CN116611223A - Accurate array response control method and device combined with white noise gain constraint - Google Patents
Accurate array response control method and device combined with white noise gain constraint Download PDFInfo
- Publication number
- CN116611223A CN116611223A CN202310495477.0A CN202310495477A CN116611223A CN 116611223 A CN116611223 A CN 116611223A CN 202310495477 A CN202310495477 A CN 202310495477A CN 116611223 A CN116611223 A CN 116611223A
- Authority
- CN
- China
- Prior art keywords
- white noise
- noise gain
- constraint condition
- matrix
- filter coefficient
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 230000004044 response Effects 0.000 title claims abstract description 82
- 238000000034 method Methods 0.000 title claims abstract description 78
- 238000004364 calculation method Methods 0.000 claims abstract description 19
- 239000011159 matrix material Substances 0.000 claims description 80
- 239000013598 vector Substances 0.000 claims description 26
- 238000000354 decomposition reaction Methods 0.000 claims description 9
- 238000005070 sampling Methods 0.000 claims description 3
- 230000017105 transposition Effects 0.000 claims description 2
- 238000013461 design Methods 0.000 abstract description 55
- 238000002474 experimental method Methods 0.000 description 12
- 238000012545 processing Methods 0.000 description 9
- 230000003044 adaptive effect Effects 0.000 description 7
- 230000006872 improvement Effects 0.000 description 5
- 238000004088 simulation Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 4
- 238000005457 optimization Methods 0.000 description 4
- 230000015572 biosynthetic process Effects 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 230000007246 mechanism Effects 0.000 description 3
- 238000003786 synthesis reaction Methods 0.000 description 3
- NAWXUBYGYWOOIX-SFHVURJKSA-N (2s)-2-[[4-[2-(2,4-diaminoquinazolin-6-yl)ethyl]benzoyl]amino]-4-methylidenepentanedioic acid Chemical compound C1=CC2=NC(N)=NC(N)=C2C=C1CCC1=CC=C(C(=O)N[C@@H](CC(=C)C(O)=O)C(O)=O)C=C1 NAWXUBYGYWOOIX-SFHVURJKSA-N 0.000 description 2
- 238000004590 computer program Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 230000001629 suppression Effects 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- KLDZYURQCUYZBL-UHFFFAOYSA-N 2-[3-[(2-hydroxyphenyl)methylideneamino]propyliminomethyl]phenol Chemical compound OC1=CC=CC=C1C=NCCCN=CC1=CC=CC=C1O KLDZYURQCUYZBL-UHFFFAOYSA-N 0.000 description 1
- 241000712899 Lymphocytic choriomeningitis mammarenavirus Species 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 201000001098 delayed sleep phase syndrome Diseases 0.000 description 1
- 208000033921 delayed sleep phase type circadian rhythm sleep disease Diseases 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 238000009533 lab test Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03H—IMPEDANCE NETWORKS, e.g. RESONANT CIRCUITS; RESONATORS
- H03H17/00—Networks using digital techniques
- H03H17/0063—R, L, C, simulating networks
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03H—IMPEDANCE NETWORKS, e.g. RESONANT CIRCUITS; RESONATORS
- H03H17/00—Networks using digital techniques
- H03H17/02—Frequency selective networks
- H03H17/06—Non-recursive filters
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/04—Constraint-based CAD
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03H—IMPEDANCE NETWORKS, e.g. RESONANT CIRCUITS; RESONATORS
- H03H17/00—Networks using digital techniques
- H03H2017/0072—Theoretical filter design
- H03H2017/0081—Theoretical filter design of FIR filters
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Abstract
The application provides a method and a device for controlling accurate array response by combining white noise gain constraint, wherein the method is suitable for a time domain beam former, and comprises the following steps: initializing a filter coefficient, and setting an initial white noise gain constraint condition and a sidelobe level constraint condition; selecting a position to be controlled in the wave beam pattern, adding a virtual interference source at a corresponding frequency and angle, updating a filter coefficient, calculating white noise gain of the wave beam former, and if the white noise gain constraint condition is not met, modifying a side lobe level constraint condition to update the filter coefficient again until the filter coefficient is obtained through calculation and meets the white noise gain constraint condition and the side lobe level constraint condition at the same time or stops iteration after the set iteration times are reached, so that the final wave beam pattern is obtained. The application has the advantages that: any array microphone array that can be used in a time domain structure; the method realizes the accurate control of the response of the designated position in the beam pattern and has flexible beam design performance.
Description
Technical Field
The application belongs to the field of acoustic signal processing, and particularly relates to a method and a device for controlling accurate array response by combining white noise gain constraint.
Background
The microphone array is composed of a plurality of microphones which are arranged according to a certain rule, and is widely applied to the field of acoustic signal processing, such as voice interaction, teleconferencing and the like. The microphone array can extract the expected signals in specific directions, simultaneously restrain interference and environmental noise in different directions, and has excellent spatial filtering performance. Because of the wide frequency band of the voice signal, wideband beamforming techniques are required for processing. In the time domain and frequency domain implementation method of broadband beam forming, the time domain method adopts a structure of summation after filtering, and is often applied to scenes requiring low processing time delay. The beamformers can be divided into adaptive beamformers and fixed beamformers depending on whether the filter coefficients depend on the received data. Typical adaptive beamformers include linear constrained minimum variance beamformers (Linearly Constrained Minimum Variance, LCMV) and least mean square undistorted response (Minimum Variance Distortionless Response, MVDR) beamformers, among others. The fixed beamformers include Delay-and-Sum (DAS) beamformers, super-directive (SD) beamformers, and the like. In practical applications, the adaptive beamformer has a large calculation amount and may damage the target signal when there is an error, so in application scenarios requiring low calculation amount and high robustness, the fixed beamformer is often used more. For a fixed beamformer, it is crucial how the filter coefficients are designed to obtain the ideal beam pattern.
