical characteristics of signals change over time, while traditional methods typically use fixed filter weights, which cannot be adjusted in a timely manner according to environmental changes, leading to a decline in filtering effectiveness.

[0006] Therefore, a new signal filtering method is needed to solve the above problems and improve the performance and efficiency of array signal processing. Summary of the Invention

[0007] The purpose of this invention is to provide a signal filtering method based on array signal processing technology. This method can reduce computational complexity, compensate for array errors, and adaptively and dynamically update filter weights, thereby improving the accuracy and real-time performance of signal filtering. It is applicable to array signal processing scenarios in various complex environments.

[0008] To achieve the above objectives, the present invention provides the following technical solution:

[0009] A signal filtering method based on array signal processing technology includes the following steps:

[0010] (1) Construct an array signal receiving model, collect spatial signal samples and perform preprocessing;

[0011] (2) Estimate the covariance matrix based on the received signal and reconstruct the low-dimensional covariance matrix through eigenvalue decomposition to reduce computational complexity;

[0012] (3) Design a space-time joint filter, construct an optimization objective function that includes covariance matrix constraints and target direction guidance vector constraints, and use a fast convergence algorithm to solve for the optimal filter weight vector;

[0013] (4) An array error compensation mechanism is introduced to reduce the impact of non-ideal factors such as amplitude and phase errors and mutual coupling effects on filtering performance through iterative correction algorithms;

[0014] (5) Dynamically update the covariance matrix and filter weights according to real-time environmental changes, and output the filtered signal.

[0015] Furthermore, the specific method for reconstructing the covariance matrix in step (2) is as follows:

[0016] Perform eigenvalue decomposition on the covariance matrix R of the received signal:

[0017] R=UΣU H

[0018] Where U is the eigenvector matrix and Σ is the eigenvalue diagonal matrix; the low-dimensional covariance matrix R′ is reconstructed by selecting the eigenvalues ​​and eigenvectors corresponding to the signal subspace:

[0019] R′=U s Σ s U sH

[0020] In the formula U s and Σ s These are the eigenvector matrix and the diagonal matrix of eigenvalues ​​of the signal subspace, respectively.

[0021] Furthermore, the objective function optimized in step (3) is:

[0022]

[0023] Where w is the filter weight vector, α(θ) O ) is the target direction guidance vector, θ O The incident angle of the target signal is denoted as ; the fast convergence algorithm is an improved normalized least mean square algorithm or an iterative algorithm based on QR decomposition.

[0024] Furthermore, the array error compensation mechanism in step (4) is implemented as follows:

[0025] (a) Construct an array error model, including the amplitude and phase error matrix (Γ) and the mutual coupling matrix C;

[0026] (b) Update the error parameters using the least squares iterative algorithm or gradient descent method, so that the compensated steering vector a'(θ) satisfies:

[0027] a′(θ)=ΓCa(θ)

[0028] (c) Substitute the compensated steering vector into the optimization objective function of step (3) and resolve the filter weights.

[0029] Furthermore, the dynamic update in step (5) is implemented as follows:

[0030] (a) Trigger covariance matrix update according to a preset time window or signal-to-noise ratio threshold;

[0031] (b) The historical covariance matrix and the current sampled data are weighted and fused using the moving average method or the exponential forgetting factor method to generate the updated covariance matrix R. new ;

[0032] (c) Based on R new Repeat steps (2)-(4) to achieve adaptive adjustment of filter weights.

[0033] Furthermore, the array signal receiving model is a uniform linear array, a uniform circular array, or a planar array, and the number of array elements M satisfies M≥4.

[0034] Furthermore, the preprocessing includes at least one of bandpass filtering, gain control, and DC component removal.

[0035] Furthermore, the method can be applied to scenarios such as communication system anti-jamming, radar target detection, sonar signal processing, or wireless sensor network positioning.

