Uniform area array constant beam width linear constraint minimum variance beam forming device

Abstract

The utility model discloses a uniform area array constant beam width linear constraint minimum variance wave beam forming device. The device comprises a plane array with microphone array elements uniformly distributed, a first-stage plane sub-array signal processing module, a second-stage plane sub-array signal processing module and a signal conversion processor. According to the utility model, two stages of sub-arrays are respectively and independently designed, the antenna has the advantages of high fixed beam robustness, low complexity and strong adaptive beam interference noise suppression capability, simultaneously meets the requirements of constant beam width in an expected direction and wide null suppression in an interference direction, is simple in design and low in complexity, and can support applications with high real-time requirements.

Description

TECHNICAL FIELD

[0001] The utility model relates to microphone array signal acquisition processing technical field, concretely relates to a kind of uniform surface array constant beam width linearly constrained minimum variance beam forming device. BACKGROUND

[0002] The existing broadband beam forming technology is mainly divided into two categories of fixed beam forming technology and adaptive beam forming technology.Constant beam width fixed beam forming technology realizes the frequency invariance of beam width on wide frequency band, and avoids signal distortion when the beam direction is not accurately aligned with the actual direction of the desired signal source, but this kind of technology is difficult to form null to suppress interference signals in a specific direction.The other kind of adaptive beam forming technology mainly uses linearly constrained minimum variance (LCMV) beam forming technology, which can form null in the direction of interference source to block directional interference, but cannot maintain constant beam width on wide frequency band, and the sidelobe is high in the non-interference direction, especially when the number of array elements is large, resulting in high complexity of covariance matrix inversion and poor robustness, which is difficult to realize in practical application.The latest beam forming technology applies constraint conditions on the desired direction and interference direction, and then uses software cxv tool to solve the optimization problem to obtain weight coefficients.This kind of technology can better control beam and null, but still has the problems of high calculation complexity and poor robustness, and when the direction of target sound source or interference sound source changes in real time, it is necessary to call the software tool again to solve the optimization problem, so the flexibility is not high, and it is difficult to apply in occasions with high real-time requirements. UTILITY MODEL CONTENT

[0003] In view of the deficiencies of the prior art, the utility model aims at providing a kind of uniform surface array constant beam width linearly constrained minimum variance beam forming device.

[0004] In order to realize the above-mentioned purpose, the utility model adopts the following technical solutions:

[0005] A kind of uniform surface array constant beam width linearly constrained minimum variance beam forming device, characterized by, including the plane array that microphone array element is uniformly distributed, first level plane subarray signal processing module, second level plane subarray signal processing module and signal conversion processor;

[0006] The plane array is used to collect sound signals in the environment, and each frame of sound signal is converted into a frequency domain signal;

[0007] The plane array is divided into a plurality of first level plane subarrays, and the first level plane subarray signal processing module is used to perform beam forming on each first level plane subarray.The second level plane subarray signal processing module is used to perform beam forming on the frequency domain signal output by the first level plane subarray signal processing module to obtain the final beam forming output frequency domain signal;

[0008] The signal conversion processor is configured to convert the beamforming output frequency domain signal into a time-frequency signal.

[0009] Further, the microphone array elements are omnidirectional microphones.

[0010] The utility model discloses beneficial effect lies in: the utility model separately designs two stage subarray, possesses fixed beam robustness high, low complexity and adaptive beam interference noise suppression ability strong's advantage, meets the requirement of constant beam width on the expected direction and the wide null suppression of forming on the interference direction simultaneously, and the design is simple and low complexity, can support the application of high real -time requirement. BRIEF DESCRIPTION OF DRAWINGS

[0011] Figure 1 It is the overall flow chart of the utility model embodiment 1 method;

[0012] Figure 2 It is the plane array and sound source direction schematic drawing of the utility model embodiment 1 method;

[0013] Figure 3 It is the first level plane subarray and second level plane subarray construction schematic drawing of the utility model embodiment 1 method;

[0014] Figure 4 It is the device structure schematic drawing of the utility model embodiment 2. DETAILED DESCRIPTION

[0015] The utility model will be further described in conjunction with the drawings, and it is to be explained that the embodiment is the prerequisite of the technical scheme, gives detailed implementation mode and specific operation process, but the protection scope of the utility model is not limited to the embodiment.

[0016] Embodiment 1

[0017] The embodiment provides a uniform surface array constant beam width linear constraint minimum variance beamforming method, as shown in Figure 1 The method comprises the following steps:

[0018] Step 1, the microphone array elements are uniformly distributed in the plane array to collect the sound signals in the environment, and each frame of sound signal collected by the plane array is converted to the frequency domain by short-time Fourier transform processing.