The array beam design problem is also known as an array beam pattern synthesis problem. Over the past few decades, a great deal of research has been devoted to array beam pattern synthesis. Since 1990, many students selected the null mechanism introduced into the adaptive beamforming method for the design of narrowband beamformers. The method of notch noise is proposed by Olen and Compton, however, the method adopts an iterative mode when selecting virtual interference power, and the design efficiency is low. For narrow-band beam design, zhang Xuejing and other scholars propose an optimal accurate array response control algorithm (Optimal Precise Array Response Control, OPARC), and accurate control of the array response can be achieved by iteratively updating a narrow-band virtual interference and noise covariance matrix (Virtual Interference-plus-Noise Covariance Matrix, VINCM). However, the above method is applicable to narrowband signals only, and cannot be applied to wideband beamformers with time domain structures. For wideband beamformers, yan Shefeng, etc. of the time-domain filtered and summed structure, it is proposed to convert the array beam pattern synthesis problem into a second order cone planning problem (Second Order Cone Programming, SOCP) and solve to obtain the optimal filter coefficients by using a toolkit based on the Interior-Point Method (IPM), such as CVX. However, the beam design method based on the optimization algorithm often uses a tool kit based on the interior point method to solve, the calculation time is long, and the situation that no solution exists can occur under the unreasonable constraint condition. Meanwhile, a beam design method based on an optimization algorithm presets a desired beam pattern, and a Side Lobe Level (SLL) and a White Noise Gain (WNG) of the desired beam pattern need to be determined through multiple pre-experiments. Therefore, how to design a broadband beam without pre-experiments is a problem to be studied.
Disclosure of Invention
The application aims to overcome the defect that the prior art needs to perform multiple pre-experiments when performing broadband array beam design.
In order to achieve the above object, the present application proposes an accurate array response control method in combination with white noise gain constraint, which is applicable to a time domain beamformer, which is followed by a set of finite impulse response filters for each microphone;
the method comprises the following steps:
step S1: initializing a filter coefficient, and setting an initial white noise gain constraint condition and a sidelobe level constraint condition;
step S2: selecting a position to be controlled in a beam pattern, adding a virtual interference source at a corresponding frequency and angle, updating a filter coefficient, calculating white noise gain of the beam former, and if the white noise gain constraint condition is not met, modifying a side lobe level constraint condition to update the filter coefficient again until the filter coefficient is obtained through calculation and meets the white noise gain constraint condition and the side lobe level constraint condition or stops iteration after the set iteration times are reached, so as to obtain a final filter coefficient.
As an improvement of the above method, the initializing the filter coefficient specifically includes:
constructing an initial broadband virtual interference and noise covariance matrix as a unit matrix; calculating initial filter coefficient h 0 ：
wherein ,representing the target direction constraint matrix, θ 0 Indicates the direction of sound source +.>Representing the operation of taking the real part,/->Representing an imaginary part taking operation, f k K=1, …, K represents the frequency corresponding to the kth frequency point to be controlled, and K represents the number of frequencies to be controlled; t represents matrix transposition; u (f) K ,θ 0 ) Representing the position in the direction theta 0 With frequency f K Steering vector of the time domain beamformer:
wherein ,represents the kronecker product; e (f) k ) Representing an L-dimensional fourier transform factor, L representing the order of the finite impulse response filter; a (f) k ,θ 0 ) M represents the frequency domain guide vector of M dimension, M represents the number of microphones;
e(f k )＝[1,exp(-j2πf k /f s ),...,exp(-j2π(L-1)f k /f s )] T
a(f k ,θ 0 )＝exp[-j2πf k sin(θ 0 )d/c]
wherein the M-dimensional vector d contains the distance between each microphone and the reference position, c represents the speed of sound, f s Representing the sampling frequency;
g＝[1,...,1,0,...,0]consists of K1 and K0;representing ML x ML identity matrix.
As an improvement of the above method, the positions in the selected beam pattern that need to be controlled are specifically:
wherein ,fj ，θ j The frequency and the angle of the jth iteration control point are respectively; b is a control frequency band; Θ is the control angle range; p (P) d (f, θ) is the desired array response;
normalized array power response P j (f, θ) is:
wherein ,hj Representing the filter coefficients at the jth iteration; θ 0 Representing the direction of the sound source; f (f) 0 Representing the normalized reference frequency.
As an improvement of the above method, the updating of the filter coefficients is specifically:
h j ＝H j [(S j ) 4 ,(S j ) 3 ,(S j ) 2 ,S j ,1] T
wherein ,Sj The power representing the virtual interference is calculated by the following formula:
[(S j ) 4 ,(S j ) 3 ,(S j ) 2 ,S j ,1]G j [(S j ) 4 ,(S j ) 3 ,(S j ) 2 ,S j ,1] T ＝0
wherein the coefficient matrixThe superscript H represents the conjugate transpose of the matrix; p (P) d (f j ,θ j ) Representing the expected array response; matrix H j ∈R ML×5 As a known real coefficient matrix, the unknown coefficient S is extracted by substituting the formula (A1) and the formula (A2) into the formula (A3) j And (3) calculating to obtain:
wherein ,represents the jth iterationThe covariance matrix of the virtual interference is calculated as:
(Vector)a first column and a second column vector of the ML x 2-dimensional matrix on the right side of the equation, respectively; by matrix of 2 x 2 dimensionsPerforming feature decomposition to obtain a feature value lambda j1 、λ j2 ，E j An ML x 2-dimensional matrix composed of feature vectors;
(Vector)a first column and a second column vector of the ML x 2-dimensional matrix on the right side of the equation, respectively; by matrix of 2 x 2 dimensionsPerforming feature decomposition to obtain a feature value gamma j1 、γ j2 ，F j A matrix of feature vectors.
As an improvement of the above method, the calculating the white noise gain of the beam former modifies the side lobe level constraint condition to update the filter coefficient if the white noise gain constraint condition is not satisfied, until the calculated filter coefficient satisfies the white noise gain constraint condition and the side lobe level constraint condition or reaches the set iteration number, and then stops iterating specifically:
when the white noise gain of the obtained filter coefficient is smaller than the set value, if the array response P is expected d (f j ,θ j ) If the set value is smaller than the set value, setting the expected array response to be increased by 1dB, and updating the filter coefficients again;
if the number of iterations is equal to the set point, or the array response P is expected d (f j ,θ j ) Equal toSet point and side lobe level of current beam pattern is lower than expected array response P d (f j ,θ j ) The iteration is exited and the final output is obtained;
otherwise, step S2 is re-executed.
