An Adaptive Wideband Weighted Beamforming Method

Abstract

The present invention discloses an adaptive broadband weighted beamforming method, comprising the following steps: calculating the average energy of each sub-band based on the acquired sub-band array data; setting adjustable control parameters, and calculating the corresponding amplitude weighting operator of each diagonal loading amount based on the average energy of each sub-band; performing amplitude weighting processing on the sub-band array data to obtain the array data after amplitude weighting of each sub-band; iteratively calculating the inverse of the corresponding covariance matrix by using the array data after amplitude weighting of each sub-band; and calculating the broadband spatial spectrum by using the minimum variance distortionless beamformer based on the inverse of the sub-band covariance matrix. The present invention performs amplitude weighting processing based on the diagonal loading amount on the frequency-domain array data of the array, and combines the iterative inversion method, which can ensure the stability of inversion while improving the broadband detection performance, and the calculation efficiency is greatly improved. The present invention can solve the problem that the frequency components with good detection performance are overloaded, thus affecting the broadband detection information.

Description

Technical Field

[0001] The present invention belongs to the field of sonar signal processing, and relates to the research on the adaptive broadband vigilance of sonar arrays. More specifically, it relates to an adaptive broadband weighted beamforming method. Background Art

[0002] For the signal detection problem of large-aperture or large-scale sonar arrays, in addition to the computational requirements, the factors restricting the application of the adaptive broadband processing method (MVDR) also include the problem of estimating the array covariance matrix. According to the existing theory, only when the number of snapshots is greater than or equal to twice the number of array elements, the performance loss of the actual MVDR can be controlled within 3 dB. For multi-element large-scale sonars, due to factors such as target movement and signal stationarity, it is difficult to meet the number of snapshots for estimating the array covariance matrix, resulting in a decline in the signal detection ability of MVDR. At this time, it is necessary to ensure both the rapid convergence of the covariance matrix and the reliable and robust engineering implementation of the signal processor, which requires diagonal loading. In engineering applications, in order to reduce hardware requirements and improve computational efficiency, an iterative method is generally used to calculate the inverse of the covariance matrix. However, when the same fixed amount is used for diagonal loading of the covariance matrices at all frequencies, it is easy to cause overloading of some frequency components with good detection performance, affecting the broadband detection performance. Therefore, it is necessary to adaptively adjust the amplitude of the loading amount by combining the data of each snapshot and each frequency point to improve the detection performance of the adaptive broadband vigilance. Summary of the Invention

[0003] One of the purposes of the present invention is to provide an adaptive broadband weighted beamforming method to solve the problem that the conventional MVDR method in the background art is prone to overloading of frequency components with good detection performance, thus affecting the broadband detection information.

[0004] To achieve the above purpose, the present invention provides the following technical solutions:

[0005] An adaptive broadband weighted beamforming method includes the following steps: calculating the average energy of each sub-band based on the obtained sub-band array data; setting adjustable control parameters, and calculating the corresponding amplitude weighted operator of each diagonal loading amount based on the average energy of each sub-band; performing amplitude weighting processing on the sub-band array data to obtain the sub-band array data after amplitude weighting; iteratively calculating the inverse of the corresponding covariance matrix using the sub-band array data after amplitude weighting; calculating the broadband spatial spectrum using the minimum variance distortionless beamformer based on the inverse of the sub-band covariance matrix.

[0006] Preferably, the method for obtaining the sub-band array data includes the following steps: performing a fast Fourier transform on the time-domain data in the element domain; dividing the transformed result into several narrow sub-bands that meet the preset narrowband assumption in the frequency domain; taking the element-domain frequency-domain data of each sub-band within the processing frequency band.

[0007] Preferably, the calculation formula for the average energy Xeg(f) of the sub-band is as follows:

[0008]

[0009] where N is the number of array elements, X(n, f) is the array element domain frequency domain data of the nth array element at frequency f, and n is the array element number.