[0036] Beneficial effects of this invention:

[0037] Reduced computational complexity: By reconstructing the low-dimensional covariance matrix through eigenvalue decomposition, the dimensionality of matrix operations is reduced, significantly reducing the amount of computation and improving the real-time performance of signal processing, making it particularly suitable for application scenarios with high real-time requirements.

[0038] Compensating for array errors: Introducing an array error compensation mechanism can effectively reduce the impact of non-ideal factors such as amplitude and phase errors and mutual coupling effects on filtering performance, improve the accuracy of target signal extraction, and enhance the stability and reliability of the system.

[0039] Adaptive dynamic update: The covariance matrix and filter weights are dynamically updated according to real-time environmental changes, enabling the system to adaptively adjust the filtering parameters and maintain good filtering performance at all times, making it suitable for complex and ever-changing real-world environments.

[0040] Wide range of applications: This invention can be applied to various scenarios such as anti-interference of communication systems, radar target detection, sonar signal processing or wireless sensor network positioning, and has broad application prospects and practical value. Attached Figure Description

[0041] Figure 1 This is a flowchart illustrating a signal filtering method based on array signal processing technology according to the present invention. Detailed Implementation

[0042] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0043] The present invention provides a signal filtering method based on array signal processing technology, comprising the following steps:

[0044] (1) Construct an array signal receiving model, collect spatial signal samples and perform preprocessing.

[0045] The array signal receiving model can be a uniform linear array, a uniform circular array, or a planar array, and the number of array elements M satisfies M≥4. This choice of multiple array forms makes the present invention more widely applicable and can be flexibly selected according to different application scenarios and actual needs.

[0046] The preprocessing includes at least one of bandpass filtering, gain control, and DC component removal. Preprocessing can remove noise, interference, and DC components from the signal, improve the signal quality, and provide more accurate input for subsequent processing.

[0047] (2) Estimate the covariance matrix based on the received signal, and reconstruct the low-dimensional covariance matrix through eigenvalue decomposition to reduce computational complexity. Perform eigenvalue decomposition on the covariance matrix R of the received signal:

[0048] R=UΣU H

[0049] Where U is the eigenvector matrix and Σ is the eigenvalue diagonal matrix; the low-dimensional covariance matrix R′ is reconstructed by selecting the eigenvalues ​​and eigenvectors corresponding to the signal subspace:

[0050] R′=U s Σ s U sH

[0051] In the formula U s and Σ s These are the eigenvector matrix and the diagonal matrix of eigenvalues ​​of the signal subspace, respectively.

[0052] By reconstructing the low-dimensional covariance matrix, the dimension of the matrix to be processed is reduced, thereby significantly reducing computational complexity and improving processing efficiency.

[0053] (3) Design a space-time joint filter, construct an optimization objective function that includes covariance matrix constraints and target direction steering vector constraints, and use a fast convergence algorithm to solve for the optimal filter weight vector.

[0054] The optimization objective function is:

[0055]

[0056] Where w is the filter weight vector, α(θ) O ) is the target direction guidance vector, θ O The target signal incident angle is denoted by . This optimized objective function comprehensively considers the constraints of the covariance matrix and the target direction steering vector, enabling more accurate extraction of the target signal.

[0057] The fast convergence algorithm is an improved normalized least mean square algorithm or an iterative algorithm based on QR decomposition. These algorithms have the characteristic of fast convergence and can solve for the optimal filter weight vector in a short time, thus meeting the real-time requirements.

[0058] (4) An array error compensation mechanism is introduced, and the influence of non-ideal factors such as amplitude and phase errors and mutual coupling effects on the filtering performance is reduced through iterative correction algorithms.

[0059] Construct an array error model, including the amplitude and phase error matrix Γ and the mutual coupling matrix C;

[0060] The error parameters are updated using a least squares iterative algorithm or gradient descent method, so that the compensated steering vector a'(θ) satisfies:

[0061] a′(θ)=ΓCa(θ)

[0062] Substitute the compensated steering vector into the optimization objective function of step (3) to resolve the filter weights; by introducing an array error compensation mechanism, the influence of non-ideal factors on the filtering performance can be effectively reduced, and the accuracy and stability of filtering can be improved.