The application also provides a precise array response control device combined with white noise gain constraint, which is realized based on the method, and comprises a time domain beam former and a precise array response control system;
the time domain beamformer is followed by a set of finite impulse response filters after each microphone;
the precision array response control system includes:
the initialization module is used for initializing the filter coefficient and setting an initial white noise gain constraint condition and a sidelobe level constraint condition; and
the iterative calculation module is used for selecting a position to be controlled in the wave beam pattern, adding a virtual interference source at a corresponding frequency and angle, updating a filter coefficient, calculating white noise gain of the wave beam former, and if the white noise gain constraint condition is not met, modifying a side lobe level constraint condition to update the filter coefficient again until the filter coefficient is obtained through calculation and meets the white noise gain constraint condition and the side lobe level constraint condition or stops iteration after the set iteration times are reached, so that the final filter coefficient is obtained.
Compared with the prior art, the application has the advantages that:
the application provides a novel accurate array response control method combined with white noise gain constraint, which can be used for any array microphone array with a time domain structure. The method introduces a null design mechanism of a narrow-band self-adaptive beam forming method into broadband beam design, and realizes accurate control of the response of a designated position of a beam pattern by constructing a broadband virtual interference and noise covariance matrix (Virtual Interference-plus-Noise Covariance Matrix, VINCM). When updating the filter coefficient, the method adjusts the expected wave beam according to White Noise Gain (WNG), and has flexible wave beam design performance. After the VINCM is updated for a plurality of times, a broadband array beam pattern meeting the requirements can be designed. In the practical experiment of a semi-anechoic room, the designed beam forming device has better interference suppression performance compared with the traditional fixed beam forming device, and has important application value.
Drawings
Fig. 1 is a schematic diagram of a time domain beamformer architecture;
FIG. 2 is a flow chart of a method of accurate array response control incorporating white noise gain constraints;
FIG. 3 shows an initial beam pattern for a uniform sidelobe beam design;
fig. 4 shows a beam pattern after 450 iterations of a uniform side lobe beam design;
FIG. 5 is a diagram of the initial expected response of a uniform side lobe beam design;
FIG. 6 is a graph of expected response after an iteration of a uniform side lobe beam design;
FIG. 7 shows a-60 null beam design;
FIG. 8 shows a non-uniform side lobe beam design;
FIG. 9 shows a beam pattern under the SLLmax-30 dB constraint;
FIG. 10 shows a beam pattern under the SLLmax-20 dB constraint;
FIG. 11 shows a beam pattern under the SLLmax-50 dB constraint;
FIG. 12 shows a beam pattern under the SLLmax-40 dB constraint;
fig. 13 shows WNG contrast for the beamformer under different sidelobe constraints.
Detailed Description
The technical scheme of the application is described in detail below with reference to the accompanying drawings.
In order to realize broadband array beam design without pre-experiment, the application provides a precise array response control method (WNG-Constrained Precise Array Response Control, WCPARC) and a device combined with white noise gain constraint. Firstly, the method is suitable for any array type array, can realize accurate control of specific position response in each iteration by constructing broadband VINCM, and obtains a beam former meeting the requirement through repeated iteration design. In the iterative process, the white noise gain of the current beam former is calculated, and the expected response value is correspondingly adjusted, so that the white noise gain is controlled. The method of the application does not need to adopt a pre-experiment to determine the expected beam pattern, and has flexible beam design performance.
1. Signal model
The accurate array response control method and device combined with white noise gain constraint are suitable for a time domain beam former adopting a filtered summation structure. The time domain beamformer is followed by a set of finite impulse response filters (Finite Impulse Response, FIR) after each microphone, the structure of which is shown in fig. 1. Let the array be formed by M microphones arranged linearly, the received signal of the mth microphone can be expressed as:
where N is the number of sound sources, s i (. Cndot.) is the ith sound source signal, θ i For the incoming wave direction of the ith sound source, τ m (θ i ) For propagation delay of the ith sound source signal to the mth microphone, n m (t) is spatially uncorrelated white noise received by the mth microphone. Each microphone is followed by an L-order FIR filter, and the first-order coefficient of the FIR filter corresponding to the mth microphone is set as h m,l Then the overall filter coefficient consisting of M times L coefficientsThe method comprises the following steps:
h＝[h 1,1 ,h 2,1 ,...,h M,1 ,...,h 1,L ,h 2,L ,…,h M,L ] T (2)
wherein ,(·)T Representing the transpose operation.
For far field signals at a frequency f in the direction θ, the steering vector of the time domain beamformer can be expressed as wherein />Representing the kronecker product, the L-dimensional fourier transform factor e (f) and the M-dimensional frequency domain guide vector a (f, θ) can be expressed as
Wherein, the M-dimensional vector d represents the distance between the microphone and the reference position, c is the sound velocity, f s Is the sampling frequency. Let sound source direction be theta 0 Normalized reference frequency f 0 The normalized array power response is:
in calculating the interference and noise covariance matrix, matrices of different value ranges are obtained in narrowband and wideband applications. The interference and noise covariance matrix used in the narrowband scenario is a complex matrix, while the real matrix is used in the wideband time domain beamformer. First, a noise covariance matrix is calculated. Assuming that only spatially uncorrelated white noise is present, the noise covariance matrix may be expressed asWherein the subscript n represents noise, ">Is the noise power, I ML Is an ml×ml dimensional identity matrix (ML represents mxl). And secondly, calculating an interference covariance matrix. By utilizing the fourier transform property, the wideband signal can be considered to have symmetrical power at both the positive and negative frequencies, resulting in a real matrix. Let the interference direction set be Θ I Subscript I represents interference (interference) with an interference signal band of [ f l ,f u ]S (f, θ) is the interference power corresponding to the frequency and angle, so that the interference and the angle are simplifiedThe noise covariance matrix is
wherein ,representing the operation of taking the real part,/->Representing an imaginary part taking operation.