[0019] The space topology structure is used to collect the sound signals by the microphone array elements uniformly distributed in the plane array, Figure 2 It is the plane array and sound source direction schematic drawing, and the microphone array elements are uniformly distributed on the two-dimensional plane of x-axis and y-axis, and are composed of MxN omnidirectional microphones, M is the row number of the plane array in the y-axis direction, N is the column number of the plane array in the x-axis direction, and the spacing of the microphone array elements on the x-axis and y-axis is δx and δ y . Taking the microphone element in the upper left corner of the planar array as the reference point and the origin of the coordinate system, each microphone element can be represented by the serial number (m, n), m = 0, 1, …, M-1, n = 0, 1, …, N-1, where m and n represent the serial numbers of the rows and columns of the microphone element in the planar array, respectively. The spatial position coordinates of the microphone element (m, n) in the coordinate system can be represented as (nδ x , mδ y , 0), and the time-domain signal collected by the microphone element (m, n) is converted into the frequency-domain form X mn (k) through short-time Fourier transform, where k is the serial number of the frequency point f k , f k = kf s / N FFT , f s is the sampling rate, and N FFT is the Fourier transform length. Considering the far-field acoustic model, let θ and φ represent the elevation angle and azimuth angle of the sound source, respectively. The steering matrix of the M × N planar array is defined as:

[0020]

[0021] where j represents the imaginary unit, D(f, θ, φ) is the array steering matrix, [D(f, θ, φ)] mn represents the element in the mth row and nth column of the array steering matrix, and the wave number parameter c is the propagation speed of sound in the environment, and f represents the frequency. By stacking the columns of the array steering matrix into a vector representation, the corresponding array steering vector is obtained:

[0022]

[0023] In this embodiment, M = 8, N = 10, δ x = 0.04 meters, δ y = 0.05 meters, there is a target sound source s in the space, the elevation angle and azimuth angle of the target sound source s are (θ s , φ s ) = (10°, 180°), and the elevation angle and azimuth angle of the two interference sound sources j1 and j2 are and

[0024] Step 2: The planar array is regarded as a cascade representation of two-level planar subarrays, and the weight matrix of the planar array can be represented in the form of two-dimensional convolution of the weight matrices of the two-level planar subarrays. The dimension parameters of the two-level planar subarrays are set respectively to obtain the corresponding subarray steering matrix and subarray steering vector. The specific process is as follows:

[0025] The plane array beamforming is a spatial linear filtering of the array frequency domain signal, and the spatial filtering weight coefficient of the plane array can be represented by a weight matrix W:

[0026]

[0027] The weight matrix is stacked by column to obtain a weight vector of the plane array:

[0028] w = [w 00 ,…,w m0 ,…,w (M-1)0 ,…,w mn ,…,w (M-1)(N-1) ] T (4)

[0029] The beam pattern corresponding to the weight vector of the plane array is B(θ,φ,w) = d H (θ,φ)w, the plane array is regarded as a cascade of two-level plane subarrays, and the beam pattern thereof is equivalent to the multiplication of two sub-beam patterns B(θ,φ,w) = B1(θ,φ,w1)B2(θ,φ,w2), the two-level plane subarrays are designed as a fixed beamformer and an adaptive beamformer respectively, and meet the requirements of constant beam width, controllable sidelobe suppression and directional interference nulling, thereby reducing the design complexity. The weight vectors of the two-level plane subarrays are w1 and w2 respectively. Similar to formula (2), the steering vectors of the two-level plane subarrays are:

[0030]

[0031]

[0032] wherein M1 and N1 are the number of rows and columns of the first-level plane subarray, M2 and N2 are the number of rows and columns of the second-level plane subarray, and the beam pattern is represented in the form of Cauchy product:

[0033]

[0034] wherein w mn is the weight coefficient of the microphone element (m, n), and are the weight coefficients of the corresponding elements of the two-level plane subarrays, and for the sake of simplifying the expression, and Since the algorithm is independent and consistent at any frequency point, without loss of generality, the frequency point sequence number k or frequency f is omitted in the formula of the embodiment. According to formula (5), assuming that the row and column numbers of the planar array and the two-stage planar subarray satisfy the relationship of M=M1+M2-1 and N=N1+N2-1, the weight matrix of the planar array can be expressed in the form of two-dimensional convolution of the weight matrix of the two-stage planar subarray:

[0035]

[0036] wherein, parameters a1=max(0,m-1+M2), a2=min(m,M1-1), b1=max(0,n-1+N2) and b2=min(n,N2-1).

[0037] Therefore, the subarray beamformer can be designed individually according to the requirements of constant beam width and interference direction nulling in actual application, which can reduce the computational complexity without affecting the overall performance.

