A planar microphone sensor array and a sound source positioning method thereof

Abstract

This invention discloses a planar microphone sensor array and its sound source localization method, including a microphone sensor assembly arranged in a planar shape, and a sound source localization method used in conjunction with the assembly. The planar microphone sensor array consists of nine microphone sensors arranged in a square shape, with the central microphone sensor as the reference sensor. Based on the positional differences of the other sensors on the plane relative to this sensor, phase control array technology is used to locate the sound source. The sound source localization method applied to the planar microphone sensor array in this invention includes a non-iterative adaptive beamforming algorithm, denoted as the DM / ABF algorithm, and a low-rank approximation multiple signal classification algorithm.

Description

Technical Field

[0001] This invention relates to the field of sound source localization technology, and in particular to a planar microphone sensor array and a method for sound source localization. Background Technology

[0002] Microphone array-based sound source localization utilizes microphone arrays to pick up sound signals and, by combining the relationship between the sound source and the array structure, obtains the location information of one or more sound sources. Compared to a single microphone, a microphone array adds a spatial domain to the time and frequency domains, enhancing its ability to process sound information and making it the preferred choice for many high-quality voice communication applications. In the 1980s, microphone arrays were applied to conference systems in large conference rooms, demonstrating their unique advantages in speech signal processing. In recent years, it has become one of the research hotspots in the field of modern signal processing, with wide applications in video conferencing, speech recognition, speaker recognition, and especially sound source localization.

[0003] The effectiveness of sound source localization depends to some extent on the number and arrangement of microphone sensors, as well as the sound source localization method used for signal processing. Traditional microphone sensor arrays often use linear arrangements, and the simple arrangement structure leads to poor localization results. Therefore, this invention proposes a planar microphone sensor array and its sound source localization method to obtain better sound source localization results. Summary of the Invention

[0004] The purpose of this invention is to propose a planar microphone sensor array and its sound source localization method to achieve better sound source localization results.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A planar microphone sensor array and its sound source localization method are disclosed. The array includes a microphone array acquisition module composed of nine microphones arranged and fixedly connected to a vertical plate; a small circuit mounting box, each unit independently constructed and fixedly connected to the vertical plate and microphone sensor indicator lights; a mounting bracket, consisting of a trapezoidal base, connected to the vertical plate by screws; a microphone sensor acquisition switch, consisting of a movable button, fixedly connected to the vertical plate; three microphone sensor indicator lights, consisting of three indicator lights, fixedly connected to the small circuit mounting box; an output interface for exporting acquired data, consisting of several output lines, fixedly connected to the vertical plate; and a sound source localization method for processing the received signals.

[0007] Preferably, the two-dimensional microphone array acquisition module consists of nine electret microphones with a sensitivity of 58±3dB, a frequency range of 0Hz-30kHz, an output impedance of 2.2kΩ, and omnidirectional orientation, arranged in a square shape.

[0008] Preferably, the small circuit mounting box contains multiple integrated circuit structures, including a microphone first-stage amplification circuit, a microphone second-stage amplification circuit, an audio acquisition circuit, a sample-and-hold circuit, a filtering circuit, and an A / D conversion module. The relationship between the circuits is as follows: first, data is acquired through the audio acquisition circuit; then, the sample-and-hold circuit performs sampling and holding; the A / D conversion module performs data state conversion; then, the signal is amplified through the microphone first-stage amplification circuit and the microphone second-stage amplification circuit; and finally, the signal is filtered through the filtering circuit.

[0009] Preferably, the small circuit housing contains a first-stage amplifier circuit that utilizes a MAX8912L microphone gain chip.

[0010] Preferably, the small circuit housing has an internal second-stage amplifier circuit that utilizes an LM324 amplifier chip to provide a high-gain amplifier circuit with adjustable amplification factor, and the voltage amplification factor can reach 100-300 times.

[0011] Preferably, the small circuit box has an internal audio acquisition circuit mainly implemented by the Hengtong DAR2000 audio acquisition card, which inputs a TRS 3.5mm standard audio interface for active audio signals and outputs a PCI bus interface.