In order to minimize the interference and noise output power under the condition of ensuring that the target direction signal is not distorted, an optimization problem is set as follows:
wherein ,constraining the matrix for the target direction, f k (k=1, …, K) is the frequency corresponding to the K-th frequency point to be controlled, K is the number of frequencies to be controlled, g= [1, …,1,0, …,0]Consists of K1 s and K0 s. The closed-form solution of the optimization problem (6) is obtained by the Lagrangian multiplier method, and is as follows:
h＝(R n+I ) -1 C T (θ 0 )[C(θ 0 )(R n+I ) -1 C T (θ 0 )] -1 g (7)
the application adopts the formula (7) as a calculation mode of the beam forming device. Wherein matrix R n+I Using a virtual interference and noise covariance matrix R v Instead, and construct R in an iterative manner v To obtain h= (R) v ) -1 C T (θ 0 )[C(θ 0 )(R v ) -1 C T (θ 0 )] - 1 g。
2. Beam design method based on array response accurate control
In order to realize broadband beam design, the application introduces the design idea of nulling in the adaptive beam forming method. And setting virtual interference source distribution by constructing VINCM, and adjusting the beam pattern by using a mechanism for forming nulls by an adaptive method. When updating the VINCM, each iteration enables precise control of the location-specific array response and adjusts the desired response based on WNG. And finally obtaining the beam forming device meeting the design requirement of the beam pattern through multiple iterations.
The method of the application is divided into an initialization stage and an iterative update stage:
in the initialization stage, the initial filter coefficient is calculated by adopting the method (7), and the filter coefficient is recorded as h 0 Wherein initiallyIs an identity matrix.
In the iterative updating stage, the updating process of the weighting coefficients is divided into three steps: the first step: and selecting a position to be controlled in the beam pattern, updating a filter coefficient h in the second step, and adjusting the expected response according to WNG in the third step. Specific:
the first step: selecting a control position in the frequency-angle two-dimensional beam pattern, namely adding a single-frequency virtual interference source at a corresponding angle;
and a second step of: calculating the power of the virtual interference source according to the expected response value, and updating the filter coefficient;
and a third step of: and if the white noise gain of the obtained beam former is lower than the constraint value, the expected response value is adjusted upwards, and the second step is carried out again. Through multiple iterations, filter coefficients that meet the desired response requirements can be obtained. A flow chart of the method of the present application is shown in fig. 2.
The broadband array beam design method provided by the application has the following advantages: firstly, because the method can realize the accurate control of the response of any position in the beam pattern, when the expected beam pattern changes, the beam pattern can be quickly adjusted on the basis of the original beam pattern; meanwhile, the method dynamically adjusts the expected response value based on the white noise gain constraint, and an expected beam pattern is not required to be determined through a pre-experiment, so that the efficiency of beam design is improved.
(1) Selection of control points
The method of the application adopts a mode of updating the VINCM to carry out beam control, and adds single-frequency interference with a specific angle for each update. The maximum peak point in the sidelobe control region is taken as the control position of the next iteration when the interference is added, and the adjustment of the peak point effectively controls surrounding beams. In the j-th iteration, for the current filter coefficient h j-1 First, the beam pattern P is calculated by the equation (4) j (f, θ) and find the maximum peak point in the control region. The calculation of the control point can be expressed as:
wherein ,fj ，θ j The frequency and the angle of the jth iteration control point are respectively, B is a control frequency band, Θ is a control angle range, and P d To expect an array response.
(2) Updating of filter coefficients
After the next control position is selected by the formula (8), the method adds a virtual interference source at the corresponding frequency and angle, and adjusts the corresponding response to the expected value. When updating the filter coefficients according to the desired response, the following three steps are taken:
step 1: pair matrixAnd updating. As can be seen from the analysis of formula (5), the above step is taken care of>Adding control points (f) j ,θ j ) Covariance matrix of virtual interference, namely:
wherein ,Sj Is an unknown virtual interference power. Since the closed-form solution (7) of the filter coefficients contains the matrix of ML x ML dimensionsAnd the computational complexity required for matrix inversion is O (M 3 L 3 ) Larger computing resources are consumed. The method of the application decomposes the ML X ML dimension matrix by matrix inversion theorem, can simplify complex matrix inversion into one-time eigenvalue decomposition of the two-dimensional matrix and multiplication and addition operation of the matrix, and greatly reduces the calculation complexity. After simplification by matrix inversion theorem, the matrix is +.>The calculation of (2) can be expressed as:
wherein the vector isThe first column and the second column vector of the ML x 2-dimensional matrix on the right side of the equation, respectively. By the above simplified operation, the update matrix can be +.>Is calculated from the complexity of O (M 3 L 3 ) Reduced to O (M) 2 L 2 ). At the same time, by matrix of 2 x 2 dimensionsPerforming feature decomposition to obtain a feature value lambda j1 、λ j2 ，E j An ML x 2-dimensional matrix of feature vectors.
Step 2: pair matrixProceeding withUpdating. The closed-form solution (7) of the filter coefficients contains not only the matrix +.>There is also a +.2K-dimensional matrix +.>And the matrix inversion theorem can be used for simplification. />The update of (c) can be expressed as:
wherein ,λ j1 、λ j2 the characteristic value calculated according to the formula (10) in the step 1. By matrix of 2 x 2 dimensionsPerforming characteristic decomposition to obtain a characteristic value gamma j1 、γ j2 ，F j A matrix of feature vectors.
Step 3: for the filter coefficient h j And updating. After updating by the formulas (10) and (11), the filter coefficient h j The updated version of (c) can be expressed as:
h j ＝H j [(S j ) 4 ,(S j ) 3 ,(S j ) 2 ,S j ,1] T (12)
wherein matrix H j ∈R ML×5 As a known real coefficient matrix, the unknown coefficient S is extracted by substituting the formula (10) and the formula (11) into the formula (7) j And (5) calculating to obtain the product. By means of (12), the updating of the filter coefficients is converted into a power S for solving the virtual interference j . The depth of the nulls formed by the adaptive method followsThe increase in interference power increases monotonically, so that the virtual interference power S can be determined by the desired null depth j . Assuming that the expected response value is ρ j Substituting (12) into equation (4) and letting the left of the equation equal the expected response P d (f j ,θ j ) The equation can be reduced to:
[(S j ) 4 ,(S j ) 3 ,(S j ) 2 ,S j ,1]G j [(S j ) 4 ,(S j ) 3 ,(S j ) 2 ,S j ,1] T ＝0 (13)
wherein the coefficient matrixThe superscript H represents the conjugate transpose of the matrix. Since equation (13) is not closed-form solved, it needs to be converted into a lower-order equation for solving. Coefficient matrix G with rank 2 by eigen decomposition j Equation (13) can be reduced to two unitary four-way equations for solving:
wherein ,is a five-dimensional matrix G j Characteristic value decomposition of Q j Matrix formed by eigenvectors, Q j,1 And Q is equal to j,2 Feature vectors of the first and second columns, respectively, ψ j Diagonal matrix for eigenvalues and only ψ j (1, 1) and ψ j (2, 2) is a non-zero element.