[0010] Preferably, the calculation formula for the amplitude weighting operator α() of the diagonal loading amount is as follows:

[0011] α(f) = 1.0 / (Xeg(f) * λ)

[0012] where λ is the adjustable control parameter, and Xeg(f) is the average energy at frequency f.

[0013] Preferably, the calculation formula for the amplitude weighting process is as follows:

[0014] Y(:, f) = α(f) * X(:, f)

[0015] where Y(:, f) is the array element domain frequency domain data after amplitude weighting at frequency f, and X(:, f) is the array element domain frequency domain data at frequency f.

[0016] Preferably, the iterative calculation formula is as follows:

[0017]

[0018] where and are the inverse of the current iteration and the inverse of the previous iteration respectively, j is the number of iterations, H is the conjugate transpose, and Y(:, f) is the array element domain frequency domain data after amplitude weighting at frequency f.

[0019] Preferably, calculating the broadband spatial spectrum using the minimum variance distortionless beamformer includes the following steps: calculating the array steering vector of each sub-band according to the base array element coordinates, spatial scanning azimuth, and processing frequency band; calculating the adaptive broadband weighted spatial spectrum of the corresponding sub-band using the array steering vector of each sub-band and the inverse of the covariance matrix; performing weighted summation processing on the adaptive broadband weighted spatial spectra of all sub-bands to obtain the broadband spatial spectrum.

[0020] Preferably, the calculation formula for the array steering vector a_scan(:, f) is as follows:

[0021] a_scan(:, f) = exp(j2πf(xcosθ + ysinθ) / c)

[0022] where f is the processing frequency, (x, y) is the coordinate vector of the base array element, θ is the scanning azimuth vector, c is the speed of sound, and j is the imaginary number.

[0023] Preferably, the calculation formula of the adaptive broadband weighted spatial spectrum Pmvdr(f, θ) of the sub-band is as follows:

[0024]

[0025] where a_scan(:, f) is the array steering vector at frequency f, and H is the conjugate transpose, is the inverse of the covariance matrix.

[0026] Preferably, the weighted summation formula of the adaptive broadband weighted spatial spectra of all sub-bands is as follows:

[0027] Mvdr(θ) = ∑ f Pmvdr(f, θ).

[0028] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0029] In the present invention, amplitude weighting processing based on diagonal loading amount is performed on the frequency-domain array data of the array, and combined with the iterative inversion method, it can ensure stable inversion while improving the broadband detection performance, and the calculation efficiency is greatly improved. The present invention can solve the problem that the frequency components with good detection performance are overloaded, thus affecting the broadband detection information. Description of the Drawings

[0030] Figure 1 Flowchart of the adaptive broadband weighted beamforming method of the present invention.

[0031] Figure 2 Processing results of sea trial data of a certain side array.

[0032] Figure 3 Processing results of snapshots at two moments of 120 s and 300 s. Detailed Embodiments

[0033] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0034] An adaptive broadband weighted beamforming method specifically includes the following five steps:

[0035] Step 1: Based on the obtained sub-band array element domain frequency domain data X(:, f), calculate the average energy Xeg(f) of each sub-band,

[0036]

[0037] Where N is the number of array elements, X(n, f) is the array element domain frequency domain data of the nth array element at frequency f, and n is the array element number.

[0038] The method for obtaining the sub-band array data in Step 1 includes the following three sub-steps:

[0039] Step 1-1: Perform a fast Fourier transform on the time domain data in the array element domain;

[0040] Step 1-2: Divide the transformation result into several narrow sub-bands that meet the preset narrowband assumption in the frequency domain;

[0041] Step 1-3: Take the array element domain frequency domain data of each sub-band within the processing frequency band.