[0063] (5) Dynamically update the covariance matrix and filter weights according to real-time environmental changes, and output the filtered signal.

[0064] The covariance matrix is ​​updated by triggering a preset time window or signal-to-noise ratio threshold. This triggering method can flexibly adjust the timing of the covariance matrix update according to the changes in the actual environment, avoid unnecessary update operations, and improve the efficiency of the system.

[0065] The historical covariance matrix and the current sampled data are weighted and fused using either the moving average method or the exponential forgetting factor method to generate an updated covariance matrix. These two methods fully consider the information from both historical and current data, making the updated covariance matrix more accurately reflect the statistical characteristics of the current signal.

[0066] Based on re-execution of steps (2)-(4), the adaptive adjustment of filter weights is realized; by dynamically updating the covariance matrix and filter weights, the present invention can adaptively respond to changes in the real-time environment and always maintain good filtering performance.

[0067] In practice:

[0068] (I) Constructing the array signal reception model and preprocessing

[0069] Assuming a uniform linear array is used as the array signal receiving model, with M=8 array elements, spatial signal samples are collected through the antenna array. The collected signals are then preprocessed, including bandpass filtering, gain control, and DC component removal. Bandpass filtering can remove noise and interference outside the signal frequency band; gain control can adjust the signal amplitude to keep it within a suitable range; and DC component removal can eliminate DC offset in the signal and improve signal quality.

[0070] (II) Covariance Matrix Reconstruction

[0071] Estimate the covariance matrix R of the preprocessed received signal, and then perform eigenvalue decomposition on R:

[0072] R=UΣU H

[0073] Based on the characteristics of the signal subspace and the noise subspace, the eigenvalues ​​and eigenvectors corresponding to the signal subspace are selected to reconstruct the low-dimensional covariance matrix R′:

[0074] R′=U s Σ s U sH

[0075] For example, suppose the eigenvalues ​​obtained through eigenvalue decomposition are arranged in descending order as λ1, λ2, ..., λ8. The first three larger eigenvalues ​​and their corresponding eigenvectors are selected to reconstruct the low-dimensional covariance matrix, which can greatly reduce the complexity of subsequent calculations.

[0076] (III) Solving for the optimal filter weight vector

[0077] Design a space-time joint filter and construct an optimization objective function:

[0078]

[0079] An improved normalized least mean square algorithm is used to solve the optimization objective function. The specific steps are as follows:

[0080] Initialize the filter weight vector w(0), and set parameters such as the iteration step size μ;

[0081] In each iteration, the error signal is calculated based on the received signal sample and the current weight vector;

[0082] The weight vector w(n+1) is updated based on the error signal and the improved normalized least mean square algorithm: w(n+1) = w(n) - μe(n)x(n) / (||x(n)|| 2 +δ), where e(n) is the error signal, x(n) is the input signal vector, and δ is a small positive number used to prevent the denominator from being zero;

[0083] Repeat steps 2 and 3 until the convergence condition is met, and obtain the optimal filter weight vector w.

[0084] (iv) Array error compensation

[0085] Construct an array error model, including the amplitude and phase error matrix Γ and the mutual coupling matrix C;

[0086] Assume the amplitude and phase error matrix Γ is a diagonal matrix, whose diagonal elements represent the amplitude and phase errors of each array element; the mutual coupling matrix (C) reflects the mutual coupling effect between array elements;

[0087] The error parameters are updated by the least squares iterative algorithm so that the compensated steering vector a′(θ)=ΓCa(θ) is as close as possible to the ideal steering vector; the compensated steering vector is substituted into the optimization objective function to resolve the filter weights in order to improve the filtering performance.

[0088] (V) Dynamically updating the covariance matrix and filter weights

[0089] The covariance matrix is ​​updated according to a preset time window, for example, every 100 sampling points. A moving average method is used to weight and fuse the historical covariance matrix with the current sampling data to generate the updated covariance matrix.