Considering a typical unitary fourth-order equation, the closed-form solution can be found as follows:
ax 4 +bx 3 +cx 2 +dx+e＝0
wherein, coefficient a is not equal to 0, and four closed solutions of the equation are respectively expressed as:
wherein ,
wherein ,ξ 0 ＝c 2 -3bd+12ae,ξ 1 ＝2c 3 -9bcd+27b 2 e+27ad 2 -72ace. Because the interference power is real and unique, the unique real solution in the obtained solution is the interference power value to be obtained. Obtaining interference power S j The filter coefficients can then be updated by equation (12).
(3) Adjusting the expected response based on white noise gain
Solving to obtain h j The desired response is then adjusted according to its white noise gain. For the beam former, the white noise gain reflects the robustness of the beam former under the influence of factors such as array element position errors, channel amplitude and phase errors and the like. In engineering applications, it is often necessary to set the white noise gain of the beamformer to be not less than a specific value in order to ensure its performance in a practical environment. Because the side lobe of the beam former affects the white noise gain, the method of the application proposes to restrict the white noise gain in the beam design so as to realize the simultaneous control of the side lobe level and the white noise gain. The method comprises the following specific steps:
step 1: setting a white noise gain constraint condition and a side lobe level constraint condition, wherein the white noise gain WNG is smaller than a preset value W 0 And side lobe level P d (f j ,θ j ) The smaller the better the P can be set d (f j ,θ j )≤SLL max ，SLL max A lower value may be empirically set;
step 2: solving by using an updating method of a filter coefficient through P d (f j ,θ j )≤SLL max Is to design a corresponding filter h j And calculates the corresponding white noise gain WNG at the moment, and satisfies the constraint condition WNG not more than W 0 If the design requirement is met, the calculation can be stopped;
step 3: if WNG in step 2 does not meet the constraint, the sidelobe constraint P can be adjusted d (f j ,θ j )≤SLL max I.e. to increase the constraint value SLL appropriately max The constraint value is adjusted up as follows:
P d (f j ,θ j )＝[P d (f j ,θ j )+1](dB) (16)
and recalculate the filter coefficients and the corresponding white noise gains WNG, if the constraint condition WNG is less than or equal to W 0 If not, the design is stopped, and if not, the constraint value SLL is continuously increased max Repeating the steps until h is calculated j And simultaneously, the white noise gain constraint condition and the side lobe level constraint condition are met, or iteration is stopped after the set iteration times are reached.
(4) Beam pattern design simulation
In order to verify the effectiveness of the method provided by the application, a 10-element uniform linear array with an array element spacing of 2.8cm is taken as an example, and a beam design simulation experiment is carried out. Firstly, carrying out beam design simulation under different design requirements, respectively considering beam designs of uniform side lobes, nulls and non-uniform side lobes, and comparing an initial state in the beam design of the uniform side lobes with an expected beam pattern after convergence is completed. And then, comparing white noise gains of the obtained wave beam forming device under different maximum side lobe level constraint, and analyzing the influence of the maximum side lobe level constraint value on the wave beam design effect.
Beam design effect under different desired beam pattern conditions:
firstly, considering the design of uniform side lobe wave beams, the design frequency band is [2000Hz,8000Hz]The expected direction is 0 DEG, WNG constraint is 9.7dB, the initial expected side lobe level is set to be-50 dB, and the maximum side lobe level SLL max Is-30 dB. When WCPARC is adopted for beam design, an initial beam pattern and a beam pattern after 450 iterations are shown in fig. 3 and 4, and a side lobe of the designed beam pattern is lower than-30 dB. Comparison of initial expected response with post-iteration expected response as shown in fig. 5 and 6, since WCPARC adjusts the expected response according to WNG during the iteration, the post-iteration expected response changes from the initial expected response.
In null beam design simulation, the desired direction is set to 0 °, the design band is [2500hz,8000hz ], and the WNG constraint is 0dB. The angle of the nulls is expected to be 50 deg., and the depth-50 dB. The beam pattern obtained by the WCPARC method is shown in figure 7, and the desired sidelobe level of-30 dB is still maintained at [2500Hz,8000Hz ] while the null design requirements are met.
Then, a non-uniform sidelobe beam design was performed with a desired direction of 10 °, a design band of [2000hz,5500hz ], and a WNG constraint of 0dB. The resulting beam pattern is shown in fig. 8, where the desired side lobes are marked with dashed lines. The simulation experiment result of the beam design shows that the method can effectively realize the broadband beam design of the time domain beam forming device under different requirements.
White noise gain contrast obtained under different maximum sidelobe level constraints:
in order to analyze the influence of sidelobe level constraint values on the wave beam design effect of the WCPARC method, different constraint values SLL are adopted max And (5) carrying out uniform sidelobe beam design under the condition. Setting WNG constraint to 9.0dB, consider SLL max The beam patterns of the corresponding beam formers and DAS beam formers are obtained by { -50, -40, -30, -20} dB in sequence, as shown in figures 9-12, and white noise gain pairs such as shown in figure 13.
The beam pattern results of FIGS. 9-12 show that, at different SLLs max Under the condition, the method of the applicationA beam pattern can be designed to meet the desired value. Meanwhile, different SLLs in FIG. 13 max Different white noise gains are obtained, and the white noise gain can be seen to follow SLL max And increases with increasing size. When SLL max Up to-20 dB, it can be seen that the noise gain now satisfies the constraint that globally is greater than 9 dB. The result shows that the method can control the side lobe level and the white noise gain simultaneously, and different wave beam design effects are realized by adjusting the side lobe level constraint value.