[0042] In Step 1-1 of the present invention, the fast Fourier transform is a general term for efficient and fast calculation methods that use a computer to calculate the discrete Fourier transform; the discrete Fourier transform (DFT) is a form in which the Fourier transform is discrete in both the time domain and the frequency domain, and the sampling of the time domain signal is transformed into the sampling in the frequency domain of the discrete-time Fourier transform (DTFT). In form, the sequences at both ends (in the time domain and the frequency domain) of the transformation are of finite length, but in fact, both of these two groups of sequences should be considered as the principal value sequences of discrete periodic signals. Even when performing DFT on a finite-length discrete signal, it should be regarded as a periodic signal after periodic extension and then perform the transformation.

[0043] In Step 1-2 of the present invention, the preset narrowband width of the narrow sub-band needs to be determined according to factors such as the bandwidth of the processing frequency band, and those skilled in the art can set it according to the actual situation.

[0044] Step 2: Set the adjustable control parameter λ = 0.1, and calculate the corresponding amplitude weighting operator α(f) of each diagonal loading amount based on the average energy Xeg(f) of each sub-band,

[0045] α(f) = 1.0 / (Xeg(f) * λ)

[0046] Where Xeg(f) is the average energy at frequency f.

[0047] In Step 2 of the present invention, the specific value of the adjustable control parameter can be set according to the actual situation.

[0048] Step 3: Perform amplitude weighting processing on the array element domain frequency domain data X(:, f) of each sub-band to obtain the array element domain frequency domain data Y(:, f) after amplitude weighting of each sub-band,

[0049] Y(:, f) = α(f) * X(:, f)

[0050] Among them, Y(:, f) is the element-domain frequency-domain data after amplitude weighting at frequency f, and X(:, f) is the element-domain frequency-domain data at frequency f.

[0051] Step 4: Iteratively calculate the inverse of each corresponding covariance matrix using the amplitude-weighted element-domain frequency-domain data.

[0052]

[0053] Among them, and are the inverse of the current iteration and the inverse of the previous iteration respectively, j is the number of iterations, H is the conjugate transpose, and Y(:, f) is the element-domain frequency-domain data after amplitude weighting at frequency f.

[0054] Step 5: Based on the inverse of the sub-band covariance matrix, calculate the broadband spatial spectrum using the minimum variance distortionless beamformer, which specifically includes the following 3 sub-steps.

[0055] Step 5-1: Calculate the array steering vector a_scan(:, f) of each sub-band according to the base array element coordinates, spatial scanning azimuth, and processing frequency band.

[0056] a_scan(:, f) = exp(j2πf(xcosθ + ysinθ) / c)

[0057] Among them, f is the processing frequency, (x, y) is the coordinate vector of the base array element, θ is the scanning azimuth vector, c is the speed of sound, and j is the imaginary number.

[0058] Step 5-2: Calculate the adaptive broadband weighted spatial spectrum Pmvdr(f, θ) of the corresponding sub-band using the array steering vector and the inverse of the covariance matrix of each sub-band.

[0059]

[0060] Among them, a_scan(:, f) is the array steering vector at frequency f, H is the conjugate transpose, is the inverse of the covariance matrix.

[0061] Step 5-3: Perform weighted summation processing on the adaptive broadband weighted spatial spectra of all sub-bands to obtain the broadband spatial spectrum Mvdr(θ).

[0062]

[0063] The present invention uses the sea trial test data of a certain side array for processing and verification. The processing frequency band is 3k - 5kHz, the integration time is 4s, the scanning azimuth is cosine scanning from 0 to 180°. Generally, the MVDR algorithm is used for adaptive beamforming processing. Due to the large scale of the array elements, the number of snapshots corresponding to 4s integration cannot meet twice the number of array elements, and the array covariance matrix is singular. To ensure the operation of the algorithm, the covariance matrix is processed by diagonal loading (adding a fixed value to the diagonal, here taking 1).