[0090] Assume the historical covariance matrix is ​​R old The covariance matrix estimated from the current sampled data is R. new The updated covariance matrix R new =αR old +(1-α)R new , where α is a weighting factor with a value range of [0,1]; based on the updated covariance matrix, steps (ii)-(iv) are executed again to achieve adaptive adjustment of the filter weights and output the filtered signal.

[0091] The present invention and its embodiments are described above. This description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if a person skilled in the art is inspired by this and, without departing from the purpose of the present invention, designs structures and embodiments similar to this technical solution without inventiveness, they shall fall within the scope of protection of the present invention.

Claims

1. A signal filtering method based on array signal processing technology, characterized in that, Includes the following steps: (1) Construct an array signal receiving model, collect spatial signal samples and perform preprocessing; (2) Estimate the covariance matrix based on the received signal and reconstruct the low-dimensional covariance matrix through eigenvalue decomposition to reduce computational complexity; (3) Design a space-time joint filter, construct an optimization objective function that includes covariance matrix constraints and target direction guidance vector constraints, and use a fast convergence algorithm to solve for the optimal filter weight vector; (4) An array error compensation mechanism is introduced to reduce the impact of non-ideal factors such as amplitude and phase errors and mutual coupling effects on filtering performance through iterative correction algorithms; (5) Dynamically update the covariance matrix and filter weights according to real-time environmental changes, and output the filtered signal.

2. The signal filtering method based on array signal processing technology according to claim 1, characterized in that: The specific method for reconstructing the covariance matrix in step (2) is as follows: Perform eigenvalue decomposition on the covariance matrix R of the received signal: R=UΣU H Where U is the eigenvector matrix and Σ is the eigenvalue diagonal matrix; the low-dimensional covariance matrix R′ is reconstructed by selecting the eigenvalues ​​and eigenvectors corresponding to the signal subspace: R′=U s Σ s IN sH In the formula U s and Σ s These are the eigenvector matrix and the diagonal matrix of eigenvalues ​​of the signal subspace, respectively.

3. The signal filtering method based on array signal processing technology according to claim 1, characterized in that: The objective function to be optimized in step (3) is: Where w is the filter weight vector, α(θ) O ) is the target direction guidance vector, θ O The incident angle of the target signal is denoted as ; the fast convergence algorithm is an improved normalized least mean square algorithm or an iterative algorithm based on QR decomposition.

4. The signal filtering method based on array signal processing technology according to claim 1, characterized in that: The array error compensation mechanism in step (4) is implemented as follows: (a) Construct an array error model, including the amplitude and phase error matrix (Γ) and the mutual coupling matrix C; (b) Update the error parameters using the least squares iterative algorithm or gradient descent method, so that the compensated steering vector a'(θ) satisfies: a′(θ)=ΓCa(θ) (c) Substitute the compensated steering vector into the optimization objective function of step (3) and resolve the filter weights.

5. The signal filtering method based on array signal processing technology according to claim 1, characterized in that: The dynamic update in step (5) is implemented as follows: (a) Trigger covariance matrix update according to a preset time window or signal-to-noise ratio threshold; (b) The historical covariance matrix and the current sampled data are weighted and fused using the moving average method or the exponential forgetting factor method to generate the updated covariance matrix R. new ; (c) Based on R new Repeat steps (2)-(4) to achieve adaptive adjustment of filter weights.

6. A signal filtering method based on array signal processing technology according to any one of claims 1-5, characterized in that: The array signal receiving model is a uniform linear array, a uniform circular array, or a planar array, and the number of array elements M satisfies M≥4.

7. A signal filtering method based on array signal processing technology according to any one of claims 1-5, characterized in that: The preprocessing includes at least one of bandpass filtering, gain control, and DC component removal.

8. A signal filtering method based on array signal processing technology according to any one of claims 1-5, characterized in that: The method is applicable to scenarios such as communication system anti-jamming, radar target detection, sonar signal processing, or wireless sensor network positioning.

Images