[0012] Preferably, in the sample-and-hold circuit section of the small circuit mounting box, the LF398 is used as the sample-and-hold chip.

[0013] Preferably, in the sample-and-hold circuit, the logic control signal of the LF398 chip is provided by the PCI-6070E acquisition card, ensuring that the signal sampling and A / D conversion can work normally.

[0014] Preferably, the sound source localization method for processing the signal received by the microphone sensor includes a non-iterative adaptive beamforming algorithm ABF, denoted as the DM / ABF algorithm, which is based on a combination of direction-of-arrival (DOA) estimation and the method of moments (MoM), to enhance the signal-to-noise ratio of the signal received by the planar microphone array in the invention.

[0015] Preferably, the sound source localization method for processing the signal received by the microphone sensor includes a low-rank approximation multiple signal classification algorithm. This algorithm decomposes the signal feature values ​​with high signal-to-noise ratio into a noise subspace and a signal subspace. The noise subspace after low-rank approximation is used to replace the noise subspace of the original multiple signal classification algorithm. By utilizing the orthogonality between the noise subspace and the signal subspace, the algorithm traverses spatial angles to find the maximum value of the spectral function, thereby estimating the direction of arrival of the signal with high signal-to-noise ratio.

[0016] Compared with the prior art, the beneficial effects of the present invention are:

[0017] 1. The microphone sensor array of the present invention adopts a square planar distribution arrangement. Compared with the traditional linear microphone array based on a small number of microphones, it increases the number of microphone sensors and changes the arrangement shape of the microphone sensors. It uses high-dimensional phase control array technology to process the received signal. The larger number of sensors and the more complex sensor arrangement structure are conducive to obtaining more accurate sound source localization results.

[0018] 2. The small circuit housing included in this invention includes an amplifier circuit that effectively amplifies the acquired sound signal multiple times, strengthening the useful signal. At the same time, the filter circuit can effectively filter the received signal, weakening noise or unwanted non-purposeful signals, thus improving the quality of the received signal.

[0019] 3. The sound source localization method using the DM / ABF algorithm combined with the rank approximation multiple signal classification algorithm can purify the received sound source signal, enhance the signal-to-noise ratio of the received signal, and obtain a more accurate localization effect. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of a planar microphone sensor array device.

[0021] Figure 2 This is a flowchart illustrating the overall workflow of the present invention.

[0022] Figure 3 This is a schematic diagram of the arrangement of nine microphone sensors in a planar microphone sensor array.

[0023] Figure 4 This is a schematic diagram of the sensor array of the present invention receiving far-field incoming wave signals.

[0024] Figure 5 This is a block diagram illustrating the principle of the DM / ABF algorithm in the sound source localization method of the present invention.

[0025] Figure 6 This is a flowchart illustrating the overall process of the sound source localization method of the present invention.

[0026] Figure 1 In the middle, 1-microphone array acquisition module, 2-small circuit mounting box, 3-fixed bracket, 4-screw, 5-microphone sensor acquisition switch, 6-microphone sensor indicator light, 7-output interface for exporting acquired data, 8-upright board. Detailed Implementation

[0027] The technical solution of this patent will be described in further detail below.

[0028] The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain this patent, and should not be construed as limiting this patent.

[0029] In the description of this patent, it should be understood that the terms "sensor", "array", "sound source localization", "amplifier circuit", "acquisition circuit", "filter circuit", "A / D conversion", "multiple signal classification", "method of moments", "adaptive beamforming algorithm", "eigenvalue", "singular value", and other terms related to signal reception and processing are used only for the convenience of describing this patent and simplifying the description, and should not be construed as limiting this patent.

[0030] In the description of this patent, it should be noted that, unless otherwise expressly specified and limited, the terms "switch," "output line," "box," and "microphone" should be interpreted broadly. Those skilled in the art can understand the specific meaning of these terms in this patent based on the specific circumstances.