[0064] Figure 2 The processing results of the sea trial data of a certain side array are given. The length of the processed data is 6 minutes, attached Figure 2 (a) is the processing result of the present invention using the broadband weighted beamforming method, attached Figure 2 (b) is the processing result of the general adaptive beamforming method (diagonal loading MVDR algorithm); by comparing the processing results of the two methods, it can be found that for the three weak targets with the starting azimuth near 100° and the fast - moving weak target near 70° in the box, from the continuity of the target detection trajectory and the snapshot results, the detection performance of the present invention is better than that of the diagonal loading MVDR algorithm.

[0065] Figure 3 (a) is the snapshot processing result at 120s, Figure 3 (b) is the snapshot processing result at two moments of 300s. For the fast - moving target and the three weak targets with continuous trajectories marked in the figure, the detection performance of the adaptive broadband weighted beamforming method of the present invention is better than that of the general adaptive beamforming method.

[0066] The above - mentioned are only the preferred embodiments of the present invention and are not used to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, for those skilled in the art, they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An adaptive broadband weighted beamforming method, characterized in that, It includes the following steps: Step 1: Based on the obtained sub-band array data X(:, f), calculate the average energy Xeg(f) of each sub-band, where N is the number of array elements, X(n, f) is the array element domain frequency domain data of the nth array element at frequency f, and n is the array element number; The method for obtaining the sub-band array data in Step 1 includes the following three sub-steps: Step 1-1: Perform a fast Fourier transform on the time domain data in the array element domain; Step 1-2: Divide the transformation result into several narrow sub-bands in the frequency domain that satisfy the preset narrowband assumption; Step 1-3: Take the array element domain frequency domain data of each sub-band within the processing frequency band; Step 2: Set the adjustable control parameter λ, and based on the average energy Xeg(f) of each sub-band, calculate the corresponding amplitude weighting operator α(f) of each diagonal loading amount, α(f) = 1.0 / (Xeg(f) * λ) where α(f) is the amplitude weighting operator of the diagonal loading amount, and Xeg(f) is the average energy at frequency f; Step 3: Perform amplitude weighting processing on the array element domain frequency domain data X(:, f) of each sub-band to obtain the array element domain frequency domain data Y(:, f) after amplitude weighting of each sub-band, Y(:, f) = α(f) * X(:, f) where Y(:, f) is the array element domain frequency domain data after amplitude weighting at frequency f, and X(:, f) is the array element domain frequency domain data at frequency f; Step 4: Iteratively calculate the inverse of the corresponding covariance matrix using the array element domain frequency domain data after amplitude weighting of each sub-band, where \(R_j(f)\) is the covariance matrix of the current iteration, and are the inverse of the covariance matrix of the current iteration and the inverse of the covariance matrix of the previous iteration, respectively, \(j\) is the number of iterations, \(H\) is the conjugate transpose, and \(Y(:,f)\) is the element-domain frequency-domain data weighted by the amplitude at frequency \(f\); Step 5: Based on the inverse of the sub-band covariance matrix, use the minimum variance distortionless beamformer to calculate the broadband spatial spectrum, including the following steps: Step 5-1: According to the array element coordinates of the base array, the spatial scanning azimuth, and the processing frequency band, calculate the array steering vector a_scan(:, f) of each sub-band, a_scan(:, f) = exp(j2πf(xcosθ + ysinθ) / c) where f is the processing frequency, (x, y) is the coordinate vector of the base array elements, θ is the scanning azimuth vector, c is the speed of sound, and j is the imaginary number; Step 5-2: Use the array steering vector of each sub-band and the inverse of the covariance matrix to calculate the adaptive broadband weighted spatial spectrum Pmvdr(f, θ) of the corresponding sub-band, where a_scan(:, f) is the array steering vector at frequency f, H is the conjugate transpose, is the inverse of the covariance matrix; Step 5-3: Perform weighted summation processing on the adaptive broadband weighted spatial spectra of all sub-bands to obtain the broadband spatial spectrum Mvdr(θ),

Images