(5) Experiment verification
To further verify the effectiveness of the designed beamformer in the actual scenario, actual experiments were performed using the voice data recorded in the semi-anechoic room. The experiment adopts a 10-element uniform linear array with the array element spacing of 2.8cm, and the sound pressure level of environmental noise in a semi-anechoic chamber is about 20dB. The target direction is set to be 0 DEG, the range of the direction of the interference sound source is [ -90 DEG, -50 DEG ] [50 DEG, 90 DEG ] U.S. 50 DEG, and the interval is 10 DEG, and 10 directions are all used. Each experiment sets a target sound source and an interference sound source, and the signal-to-interference ratio is { -10, -5,0,5,10} dB. The distance from the sound source to the array is 3m, and the heights of the center of the sound source and the center of the array are consistent and are 0.9m. Four beamformers were used in the experiment, respectively: the DAS beam forming device, the SD beam forming device and the uniform side lobe beam forming device and the null beam forming device are obtained by adopting WCPARC design.
The experimentally measured average speech quality perception assessment (Perceptual Evaluation of Speech Quality, PESQ) and Short-term objective intelligibility (Short-Time Objective Intelligibility, STOI) in 10 interference directions are shown in table 1. In the two objective indexes, the null wave beam forming device adopting WCPARC obtains the optimal result, which shows that the wave beam forming device obtained by design has good interference suppression performance.
Table 1 objective index improvement contrast at different input signal-to-interference ratios in semi-anechoic laboratory experiments
In summary, the accurate array response control method combining white noise gain constraint provided by the application comprises the following specific steps:
step 1: the beamformer is initialized and the initial VINCM is set as a unit array. Calculating initial filter coefficients h according to (7) 0 Determining an expected response P based on design requirements d (f j ,θ j )；
Step 2: calculating a next control point according to formula (8), calculating an updated version of the filter coefficients using formula (12);
step 3: according to P d (f j ,θ j ) Constructing equation (14) and solving for virtual interference power S j Calculating filter coefficients using equation (12);
step 4: when the white noise gain of the obtained filter coefficient is smaller than the constraint value, two cases are divided: if P d (f j ,θ j ) Less than the maximum value, let P d (f j ,θ j )＝[P d (f j ,θ j )+1](dB) and returns to step 3; if P d (f j ,θ j ) If the maximum value is equal to the maximum value, executing the step 5;
step 5: and updating the matrix using (10)Update +.>
Step 6: if the iteration number is equal to the set value or the sidelobe level of the current beam pattern is lower than the expected value, namely P (f, theta) is less than or equal to P d (f, θ), (f ε B, θ ε Θ), then the iteration is exited and the final output is obtained; otherwise, returning to the step 2.
The application also provides a precise array response control device combined with white noise gain constraint, which is realized based on the method, and comprises a time domain beam former and a precise array response control system;
the time domain beamformer is followed by a set of finite impulse response filters after each microphone;
the precision array response control system includes:
the initialization module is used for initializing the filter coefficient and setting an initial white noise gain constraint condition and a sidelobe level constraint condition;
the iterative calculation module is used for selecting a position to be controlled in the wave beam pattern, adding a virtual interference source at a corresponding frequency and angle, updating a filter coefficient, calculating white noise gain of the wave beam former, and if the white noise gain constraint condition is not met, modifying a side lobe level constraint condition to update the filter coefficient again until the filter coefficient is obtained through calculation and meets the white noise gain constraint condition and the side lobe level constraint condition or stops iteration after the set iteration times are reached, so that the final filter coefficient is obtained.
The present application may also provide a computer apparatus comprising: at least one processor, memory, at least one network interface, and a user interface. The various components in the device are coupled together by a bus system. It will be appreciated that a bus system is used to enable connected communications between these components. The bus system includes a power bus, a control bus, and a status signal bus in addition to the data bus.
The user interface may include, among other things, a display, a keyboard, or a pointing device. Such as a mouse, track ball, touch pad, touch screen, or the like.
It will be appreciated that the memory in the disclosed embodiments of this application can be either volatile memory or nonvolatile memory, or can include both volatile and nonvolatile memory. The nonvolatile Memory may be a Read-Only Memory (ROM), a Programmable ROM (PROM), an Erasable PROM (EPROM), an Electrically Erasable EPROM (EEPROM), or a flash Memory. The volatile memory may be random access memory (Random Access Memory, RAM) which acts as an external cache. By way of example, and not limitation, many forms of RAM are available, such as Static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double Data Rate SDRAM (Double Data Rate SDRAM), enhanced SDRAM (ESDRAM), synchronous DRAM (SLDRAM), and Direct RAM (DRRAM). The memory described herein is intended to comprise, without being limited to, these and any other suitable types of memory.
In some implementations, the memory stores the following elements, executable modules or data structures, or a subset thereof, or an extended set thereof: an operating system and application programs.
The operating system includes various system programs, such as a framework layer, a core library layer, a driving layer, and the like, and is used for realizing various basic services and processing hardware-based tasks. Applications, including various applications such as Media Player (Media Player), browser (Browser), etc., are used to implement various application services. The program implementing the method of the embodiment of the present disclosure may be contained in an application program.
In the above embodiment, the processor may be further configured to call a program or an instruction stored in the memory, specifically, may be a program or an instruction stored in an application program:
the steps of the above method are performed.
The method described above may be applied in a processor or implemented by a processor. The processor may be an integrated circuit chip having signal processing capabilities. In implementation, the steps of the above method may be performed by integrated logic circuits of hardware in a processor or by instructions in the form of software. The processor may be a general purpose processor, a digital signal processor (Digital Signal Processor, DSP), an application specific integrated circuit (Application Specific Integrated Circuit, ASIC), a field programmable gate array (Field Programmable gate array, FPGA) or other programmable logic device, discrete gate or transistor logic device, discrete hardware components. The methods, steps and logic blocks disclosed above may be implemented or performed. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The steps of a method as disclosed above may be embodied directly in hardware for execution by a decoding processor, or in a combination of hardware and software modules in a decoding processor. The software modules may be located in a random access memory, flash memory, read only memory, programmable read only memory, or electrically erasable programmable memory, registers, etc. as well known in the art. The storage medium is located in a memory, and the processor reads the information in the memory and, in combination with its hardware, performs the steps of the above method.
It is to be understood that the embodiments described herein may be implemented in hardware, software, firmware, middleware, microcode, or a combination thereof. For a hardware implementation, the processing units may be implemented within one or more application specific integrated circuits (Application Specific Integrated Circuits, ASIC), digital signal processors (Digital Signal Processing, DSP), digital signal processing devices (DSP devices, DSPD), programmable logic devices (Programmable Logic Device, PLD), field programmable gate arrays (Field-Programmable Gate Array, FPGA), general purpose processors, controllers, microcontrollers, microprocessors, other electronic units designed to perform the functions described herein, or a combination thereof.