[0031] Implementation

[0032] A planar microphone sensor array includes a two-dimensional microphone array acquisition module consisting of nine sound microphone sensors arranged in a square shape, a small circuit mounting box, a fixing bracket, screws connecting the bracket and the planar microphone array acquisition module, a microphone sensor acquisition switch, and an output interface for exporting acquired data; and a sound source localization method for processing signals acquired by the microphone array.

[0033] The planar microphone array acquisition module consists of nine electret microphones with a sensitivity of 58±3dB, a frequency range of 0Hz-30kHz, an output impedance of 2.2kΩ, omnidirectional orientation, and an operating voltage of 4.5V, arranged in combination.

[0034] As a further aspect of the present invention: the microphone sensor assembly is arranged in a square shape.

[0035] As a further aspect of the present invention: the small circuit mounting box includes a microphone first-stage amplification circuit, a microphone second-stage amplification circuit, an audio acquisition circuit, a sample-and-hold circuit, a filtering circuit, and an A / D conversion module.

[0036] As a further aspect of the present invention: the small circuit mounting box, the first stage amplifier circuit inside it utilizes a MAX8912L microphone gain chip to provide a stable microphone voltage bias and a 20dB fixed gain, which has the advantages of low noise and high bandwidth.

[0037] As a further aspect of the present invention: the small circuit mounting box has an internal second-stage amplifier circuit that utilizes an LM324 amplifier chip to provide a high-gain amplifier circuit with adjustable amplification factor, and the voltage amplification factor can reach 100-300 times.

[0038] As a further aspect of the present invention: the small circuit box is equipped with an audio acquisition circuit mainly implemented by the Hengtong DAR2000 audio acquisition card, whose main features include: input TRS 3.5mm standard audio interface, active audio signal, and output PCI bus interface.

[0039] As a further aspect of the present invention: the audio acquisition card supports 30 audio inputs, real-time recording of multiple audio channels, supports audio compression formats such as ADPCM and MP3, exports standard WAV format audio files, and supports audio sampling frequencies such as 8K, 16K, and 32K.

[0040] As a further aspect of the present invention: the main function of the sample-and-hold circuit inside the small circuit housing is to ensure that the signals of each microphone channel can be sampled synchronously. The present invention uses LF398 as the sample-and-hold chip.

[0041] As a further aspect of the present invention: the logic control signal of the LF398 chip in the sample-and-hold circuit is provided by the PCI-6070E acquisition card, ensuring that the signal sampling and A / D conversion can work normally.

[0042] As a further aspect of the present invention, the microphone sensor acquisition switch can control the supply and stop of the sensor power supply.

[0043] As a further aspect of the present invention: when the indicator light is bright, it indicates that the planar microphone sensor array is in working condition; when the indicator light is dim, it indicates that the planar microphone sensor array is in a closed condition.

[0044] As a further aspect of the present invention: the output line should be connected to a PC during operation to transmit the received and processed data to the PC for display.

[0045] As a further aspect of the present invention: the sound source localization method uses a non-iterative adaptive beamforming algorithm (ABF) based on a combination of direction of arrival (DOA) estimation and the method of moments (MoM), denoted as the DM / ABF algorithm, to enhance the signal-to-noise ratio of the planar microphone array received signal in the invention, and then uses a low-rank approximation multiple signal classification algorithm to locate the direction of the sound source signal.

[0046] First, connect the output interface 7 for exporting the acquired data to the PC. Then, point the planar microphone sensor array towards the approximate direction of the target to be detected. Next, turn on the device and press the microphone sensor acquisition switch 5. Wait for all three microphone sensor indicator lights 6 to illuminate before starting the acquisition of the target object's sound signal. The audio acquisition circuit included in the small circuit mounting box 2 will acquire the signal. The acquired signal will then be processed by the amplification circuit, filtering circuit, and A / D conversion module included in the small circuit mounting box.

[0047] Finally, the collected signals are processed and located using the sound source localization method described in this invention.