[0038] The row and column number parameters of the two-stage planar subarray are set. In the embodiment, the first-stage planar subarray adopts the fixed beam method to form the main lobe with constant beam width, and at the same time needs to reduce the sidelobe as much as possible, so the values of M1 and N1 are slightly larger, but must be smaller than M and N. The second-stage planar subarray adopts the adaptive beamforming method, which should reduce the number of microphone elements of the second-stage planar subarray as much as possible on the premise of ensuring sufficient degrees of freedom, so as to ensure that the null will not enter the main lobe region, and balance between the noise interference suppression degree and the algorithm complexity.

[0039] In the embodiment, the row and column parameters of the first-stage planar subarray are selected as M1=6 and N1=8, the first-stage planar subarray is a 6×8 planar array, and the row and column parameters of the corresponding second-stage planar subarray can be calculated as M2=M+1-M1=3 and N2=N+1-N1=3, so the second-stage planar subarray is a 3×3 planar array.

[0040] Step 3, dividing the first-stage planar subarray: selecting a first-stage planar subarray on the planar array, obtaining a plurality of first-stage planar subarrays which overlap each other by translation, and using the two-dimensional Chebyshev window constant beam width beamforming method to weight and sum the frequency domain signals of each first-stage planar subarray to obtain the frequency domain output of each first-stage planar subarray. The specific process is as follows:

[0041] 3.1) Dividing the first-stage planar subarray as Figure 3As shown, using the property of planar array translational invariance, according to the row parameter M1 and column parameter N1 set in step 2, M1 rows of continuous N1 columns are selected to form a M1 x N1 sub-array, and (M-M1+1)(N-N1+1)=M2N2 first-level planar sub-arrays can be obtained on the M x N planar array by translation, each first-level planar sub-array is represented by the serial number (m', n') m' = 0, 1, …, M2-1, n' = 0, 1, …, N2-1, wherein n' and m' represent the serial numbers of the rows and columns of each first-level planar sub-array, respectively, and the frequency domain representation of the (m, n)th microphone element in the first-level planar sub-array (m', n') is X (m'+m)(n'+n) , where m = 0, 1, …, M1-1, n = 0, 1, …, N1-1.

[0042] 3.2) Calculate the Chebyshev coefficients of each first-level planar sub-array in the x-axis direction dimension, according to the design target, the x-axis main lobe width is Calculate the approximate intermediate value Obtain the Chebyshev weighted beam zero position of each first-level planar sub-array:

[0043]

[0044] Form a beam zero position steering vector matrix:

[0045]

[0046] Where each row in the matrix represents a steering vector:

[0047]

[0048] Then the Chebyshev coefficients of each first-level planar sub-array in the x-axis direction can be expressed as Where e1 = [1, 0, …, 0] T .

[0049] 3.3) Using the same method in step 3.2), according to the y-axis main lobe width in the design target Calculate the Chebyshev coefficients h of each first-level planar sub-array in the y-axis direction dimension (y) :

[0050] Calculate the approximate intermediate value Obtain the beam zero position of each first-level planar sub-array in the y-axis direction dimension:

[0051]

[0052] Form a beam zero position steering vector matrix:

[0053]

[0054] where each column of the matrix represents a steering vector:

[0055]

[0056] The Chebyshev coefficients of each first-level planar subarray in the y-axis direction can be expressed as

[0057] 3.4) The Chebyshev coefficient vector h of the first-level planar subarray can be constructed by h (x) and h (y) , where the elements in vector h are m = 0, 1, …, M1-1, n = 0, 1, …, N1-1, where and are the elements in vector h (x) and h (y) , respectively. The two-dimensional Chebyshev windowed beamformer weight vector w1 of each first-level planar subarray is:

[0058] w1 = diag{h}d1(θ s ,φ s ) = Λd1(θ s ,φ s ) (10)

[0059] where θ s = 10° and φ s = 180° are the elevation angle and the horizontal angle of the target sound source s, d1(θ s ,φ s ) is calculated according to formula (5), and Λ = diag{h} represents a diagonal matrix with the elements of vector h on the diagonal. The Chebyshev windowed beamforming can effectively control the sidelobes while keeping the beamwidth unchanged in the frequency domain.

[0060] 3.5) Spatial filtering is performed on the frequency domain signal of each first-level planar subarray using the corresponding two-dimensional Chebyshev windowed beamforming weight vector, and the output signal of each first-level planar subarray (m', n') is:

[0061]

[0062] where is the corresponding element in the weight vector , and (·) * represents the conjugate of the elements in the parentheses.

[0063] Step 4, the output signals of each first-level planar subarray are combined to form a second-level planar subarray, a linearly constrained minimum variance beamformer with null broadening is constructed, adaptive spatial filtering processing is performed, target direction signal is enhanced, and a wide null is formed in the interference direction to suppress. The specific process is as follows:

[0064] 4.1) The output signals of the first-level planar subarray are combined to form a second-level planar subarray, and the second-level planar subarray is represented as m' = 0, 1, …, M2-1, n' = 0, 1, …, N2-1

[0065] 4.2) The signal incident direction (θ j' ,φ j' ) of the interference sound source is broadened to obtain the possible incident area Θ j' of the signal of the interference sound source, Θ j' is discretized, and then the discretized interference steering vector d2(θ j' ,φ j' ) of the incident area of the signal of the interference sound source is obtained, j' = 1, 2, …, J, wherein J represents the number of discrete interference angles, and d2(θ j' ,φ j' ) is calculated according to formula (6); and the correlation matrix of the discretized interference steering vector is constructed as:

[0066]

[0067] 4.3) Eigenvalue decomposition is performed on the correlation matrix C to obtain λ1≥λ2≥…≥λ J , wherein λ k is a characteristic value of the decomposition, v k is a corresponding eigenvector, a larger eigenvector is selected as a basis vector to form a projection matrix, a parameter K is traversed from 1 to J, and whether the following inequality is satisfied is determined:

[0068]

[0069] wherein ε is a set constant, and in the embodiment, ε = 0.01; if the inequality is satisfied, the traversal process is stopped, and K is determined as the number of selected large eigenvalues.

[0070] 4.4) The eigenvectors corresponding to the first K large eigenvalues are selected to form a projection matrix The projection matrix is used for preprocessing the second-level planar subarray to obtain a projected subarray signal The covariance matrix of the projected subarray signal is:

[0071]

[0072] Where R is the covariance matrix of the second-level planar subarray y1, and N is the number of snapshots for calculating the signal covariance matrix.

[0073] 4.5) The covariance matrix of the projected subarray signal obtained by step 4.4) and the known second-level planar subarray target signal steering vector d2(θ s ,φ s ), the second-level planar subarray null broadening linear constraint minimum variance LCMV adaptive weight vector is calculated as:

[0074]

[0075] Where d2(θ s ,φ s ) is obtained by formula (6), represents the inverse of the covariance matrix of the projected subarray signal.

[0076] 4.6) As shown in Figure 3 , the second-level planar subarray is weighted and summed to obtain the final beamforming output frequency domain signal

[0077] Step 5, the final beamforming output frequency domain signal calculated in step 4 is inverse Fourier transformed to restore each frame of time domain signal. At the same time, it is continuously determined whether the target sound source and the interference sound source of the beam pointing change in the direction of arrival. If it does not change, the weight coefficient does not need to be updated, and steps 1-4 are repeatedly executed to output the speech enhancement signal; if it has changed, only the subarray beamforming weight coefficient needs to be updated according to steps 3 and 4, and then a new speech enhancement signal is output.

[0078] Embodiment 2

[0079] This embodiment provides a uniform planar array constant beamwidth linear constraint minimum variance beamforming device for implementing the method of embodiment 1, as shown in Figure 4 , comprising a planar array in which the microphone elements are uniformly distributed, a first-level planar subarray signal processing module, a second-level planar subarray signal processing module, and a signal conversion processor.

[0080] The planar array is used to collect sound signals in the environment and convert each frame of sound signal into a frequency domain signal;

[0081] The planar array is divided into a first-level planar subarray and a second-level planar subarray, the first-level planar subarray signal processing module is used to perform beamforming on the first-level planar subarray, and the second-level planar subarray signal processing module is used to process the frequency domain signal output by the first-level planar subarray signal processing module and output the final beamforming output frequency domain signal;

[0082] The signal conversion processor is configured to convert the beamforming output frequency domain signal into a time-frequency signal.

[0083] For those skilled in the art, various corresponding changes and modifications can be made according to the above technical solutions and concepts, and all these changes and modifications should be included in the protection scope of the present application.

Claims

1. A uniform array constant beamwidth linearly constrained minimum variance beamforming device, characterized in that, It includes a planar array with uniformly distributed microphone elements, a first-level planar subarray signal processing module, a second-level planar subarray signal processing module, and a signal conversion processor; The planar array is used to acquire sound signals from the environment and convert each frame of sound signal into a frequency domain signal; The planar array is divided into multiple first-level planar sub-arrays. The first-level planar sub-array signal processing module is used to perform beamforming on each first-level planar sub-array. The second-level planar sub-array signal processing module is used to perform beamforming on the frequency domain signal output by the first-level planar sub-array signal processing module to obtain the final beamforming output frequency domain signal. The signal conversion processor is used to convert the beamforming output frequency domain signal into a time-frequency signal.

2. The apparatus according to claim 1, characterized in that, The microphone array element is an omnidirectional microphone.

Images