For a software implementation, the inventive techniques may be implemented with functional modules (e.g., procedures, functions, and so on) that perform the inventive functions. The software codes may be stored in a memory and executed by a processor. The memory may be implemented within the processor or external to the processor.
The present application may also provide a non-volatile storage medium for storing a computer program. The steps of the above-described method embodiments may be implemented when the computer program is executed by a processor.
Finally, it should be noted that the above embodiments are only for illustrating the technical solution of the present application and are not limiting. Although the present application has been described in detail with reference to the embodiments, it should be understood by those skilled in the art that modifications and equivalents may be made thereto without departing from the spirit and scope of the present application, which is intended to be covered by the appended claims.
Claims (6)
1. A method of accurate array response control incorporating white noise gain constraints, the method being applicable to a time domain beamformer followed by a set of finite impulse response filters;
the method comprises the following steps:
step S1: initializing a filter coefficient, and setting an initial white noise gain constraint condition and a sidelobe level constraint condition;
step S2: selecting a position to be controlled in a beam pattern, adding a virtual interference source at a corresponding frequency and angle, updating a filter coefficient, calculating white noise gain of the beam former, and if the white noise gain constraint condition is not met, modifying a side lobe level constraint condition to update the filter coefficient again until the filter coefficient is obtained through calculation and meets the white noise gain constraint condition and the side lobe level constraint condition or stops iteration after the set iteration times are reached, so as to obtain a final filter coefficient.
2. The method for accurate array response control in combination with white noise gain constraint of claim 1, wherein initializing filter coefficients specifically comprises:
constructing an initial broadband virtual interference and noise covariance matrix as a unit matrix; calculating initial filter coefficient h 0 ：
wherein ,representing the target direction constraint matrix, θ 0 Indicates the direction of sound source +.>Representation fetchReal part operation, < ->Representing an imaginary part taking operation, f k K=1, …, K represents the frequency corresponding to the kth frequency point to be controlled, and K represents the number of frequencies to be controlled; t represents matrix transposition; u (f) K ,θ 0 ) Representing the position in the direction theta 0 With frequency f K Steering vector of the time domain beamformer:
wherein ,represents the kronecker product; e (f) k ) Representing an L-dimensional fourier transform factor, L representing the order of the finite impulse response filter; a (f) k ,θ 0 ) M represents the frequency domain guide vector of M dimension, M represents the number of microphones;
e(f k )＝[1,exp(-j2πf k /f s ),…,exp(-j2π(L-1)f k /f s )] T
a(f k ,θ 0 )＝exp[-j2πf k sin(θ 0 )d/c]
wherein the M-dimensional vector d contains the distance between each microphone and the reference position, c represents the speed of sound, f s Representing the sampling frequency;
g＝[1,…,1,0,…,0]consists of K1 and K0;representing ML x ML identity matrix.
3. The method for accurate array response control in combination with white noise gain constraint according to claim 2, wherein the positions in the selected beam pattern to be controlled are specifically:
wherein ,fj ，θ j The frequency and the angle of the jth iteration control point are respectively; b is a control frequency band; Θ is the control angle range; p (P) d (f, θ) is the desired array response;
normalized array power response P j (f, θ) is:
wherein ,hj Representing the filter coefficients at the jth iteration; θ 0 Representing the direction of the sound source; f (f) 0 Representing the normalized reference frequency.
4. A method of accurate array response control in combination with white noise gain constraint according to claim 3 wherein said updating filter coefficients is specifically:
h j ＝H j [(S j ) 4 ,(S j ) 3 ,(S j ) 2 ,S j ,1] T
wherein ,Sj The power representing the virtual interference is calculated by the following formula:
[(S j ) 4 ,(S j ) 3 ,(S j ) 2 ,S j ,1]G j [(S j ) 4 ,(S j ) 3 ,(S j ) 2 ,S j ,1] T ＝0
wherein the coefficient matrixThe superscript H represents the conjugate transpose of the matrix; p (P) d (f j ,θ j ) Representing the expected array response; matrix H j ∈R ML×5 Is a known real coefficient matrix obtained by substituting the formulas (A1) and (A2)Entering (A3) and extracting unknown coefficients S j And (3) calculating to obtain:
wherein ,the covariance matrix of the virtual interference in the j-th iteration is represented, and the calculation formula is as follows:
(Vector)a first column and a second column vector of the ML x 2-dimensional matrix on the right side of the equation, respectively; by matrix of 2 x 2 dimensionsPerforming feature decomposition to obtain a feature value lambda j1 、λ j2 ，E j An ML x 2-dimensional matrix composed of feature vectors;
(Vector)a first column and a second column vector of the ML x 2-dimensional matrix on the right side of the equation, respectively; by matrix of 2 x 2 dimensionsPerforming feature decomposition to obtain a feature value gamma j1 、γ j2 ，F j A matrix of feature vectors.
5. The method for precise array response control combined with white noise gain constraint according to claim 4, wherein the calculating the white noise gain of the beam former modifies the side lobe level constraint condition to update the filter coefficient if the white noise gain constraint condition is not satisfied, until the filter coefficient is obtained by calculation while satisfying the white noise gain constraint condition and the side lobe level constraint condition or stopping iteration after reaching a set number of iterations, specifically:
when the white noise gain of the obtained filter coefficient is smaller than the set value, if the array response P is expected d (f j ,θ j ) If the set value is smaller than the set value, setting the expected array response to be increased by 1dB, and updating the filter coefficients again;
if the number of iterations is equal to the set point, or the array response P is expected d (f j ,θ j ) Equal to the set point and the sidelobe level of the current beam pattern is lower than the expected array response P d (f j ,θ j ) The iteration is exited and the final output is obtained;
otherwise, step S2 is re-executed.