[0048] The structure of the planar microphone sensor array device in this invention is shown in the attached figure. Figure 1 As shown

[0049] The overall workflow diagram of this invention is attached. Figure 2 As shown.

[0050] The arrangement of the nine microphone sensors of this invention is shown in the attached figure. Figure 3 As shown.

[0051] A schematic diagram of the sensor array of the present invention receiving far-field incoming wave signals is attached. Figure 4 As shown.

[0052] The principle block diagram of the DM / ABF algorithm in the sound source localization method of this invention is attached. Figure 5 As shown.

[0053] The overall flowchart of the sound source localization method of the present invention is attached. Figure 6 As shown.

[0054] The working principle of the sound source localization method described below is described in detail below.

[0055] The sound source localization method for planar microphone arrays proposed in this invention uses a non-iterative adaptive beamforming algorithm (ABF) based on a combination of direction of arrival (DOA) estimation and the method of moments (MoM), denoted as the DM / ABF algorithm, to enhance the signal-to-noise ratio of the received signal of the planar microphone array in the invention. Then, a low-rank approximation multiple signal classification algorithm is used to locate the direction of the sound source signal.

[0056] In the DM / ABF algorithm, the array element excitation or weight vector is not updated for each received signal sample.

[0057] However, the array element excitation or weight vector will be updated when the estimated direction of arrival of the received signal changes as follows:

[0058] 1. Based on the estimation of the direction of arrival of the interference signal, create the corresponding zero-value function, multiply it by the original array pattern, create depth zero values ​​in these directions, and at the same time, turn the main beam to the desired signal direction to form the desired shape pattern.

[0059] 2. The method of moments (MoM) accepts this shape as input and generates a set of linear equations. By solving these linear equations, the elemental excitations required to synthesize the desired pattern can be determined.

[0060] The received signal is weighted by the generated excitation or weight to reduce interference signals and achieve a high signal-to-noise ratio.

[0061] Furthermore, in the adaptive beamforming algorithm of the direction-of-arrival joint moment method (DM / ABF), adaptive beamforming is achieved by controlling the array excitation coefficients, or amplitude and phase, rather than processing the received signal samples.

[0062] The DM / ABF algorithm operates based on two main processes: DOA estimation and array synthesis. Its algorithm block diagram is shown below. Figure 1 As shown.

[0063] Direction of arrival (DOA) estimation is used to determine the direction of arrival of the signal received by the microphone array. Adaptive localization of signal sources requires DOA estimation.

[0064] The method of moments is a simple numerical technique that is mainly used to provide deterministic solutions to integral equations.

[0065] It takes the required pattern as input and transforms the problem into a well-conditional linear matrix equation that can be solved without any special processing.

[0066] The algorithm proposed in this invention provides the weight vector required for sensor unit excitation or synthesis of patterns, which have a main beam pointing to the desired signal and a deep space value pointing to interference.

[0067] Unlike traditional Least Mean Square (LMS) and Recursive Least Squares (RLS) algorithms, the weight vector in the DM / ABF algorithm of this invention is not updated iteratively on every sample, but only when the estimated direction of arrival of the signal changes.

[0068] Suppose there are K signal sources in total.

[0069] To make it more general and easier to change the array configuration later, we consider a uniform linear antenna array consisting of N microphone elements, with each element spacing being d. The array pattern AF(θ) can be expressed as:

[0070]

[0071] In the formula, a n λ is the excitation coefficient of the nth element; β = 2π / λ is the free space of the wavenumber; d is the fixed element spacing in the ULA.

[0072] Considering the direction (θ) I1 θ I2 , ..., θ IM The M interference signals from the microphone array in the impact invention and the signals from θ d The required sound source signal in the direction.

[0073] The specific steps of the sound source localization method of the present invention are as follows:

[0074] S1. First, the directions of the target signal and the interference signal are estimated using modified virtual singular value decomposition (MV-SVD) and DOA estimation techniques;

[0075] This DOA estimation technique has the following advantages:

[0076] 1. This method can be used for the detection of coherent and incoherent signals.

[0077] 2. This method provides high resolution and stability even at low signal-to-noise ratios.

[0078] 3. This method can detect multiple incoming wave signals of number N-1 or less.

[0079] 4. This method is superior to existing techniques such as SVD, forward / backward spatial smoothing (FBSS), virtual array spatial smoothing (VSS), and rotation invariant technique (ESPRIT).

[0080] MV-SVD technology is a combination of Virtual Array Extension (VAE), SVD, and Multiple Signal Classification (MUSIC) algorithms.

[0081] Suppose the transmitted signal S(k) is contaminated with additive white Gaussian noise v(k) with zero mean and variance σ. 2 If the value is zero, then the received signal X(k) can be expressed as:

[0082] X(k)=AS(k)+v(k) (2)

[0083] Where k = 1, 2, 3, ..., K; is the number of received signals.

[0084] A is the received interference signal and the desired signal at θ = (θ I1 θ I2 , ..., θ IM θ d The N×ξ-sized microphone array steering matrix in the direction, ξ=M+1;

[0085] A(θ)=[a(θ I1 ), a(θ I2 ), ..., a(θ) IM ), a(θ d (3)

[0086] a(θ) = [1, e -j(2π / λ)dsinθ , ..., e -j(N-1)(2π / λ)dsinθ ] T (4)

[0087] S(k)=[s1(k),s2(k),...,s ξ (k)] (5)

[0088] Furthermore, to improve the direction-of-arrival estimation capability, the MV-SVD algorithm utilizes virtual array expansion to extend the size of the original antenna array from N elements to 2N-1 virtual elements.

[0089] In this case, the original N×K size received signal sample matrix X is expanded into a larger (2N-1)×K size sample matrix X1.

[0090]

[0091] in,

[0092] X(k)=[x1(k),x2(k),...,x N (k)] T (7)

[0093] X(k)′=[x N (k) * x N-1 (k) * , ..., x2(k) * ] T (8)

[0094] Then, a large covariance matrix is ​​constructed using the enlarged sample matrix X1.

[0095] R is determined by eigenvalue decomposition. X1 The eigenvalues ​​and corresponding eigenvectors e = [e1, e2, ..., e] 2N-1 ].

[0096] Construct a new N×N covariance matrix using this eigenvector.

[0097] Determine the noise subspace

[0098] Un = e ξ+1 e ξ+2 ...e N (9)

[0099] and signal subspace

[0100] Us = e1e2...e ξ (10)

[0101] The direction of arrival of the signal impacting the microphone array is estimated using the pseudospectral function P(θ) of the MUSIC algorithm.

[0102]

[0103] S2. After estimating the direction of the incoming waves of the desired signal and the interference signal, the main wave of AF(θ) will turn to θd.

[0104] S3, for those from θ I1 θ I2 ...θ IM M interference signals in the direction are obtained by multiplying the guide pattern by M narrow-band zero-value functions f centered at these angles. null (θ) generates M zero values ​​in the guide pattern.

[0105] Based on the hyperbolic secant function f(θ)=[sech(θ)] α The zero-valued function is

[0106] f null (θ-θ Im )=1-[sech(θ-γ Im )] α (12)

[0107] Based on rectangular functions The zero-valued function is

[0108]

[0109] Based on trigonometric functions The zero-valued function is

[0110]

[0111] Where α is an exponent that controls the width of the hyperbolic secant function;

[0112] τ represents the width of the rectangular and triangular functions; X1(k)

[0113] θ Im It is the center of the zero function, corresponding to the direction of the interference signal.

[0114] For M interference signals, the shape pattern generated by M zero values ​​can be represented as:

[0115] AF sh (θ)=AF(θ)×[f null (θ-θ I1 )×f null (θ-θ I2 )×...×f null (θ-θ IM (15)

[0116] S4. In this step, an array pattern AF needs to be synthesized. syn (θ), it has a main beam pointing in the desired signal direction (θd), and is also sensitive to interference signal direction (θ). i1 θ i2 , ..., θ iM It has a deep null value.

[0117] The excitation coefficients of the composite drawing are determined using the method of moments.

[0118] According to formula (1), the synthesized radiation pattern AF syn (θ) should correspond to the desired shape pattern AF sh (θ) aligns at a height, therefore:

[0119]

[0120] Among them, w n Let be the composite excitation coefficient of the nth antenna element. The composite pattern should have the same characteristics as the shaped pattern. Equation (15) can be written in matrix form:

[0121] [W] 1×N ×[Q] N×T =[F] 1×T (17)

[0122] In the formula, T is the number of samples in the array pattern; Q is an N×T matrix containing e iβndcosθ The sample information, θ = [θ1, θ2, ..., θ T ].

[0123] [W] 1×N This refers to the comprehensive excitation coefficient or the weighted vector.

[0124] Where W = [w0, w2, ..., w N-1 ].

[0125] F is a 1×T vector containing θ = [θ1, θ2, ..., θ3]. T ] Shape array pattern.

[0126] When Q is a non-square matrix, the comprehensive excitation coefficient W can be obtained by solving the MoM linear equations using the Gaussian elimination method.

[0127] However, if Q is designed as a square matrix such that N = T, the excitation coefficient can be obtained:

[0128] W = FQ -1 (18)

[0129] Synthetic array pattern AF syn (θ) will provide a deep space region for the interference signal.

[0130] Furthermore, the array diagram can update the zero function center angle θ based on the estimated direction of arrival. Im Adaptive beamforming is performed.

[0131] S5, each antenna element x n (k) The received signals are multiplied by complex adaptive weights w n As shown below:

[0132] y n (k)=w n ·x n (k) (19)

[0133] The weighted signal y n (k) The output signal Y is obtained by combining delay and array synthesis techniques. combined Its expression is:

[0134]

[0135] Furthermore, the obtained output signal Y combined It has a higher signal-to-noise ratio compared to the original signal initially received by the microphone array.

[0136] A high signal-to-noise ratio is, to some extent, equivalent to effectively filtering the original signal and retaining the effective part of the original signal, thus maximizing the purification of the received signal.

[0137] S6. The output signal Y obtained after processing by the DM / ABF algorithm combined The input S1 replaces the original received signal X(k), and the direction of the incoming signal is accurately located by the low-rank approximation multiple signal classification algorithm in the DOA estimation algorithm.

[0138] Use Y to output the signal c Let represent it, and after autocorrelation processing, we obtain its covariance matrix R.c :

[0139] R c =E[Y c Y c H ] (twenty one)

[0140] Where H represents the matrix conjugate transpose.

[0141] The signal and noise are uncorrelated, and the noise is zero-mean white noise. Using the definition of a matrix, Y can be... c This can be expressed as the following formula:

[0142] Y c =AS c +N (22)

[0143] in,

[0144] A is the array manifold matrix, which is related to the square array shape in this invention.

[0145] N c This refers to Gaussian white noise in the output signal.

[0146] S c The signal of interest in the output signal.

[0147] Substituting can have

[0148] R c =E[(AS c +N c (AS) c +N c ) H ] (twenty three)

[0149] Performing eigenvalue decomposition on the covariance matrix, we have:

[0150]

[0151] in

[0152] R s =E[SS H (25)

[0153] Let be the correlation matrix of the signal.

[0154] R N =σ 2 I (26)

[0155] This is the correlation matrix for the noise.

[0156] σ 2 It is noise power.

[0157] I is an M×M identity matrix.

[0158] It should also be noted that, in practical applications, R is usually not directly obtainable. c Only the sample covariance matrix is ​​available.

[0159]

[0160] It is R c The maximum likelihood estimates are consistent as the number of samples L approaches infinity.

[0161] However, in reality, errors will occur due to the limited number of samples. Correspondingly, the larger the number of samples, the smaller the error.

[0162] Based on the order of the magnitude of the decomposed eigenvalues, the eigenvalues ​​and corresponding eigenvectors that are equal to the number of signals K are regarded as the signal part space, and the remaining NK eigenvalues ​​and eigenvectors are regarded as the noise part space.

[0163] Obtain the noise matrix E n :

[0164] E n =[v K+1 v K+2 , ..., v N (28)

[0165] And because

[0166] A H v1=0;i=K+1,K+2,...,N (29)

[0167] Therefore, it can be The θ change, which involves traversing various angles in space and then calculating...

[0168]

[0169] In P mu (θ) reaches its minimum value It is the estimated direction of the sound source signal.

[0170] This invention utilizes an improved multi-signal classification algorithm, a low-rank approximation multi-signal classification algorithm.

[0171] The low-rank approximation method is described in detail below.

[0172] The improved algorithm, as described above, lets I′ be an N×N order reverse identity matrix, i.e.

[0173]

[0174] And order

[0175]

[0176] in, For R c Conjugate, then with respect to R cc Perform singular value decomposition, and we have

[0177] [U, S, V] = svd(R) cc (33)

[0178] Pick

[0179] Vu=U(:,K+1:N) (34)

[0180] This is the eigenvector corresponding to the noise eigenvalue.

[0181] make

[0182] S(M, M)=0, S(M-1, M-1)=0, ...., S(M-D+1, M-D+1)=0; (35)

[0183] and take

[0184] S s =S; (36)

[0185] R=US S V′; (37)

[0186] Where V′ is the conjugate of V,

[0187] Replace the full-rank matrix R with the low-rank matrix R mentioned above. c Then perform singular value decomposition on R.

[0188] [U U S SS V V ]=svd(R) (38)

[0189] Pick

[0190] Vuu=U U (:,D+1,M) (39)

[0191] This is the eigenvector corresponding to the noise eigenvalue, i.e., the noise eigenvector.

[0192] Then, average the noise feature vectors obtained from the two tests to obtain...

[0193] V M =(Vuu+Vu) / 2 (40)

[0194] This is the processed noise feature vector.

[0195] The noise feature vector V M Substitute the original spatial spectral function P mu A new spatial spectral function is obtained.

[0196]

[0197] in θ is the azimuth angle, and θ is the elevation angle.

[0198] By performing spatial traversal on this function, the direction of arrival of the signal can be determined.

[0199] The low-rank approximation multiple signal classification method can achieve more accurate positioning results compared to the traditional MUSIC algorithm.

[0200] The beneficial effects of this invention are:

[0201] 1. The microphone sensor array of the present invention adopts a square planar distribution arrangement. Compared with the traditional linear microphone array based on a small number of microphones, it increases the number of microphone sensors and changes the arrangement shape of the microphone sensors. The larger number of sensors and the more complex sensor arrangement structure are conducive to obtaining more accurate sound source localization results.

[0202] 2. The small circuit housing included in this invention includes an amplifier circuit that effectively amplifies the acquired sound signal multiple times, strengthening the useful signal. At the same time, the filter circuit can effectively filter the received signal, weakening noise or unwanted non-purposeful signals, thus improving the quality of the received signal.

[0203] 3. The sound source localization method using the DM / ABF algorithm combined with the rank approximation multiple signal classification algorithm can purify the received sound source signal, enhance the signal-to-noise ratio of the received signal, and obtain a more accurate localization effect.

[0204] The above are merely preferred embodiments of the present invention. It should be noted that those skilled in the art can make several modifications and improvements without departing from the concept of the present invention, and these should also be considered within the scope of protection of the present invention. These will not affect the effectiveness of the implementation of the present invention or the practicality of the patent.

Claims

1. A planar microphone sensor array, characterized in that, The system includes a microphone array acquisition module (1), which consists of nine microphones arranged in a combination and is fixedly connected to a vertical plate (8); a small circuit mounting box (2), which is independently composed of circuit mounting box units and is fixedly connected to the vertical plate (8) and the microphone sensor indicator light (6); a fixed bracket (3), which is composed of a three-dimensional trapezoidal base and is connected to the vertical plate (8) by screws (4); a microphone sensor acquisition switch (5), which is composed of a movable button and is fixedly connected to the vertical plate (8); a microphone sensor indicator light (6), which is composed of three indicator lights and is fixedly connected to the small circuit mounting box (2); and an output interface (7) for exporting acquired data, which is composed of several output lines and is fixedly connected to the vertical plate (8). The microphone array acquisition module (1) consists of nine electret microphones with a sensitivity of 58±3dB, a frequency range of 0Hz-30kHz, an output impedance of 2.2kΩ, and omnidirectional orientation, arranged in a square shape. It also includes a sound source localization method for processing received signals; The sound source localization method is as follows: a non-iterative adaptive beamforming algorithm ABF, denoted as DM / ABF algorithm, based on the combination of direction of arrival (DOA) estimation and the method of moments (MoM), is used to enhance the signal-to-noise ratio of the signal received by the planar microphone array, and then a low-rank approximation multiple signal classification algorithm is used to locate the direction of the sound source signal.

2. The planar microphone sensor array according to claim 1, characterized in that, The small circuit mounting box (2) contains a variety of integrated circuit structures, including a microphone first-stage amplifier circuit, a microphone second-stage amplifier circuit, an audio acquisition circuit, a sample-and-hold circuit, a filter circuit, and an A / D conversion module. The relationship between the circuits is that the audio acquisition circuit first acquires data, then the sample-and-hold circuit performs sampling and holding, the A / D conversion module performs data state conversion, then the microphone first-stage amplifier circuit and the microphone second-stage amplifier circuit amplify the signal, and finally the filter circuit performs filtering processing on the signal.

3. The planar microphone sensor array according to claim 2, characterized in that, The small circuit mounting box (2) has a first-stage amplifier circuit inside that utilizes a MAX8912L microphone gain chip.

4. The planar microphone sensor array according to claim 2, characterized in that, The small circuit mounting box (2) has an internal second-stage amplifier circuit that uses an LM324 amplifier chip to provide a high-gain amplifier circuit with adjustable amplification factor, and the voltage amplification factor can reach 100-300 times.

5. The planar microphone sensor array according to claim 2, characterized in that, The small circuit box (2) has an internal audio acquisition circuit implemented by the Hengtong DAR2000 audio acquisition card, which inputs a TRS 3.5mm standard audio interface and an active audio signal, and outputs a PCI bus interface.

6. The planar microphone sensor array according to claim 2, characterized in that, The small circuit mounting box (2) mentioned above uses LF398 as the sample and hold chip in its sample and hold circuit part.

7. The planar microphone sensor array according to claim 6, characterized in that, The logic control signals for the LF398 chip in the sample-and-hold circuit are provided by the PCI-6070E data acquisition card, ensuring that signal sampling and A / D conversion can work normally.

8. A sound source localization method for a planar microphone sensor array, applicable to a planar microphone sensor array as described in any one of claims 1-7, characterized in that, A sound source localization method that processes signals received by a microphone sensor includes a non-iterative adaptive beamforming algorithm ABF, denoted as the DM / ABF algorithm, which combines direction-of-arrival (DOA) estimation and the method of moments (MoM) to enhance the signal-to-noise ratio of signals received by a planar microphone array.

9. The sound source localization method of the planar microphone sensor array according to claim 8, characterized in that, Sound source localization methods that process signals received by microphone sensors include low-rank approximation multiple signal classification algorithms. This algorithm decomposes the eigenvalues ​​of signals with high signal-to-noise ratios into noise subspaces and signal subspaces. It then uses the noise subspace obtained by low-rank approximation to replace the noise subspace of the original multiple signal classification algorithm. By utilizing the orthogonality between the noise subspace and the signal subspace, it traverses spatial angles to find the maximum value of the spectral function, thereby estimating the direction of arrival of signals with high signal-to-noise ratios.

Images