6. A precision array response control device incorporating white noise gain constraints, implemented on the basis of the method of claims 1-5, characterized in that the device comprises a time domain beamformer and a precision array response control system;
the time domain beamformer is followed by a set of finite impulse response filters after each microphone;
the precision array response control system includes:
the initialization module is used for initializing the filter coefficient and setting an initial white noise gain constraint condition and a sidelobe level constraint condition; and
the iterative calculation module is used for selecting a position to be controlled in the wave beam pattern, adding a virtual interference source at a corresponding frequency and angle, updating a filter coefficient, calculating white noise gain of the wave beam former, and if the white noise gain constraint condition is not met, modifying a side lobe level constraint condition to update the filter coefficient again until the filter coefficient is obtained through calculation and meets the white noise gain constraint condition and the side lobe level constraint condition or stops iteration after the set iteration times are reached, so that the final filter coefficient is obtained.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310495477.0A CN116611223B (en) | 2023-05-05 | 2023-05-05 | Accurate array response control method and device combined with white noise gain constraint |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310495477.0A CN116611223B (en) | 2023-05-05 | 2023-05-05 | Accurate array response control method and device combined with white noise gain constraint |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116611223A true CN116611223A (en) | 2023-08-18 |
| CN116611223B CN116611223B (en) | 2023-12-19 |
Family
ID=87673948
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310495477.0A Active CN116611223B (en) | 2023-05-05 | 2023-05-05 | Accurate array response control method and device combined with white noise gain constraint |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116611223B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120093344A1 (en) * | 2009-04-09 | 2012-04-19 | Ntnu Technology Transfer As | Optimal modal beamformer for sensor arrays |
| CN106842237A (en) * | 2017-01-18 | 2017-06-13 | 南京理工大学 | The quick arbitrary shape conformal Adaptive beamformer method of the major lobe of directional diagram |
| CN108595758A (en) * | 2018-03-22 | 2018-09-28 | 西北工业大学 | A kind of optimal broadband beams figure integrated approach of arbitrary form sensor array |
| CN110174656A (en) * | 2019-05-21 | 2019-08-27 | 电子科技大学 | A kind of thinned array design method and device based on frequency domain broad-band EDFA |
| CN113655436A (en) * | 2021-07-23 | 2021-11-16 | 公安部第一研究所 | Method and device for optimizing broadband beam forming by particle swarm with channel calibration |
| CN115038012A (en) * | 2022-08-10 | 2022-09-09 | 湖北工业大学 | Microphone array robust frequency invariant beam forming method based on ADMM |
-
2023
- 2023-05-05 CN CN202310495477.0A patent/CN116611223B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120093344A1 (en) * | 2009-04-09 | 2012-04-19 | Ntnu Technology Transfer As | Optimal modal beamformer for sensor arrays |
| CN102440002A (en) * | 2009-04-09 | 2012-05-02 | 挪威科技大学技术转让公司 | Optimal modal beamformer for sensor arrays |
| CN106842237A (en) * | 2017-01-18 | 2017-06-13 | 南京理工大学 | The quick arbitrary shape conformal Adaptive beamformer method of the major lobe of directional diagram |
| CN108595758A (en) * | 2018-03-22 | 2018-09-28 | 西北工业大学 | A kind of optimal broadband beams figure integrated approach of arbitrary form sensor array |
| CN110174656A (en) * | 2019-05-21 | 2019-08-27 | 电子科技大学 | A kind of thinned array design method and device based on frequency domain broad-band EDFA |
| CN113655436A (en) * | 2021-07-23 | 2021-11-16 | 公安部第一研究所 | Method and device for optimizing broadband beam forming by particle swarm with channel calibration |
| CN115038012A (en) * | 2022-08-10 | 2022-09-09 | 湖北工业大学 | Microphone array robust frequency invariant beam forming method based on ADMM |
Non-Patent Citations (3)
| Title |
|---|
| BENHONG ZHANG ET AL.: "Bayesian Complex Amplitude Estimation and Adaptive Matched Filter Detection in Low-Rank Interference", 《IEEE TRANSACTIONS ON SIGNAL PROCESSING》, vol. 55, no. 3, pages 1176 - 1182, XP011165437, DOI: 10.1109/TSP.2006.887151 * |
| 鄢社锋;马晓川;: "宽带波束形成器的设计与实现", 声学学报(中文版), no. 04 * |
| 鄢社锋;马远良;: "二阶锥规划方法对于时空域滤波器的优化设计与验证", 中国科学(E辑:信息科学), no. 02 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116611223B (en) | 2023-12-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Benesty et al. | On microphone-array beamforming from a MIMO acoustic signal processing perspective | |
| KR100878992B1 (en) | Geometric source separation signal processing technique | |
| CN110931036B (en) | Microphone array beam forming method | |
| Wang et al. | Combining superdirective beamforming and frequency-domain blind source separation for highly reverberant signals | |
| JP6987075B2 (en) | Audio source separation | |
| US10283139B2 (en) | Reverberation suppression using multiple beamformers | |
| Li et al. | Geometrically constrained independent vector analysis for directional speech enhancement | |
| Fahim et al. | PSD estimation and source separation in a noisy reverberant environment using a spherical microphone array | |
| CN113655436A (en) | Method and device for optimizing broadband beam forming by particle swarm with channel calibration | |
| CN111681665A (en) | Omnidirectional noise reduction method, equipment and storage medium | |
| CN116611223B (en) | Accurate array response control method and device combined with white noise gain constraint | |
| Yamaoka et al. | CNN-based virtual microphone signal estimation for MPDR beamforming in underdetermined situations | |
| Niwa et al. | PSD estimation in beamspace using property of M-matrix | |
| Zhao et al. | Experimental study of robust beamforming techniques for acoustic applications | |
| Markovich‐Golan et al. | Spatial filtering | |
| Gao et al. | A modified frequency weighted MUSIC algorithm for multiple sound sources localization | |
| Doblinger | Optimized design of interpolated array and sparse array wideband beamformers | |
| Guzik et al. | Wishart localization prior on spatial covariance matrix in ambisonic source separation using non-negative tensor factorization | |
| CN114242104A (en) | Method, device and equipment for voice noise reduction and storage medium | |
| Wang et al. | Speech separation and extraction by combining superdirective beamforming and blind source separation | |
| Beit-On et al. | Binaural direction-of-arrival estimation in reverberant environments using the direct-path dominance test | |
| Zhang et al. | Frequency-invariant beamformer design via ADPM approach | |
| CN113053408B (en) | Sound source separation method and device | |
| Pu et al. | A penalized inequality-constrained approach for robust beamforming with DoF limitation | |
| Hioka et al. | Under-determined source separation based on power spectral density estimated using cylindrical mode beamforming |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |