Method and system for intelligently eliminating signal noise in acoustic measurement

Abstract

The embodiment of the invention provides a method and system for intelligently eliminating signal noise in acoustic measurement, and the method comprises the steps: collecting a sound source signal through a sound sensor array, dividing an observation plane of the sound sensor array into a plurality of grid points, and calculating the value of a sound propagation vector G; the method comprises the following steps: measuring the output of an acoustic sensor array in each scene in a wind tunnel test signal acquisition process, carrying out multilayer model design, and constructing an array-signal multilayer model; carrying out iterative operation on the array-signal multilayer model, and estimating a state value of each parameter; and carrying out iterative operation on sound source signals acquired by the sound sensor array, and repeating the iterative operation on each lattice point to eliminate noise in the sound source signals so as to realize sound source localization.

Description

Technical Field

[0001] This document relates to the technical field of aerodynamic noise measurement signal processing, and in particular to a method and system for intelligently eliminating signal noise in acoustic measurement. Background Art

[0002] Acoustic sensor arrays can achieve directional filtering of spatial sound fields, accurately identify the area where noise is generated and quantify the size of the noise, and are widely used in acoustic measurements. Acoustic measurements are generally carried out in wind tunnels. When the model is in the airflow, a series of complex signal sources will be generated. Accurately identifying the area where the signal is generated and quantifying the intensity of the signal provide an important basis for noise reduction research. It should be pointed out that when performing acoustic measurements in a wind tunnel, the signals collected by the array are mixed signals, which include not only the signals we are concerned about, but also some noise, such as channel noise, incoherent and coherent background noise. In some cases, the intensity of these noises is even greater than the signal intensity, so that the sound source is masked by the noise, affecting the identification of the sound source. Therefore, the noise needs to be removed during the experiment or data processing.

[0003] In the existing array signal processing methods, the classical algorithm uses spectral cross matrix subtraction and zeroing of the diagonal elements of the spectral cross matrix to remove background noise and channel noise respectively. However, this algorithm has two disadvantages. First, since the classical algorithm is based on the idea of ​​statistical averaging, it cannot display the state of the sound source in real time; second, the effect of removing background noise coherent with the sound source is not good; the state observer and extended Kalman filter algorithm remove coherent background noise through iterative calculation, but the convergence speed is not fast enough when the background noise phase produces a mutation, and the sound source state cannot be accurately displayed in real time for a period of time after the mutation point, and the convergence factor needs to be set manually, which makes it difficult to ensure that the convergence process reaches the optimal state; the algorithm combining the state observer and the strong tracking filter improves the speed of removing coherent background noise, but the stability of this method deteriorates when channel noise exists; although the adaptive beamforming algorithm can eliminate the influence of channel noise, it cannot remove coherent background noise.

[0004] In order to better optimize the acoustic imaging results in the presence of noise, it is necessary to invent an intelligent signal processing method that can intelligently estimate the phase value in real time according to certain rules when the background noise phase changes, without being restricted by experience and parameters. In the presence of channel noise, it can intelligently remove background noise interference and accurately locate the sound source. Summary of the invention

[0005] One or more embodiments of this specification provide a method for intelligently eliminating signal noise in acoustic measurement, including:

[0006] S1. Collecting sound source signals through the acoustic sensor array, dividing the acoustic sensor array observation plane into several grid points, and calculating the value of the sound propagation vector G;

[0007] S2. Measure the output of the acoustic sensor array in each scenario during the wind tunnel test signal acquisition process, design a multi-layer model, and construct an array-signal multi-layer model;

[0008] S3. Iterate the array-signal multilayer model to estimate the state value of each parameter;

[0009] S4. Perform an iterative operation on the sound source signal collected by the sound sensor array, and repeat the iterative operation for each grid point to eliminate noise in the sound source signal and realize sound source positioning.

[0010] Furthermore, the sound source signal is collected by the sound sensor array, the observation plane of the sound sensor array is divided into a number of grid points, and the value of the sound propagation vector G is calculated specifically including:

[0011] The sound source signal collected by the sound sensor is divided into two parts of length T according to the time series. B The data block is obtained by the output signal y of the acoustic sensor in the time domain n ,n=1,2,...,N, where n represents the number of the data block;

[0012] The observation plane of the acoustic sensor array is divided into a number of grid points, the value of the acoustic propagation vector G between each array element in the array and the grid point of the observation plane is calculated, and the output value of the acoustic sensor array is expressed in the frequency domain as a propagation vector.

[0013] Furthermore, the acoustic sensor array adopts an odd-number-arm spiral layout and a seven-arm 63-element array.

[0014] Furthermore, the noise includes channel noise, coherent background noise and incoherent background noise.

[0015] Furthermore, the output of the acoustic sensor array in each scenario during the wind tunnel test signal collection process is measured, a multi-layer model is designed, and an array-signal multi-layer model is constructed, including:

[0016] Determine the scenarios to measure, including:

[0017] The first scenario: measuring the background noise incoherent with the sound source and obtaining the incoherent background noise The output is mixed with the channel noise E

[0018] Scenario 2: Measuring background noise that is incoherent with the sound source Coherent background noise And the mixed output of channel noise E If there is no Then this step degenerates to the first scenario;

[0019] Scenario 3: Measuring signal X S , incoherent background noise Coherent background noise and the mixture of channel noise E

[0020] A "multi-layer" model is designed for three scenarios in the signal acquisition process, including:

[0021] The "first layer" includes incoherent background noise and channel noise E, where channel noise E is processed as Gaussian white noise, the array output is expressed as:

[0022] The "second layer" includes incoherent background noise Coherent background noise Channel noise E, array output is expressed as: In particular, if there is no coherent background noise Then the second layer is the same as the first layer;

[0023] The "third layer" includes signal X S , incoherent background noise Coherent background noise Channel noise E, array output is expressed as:

[0024] Considering the steady-state phase difference φ0 and the phase difference disturbance φ', a state space model about φ is established on the kth data block, and the coherent background noise phase difference φ=φ0+φ';

[0025] On the kth data block, the “multi-layer” acquisition model is rewritten as a state-space expression.

[0026] Furthermore, the array-signal multilayer model is iteratively calculated to estimate the state value of each parameter:

[0027] The array-signal multilayer model is iterated using Kalman filtering and strong tracking filtering:

[0028] The sound source on the k-th data block is recursively estimated by Kalman filtering, and the coherent background noise, incoherent background noise and channel noise on the k-th data block are removed at the same time;

[0029] The change of the phase difference φ of the coherent background noise is tracked by strong tracking filtering.

[0030] Furthermore, an iterative operation is performed on the sound source signal collected by the sound sensor array, and the iterative operation is repeated for each grid point to eliminate the noise in the sound source signal. The specific method for realizing sound source localization is as follows:

[0031] For each grid point on the observation plane of the acoustic sensor array, Kalman filtering and strong tracking filtering are used to perform iterative operations in the presence of channel noise to remove incoherent background noise and coherent background noise, thereby realizing the distribution and positioning of the sound source on the observation plane.

[0032] One or more embodiments of this specification provide a system for intelligently eliminating signal noise in acoustic measurement, including:

[0033] Signal processing module: used to collect sound source signals through the sound sensor array, divide the observation plane of the sound sensor array into a number of grid points, and calculate the value of the sound propagation vector G;

[0034] Model building module: used to measure the output of the acoustic sensor array in various scenarios during the wind tunnel test signal acquisition process, perform multi-layer model design, and build an array-signal multi-layer model;

[0035] Model calculation module: used to iterate the array-signal multi-layer model and estimate the state value of each parameter;

[0036] Noise elimination module: used to perform iterative operations on the sound source signal collected by the sound sensor array, and repeat the iterative operation for each grid point to eliminate the noise in the sound source signal and realize sound source positioning.

[0037] One or more embodiments of the present specification provide an electronic device, including:

[0038] processor; and,

[0039] A memory arranged to store computer executable instructions, which, when executed, cause the processor to implement the steps of the above-mentioned method for intelligently eliminating signal noise in acoustic measurement.

[0040] One or more embodiments of the present specification provide a storage medium for storing computer-executable instructions, which, when executed, implement the steps of the above-mentioned method for intelligently eliminating signal noise in acoustic measurement.

[0041] By adopting the embodiments of the present invention, the present invention can eliminate noise interference in real time in wind tunnel acoustic measurement, the data processing process is simple and efficient, the convergence speed is fast, the influence of noise on the measurement results is effectively reduced, the imaging effect of the acoustic sensor array is significantly improved, and the method design is intelligent and various random noise interferences are removed; it can meet the needs of various measurement conditions, has strong robustness to various noises, can further improve the sound source resolution, and improve the ability to identify the sound source.

[0042] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented according to the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate one or more embodiments of this specification or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0044] Figure 1 A flow chart of a method for intelligently eliminating signal noise in acoustic measurement provided by one or more embodiments of this specification;

[0045] Figure 2 A schematic diagram of the output results of the acoustic sensor propagation vector array when measuring aerodynamic noise in a wind tunnel using a method for intelligently eliminating signal noise in acoustic measurement provided by one or more embodiments of this specification;

[0046] Figure 3 A schematic diagram of beamforming results when performing acoustic measurement in an open wind tunnel according to a method for intelligently eliminating signal noise in acoustic measurement provided by one or more embodiments of this specification;

[0047] Figure 4 A schematic diagram of beamforming results after removing background noise when performing acoustic measurement in an open wind tunnel according to a method for intelligently eliminating signal noise in acoustic measurement provided by one or more embodiments of this specification;

[0048] Figure 5 A schematic diagram of the results of improving imaging resolution by a deconvolution algorithm in a method for intelligently eliminating signal noise in acoustic measurement provided by one or more embodiments of this specification;

[0049] Figure 6 A schematic diagram of the composition of a system for intelligently eliminating signal noise in acoustic measurement provided by one or more embodiments of this specification;

[0050] Figure 7 A schematic diagram of the structure of an electronic device provided for one or more embodiments of this specification. DETAILED DESCRIPTION

[0051] In order to enable those skilled in the art to better understand the technical solutions in one or more embodiments of this specification, the following will be combined with the drawings in one or more embodiments of this specification to clearly and completely describe the technical solutions in one or more embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on one or more embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this document.

[0052] Method Embodiment

[0053] According to an embodiment of the present invention, a method for intelligently eliminating signal noise in acoustic measurement is provided. Figure 1 A flowchart of a method for intelligently eliminating signal noise in acoustic measurement provided by one or more embodiments of this specification, such as Figure 1 As shown, the method for intelligently eliminating signal noise in acoustic measurement according to an embodiment of the present invention specifically includes:

[0054] S1. Collect sound source signals through the sound sensor array, divide the observation plane of the sound sensor array into several grid points, and calculate the value of the sound propagation vector G.

[0055] All acoustic sensors synchronously collect sound pressure and start calculation. The acoustic sensor array adopts an odd-arm spiral layout. The sound source signal collected by the acoustic sensor is divided into pieces of length T according to the time series. B The data block is obtained by the output signal y of the acoustic sensor in the time domain n ,n=1,2,...,N, where n represents the number of the data block; after Fourier transform, it is expressed as Y in the frequency domain n ,n=1,2,...,N。

[0056] The observation plane of the acoustic sensor array is divided into a number of grid points, and the value of the acoustic propagation vector G between each array element in the array and the grid point of the observation plane is calculated. In this embodiment, a seven-helix layout is adopted, which includes 63 array elements, and the output value of the acoustic sensor array is expressed in the frequency domain as a propagation vector Y=GX. Specifically, the signal collected by the array is processed in the frequency domain. For each grid point on the array observation plane, G=(1 / 4πr)e -iωτ , where r∈R M×1 is the distance between the sound source and the sound sensor, and τ=r / c0 is the propagation time related to the sound propagation distance.

[0057] S2. Measure the output of the acoustic sensor array in various scenarios during the wind tunnel test signal acquisition process, design a multi-layer model, and construct an array-signal multi-layer model.

[0058] The beamforming results are performed in the frequency domain, and the measurement scenarios are determined in chronological order, including:

[0059] The first scenario: measuring the background noise incoherent with the sound source and obtaining the incoherent background noise The output is mixed with the channel noise E

[0060] Scenario 2: Measuring background noise that is incoherent with the sound source Coherent background noise And the mixed output of channel noise E If there is no Then this step degenerates to the first scenario;

[0061] Scenario 3: Measuring signal X S , incoherent background noise Coherent background noise and the mixture of channel noise E

[0062] The above three scenarios are all expressed in the frequency domain. If the second scenario does not exist The first scene and the second scene are merged; in addition, since the measurement environment has not changed, the default time is The amplitude does not change.

[0063] A "multi-layer" model is designed for three scenarios in the signal acquisition process. If it does not exist in actual measurement The "layer" related to the sound source will be deleted, and and The signals are combined and processed in the "same layer". During the experiment, it is assumed that the channel noise E is Gaussian white noise, and an array-signal "multi-layer" acquisition model is established, including:

[0064] The "first layer" includes incoherent background noise and channel noise E, where channel noise E is processed as Gaussian white noise, the array output is expressed as:

[0065] The "second layer" includes incoherent background noise Coherent background noise Channel noise E, array output is expressed as: In particular, if there is no coherent background noise Then the second layer is the same as the first layer;

[0066] The "third layer" includes signal X S , incoherent background noise Coherent background noise Channel noise E, array output is expressed as:

[0067] In the above three measurement steps, the incoherent background noise is distributed over a wide frequency, so the phase φ change is not considered. The coherent background noise needs to consider the phase φ change. The "multi-layer" model can be used to represent the k-th data block as follows:

[0068]

[0069] The above “multi-layer” design covers typical scenarios in acoustic measurement and is used to estimate the signal state and eliminate various noise interferences during data iteration.

[0070] In the process of removing coherent background noise, the phase difference φ between two background noise measurements is not a fixed value. When establishing the state space model for φ, the steady-state phase difference φ0 and the phase difference disturbance φ' must be considered. During most of the measurement time, φ = φ0, and it is only disturbed at a few points. At this time, φ = φ0 + φ'. Considering the steady-state phase difference φ0 and the phase difference disturbance φ', the state space model for φ is established on the kth data block:

[0071]

[0072] φ is the phase difference of coherent background noise;

[0073] On the kth data block, the “multi-layer” acquisition model is rewritten as a state space expression:

[0074]

[0075] in,

[0076] S3. Perform iterative operations on the array-signal multi-layer model to estimate the state value of each parameter.

[0077] The array-signal multilayer model is iterated to estimate the state value of each parameter:

[0078] The Kalman filter and strong tracking filter are used to iteratively calculate the array-signal multilayer model to estimate the state value of the unknown parameter. Among them, the sound source X on the kth data block is recursively estimated by Kalman filtering. S | k , while removing the coherent background noise on the kth data block Incoherent background noise and channel noise E| k The interference of the coherent background noise is eliminated by tracking the change of the phase difference φ of the coherent background noise through strong tracking filtering, thereby improving the speed of removing the coherent background noise.

[0079] The time update equation can be described as:

[0080]

[0081]

[0082] The state update equation can be described as:

[0083]

[0084] Where, the superscript ^ is the state estimate, * is the transpose, and the subscript φ is the parameter related to the phase difference. For a stationary random process, A is the identity matrix, P is the covariance of the estimation error, Λ is the strong tracking suboptimal fading factor, K is the Kalman filter gain, R and R φ are the covariance of channel noise and phase noise respectively,

[0085] The linear states are given respectively Nonlinear state Covariance of the estimation errors and Initial estimate of:

[0086] The initial estimate of the linear state is given by:

[0087] Nonlinear state initial estimate:

[0088] Initial estimate of the covariance of the estimation error (can also be set to other values): and

[0089] S4. Perform an iterative operation on the sound source signal collected by the sound sensor array, and repeat the iterative operation for each grid point to eliminate noise in the sound source signal and realize sound source positioning.

[0090] The collected data is iterated by using Kalman filtering and strong tracking filtering according to the data block, and the value of the gain matrix K is adjusted in real time through the suboptimal fading factor Λ to make the error covariance converges to a stable value,

[0091] Phase Estimation Error The above step S3 is repeated for each grid point on the observation surface to obtain the quasi-steady-state estimation value of all grid points. In the presence of channel noise E, the incoherent and coherent background noise Get the sound source X on the array observation plane S Distribution positioning.

[0092] Taking the measurement of aerodynamic noise in a wind tunnel as an example, in the time domain, the incoherent background noise is measured at time t1. The propagation vectors from the sound source to the sound sensor are Each channel contains channel noise e1, e2, ...e n , the array output is During the measurement, the statistical distribution of the channel noise e remains unchanged; coherent background noise is added at time t2 The corresponding sound propagation vector is The array output is Install the model x to be tested at time t3 S (t3), the corresponding sound propagation vector is r S1 、r S2 ,…r Sn , the array output is like Figure 2 shown.

[0093] Convert the above time domain signal into frequency domain for calculation, input and x S (t3) respectively corresponds to and X S Output and Corresponding to and

[0094] The simulation is based on acoustic measurements in an open wind tunnel. The airflow at the wind tunnel nozzle is 30m / s, and the nozzle produces background noise that is incoherent with the sound source. Use a monopole to simulate coherent background noise with the same frequency (3kHz) as the sound source The channel noise E has a mean of 0 and a variance of 10. -4 Gaussian white noise distribution. The beamforming result is as follows Figure 3 As shown, the presence of noise affects the sound source identification, and the imaging results need to be optimized to eliminate the impact of noise.

[0095] Through the Kalman-strong tracking filter "multi-layer" background noise intelligent removal method proposed in the present invention, the first scene is placed in the "first layer", the second scene is placed in the "second layer", and the third scene is placed in the "third layer". Through iterative calculation, the Kalman filter part removes the channel noise E, and the strong tracking filter part tracks the phase difference φ to eliminate the background noise in real time. and The beamforming result is as follows: Figure 4As shown in the figure, (a), (b), (c), and (d) show the 1st, 2nd, 3rd, and 4th data blocks respectively. In the 4th data block, the influence of noise on the sound source imaging has been basically eliminated. The Kalman-strong tracking filter intelligent method converges faster than the existing methods and is suitable for optimizing the beamforming results under noisy conditions.

[0096] The method of the present invention retains the average acoustic energy of at least one data block during the data iterative calculation process, and the imaging resolution can be improved through the deconvolution algorithm. The result is as follows: Figure 5 shown. Figure 5 (a), (b), (c), and (d) are respectively Figure 4 (a), (b), (c), and (d) in the figure correspond to each other, proving that this method can further improve the ability to identify the sound source.

[0097] The beneficial effects of the present invention are as follows:

[0098] The present invention can eliminate noise interference in real time during wind tunnel acoustic measurement. The data processing process is simple and efficient, the convergence speed is fast, the influence of noise on the measurement results is effectively reduced, and the imaging effect of the acoustic sensor array is significantly improved. The method design is intelligent and can meet the needs of various measurement conditions. It has strong robustness to various noise interferences. The deconvolution algorithm can further improve the sound source resolution and the ability to identify the sound source.

[0099] System Example

[0100] According to an embodiment of the present invention, a method and system for intelligently eliminating signal noise in acoustic measurement are provided. Figure 6 A schematic diagram of a system for intelligently eliminating signal noise in acoustic measurement provided by one or more embodiments of this specification, such as Figure 6 As shown, the system for intelligently eliminating signal noise in acoustic measurement according to an embodiment of the present invention specifically includes:

[0101] Signal processing module 60: used to collect sound source signals through the sound sensor array, divide the observation plane of the sound sensor array into a number of grid points, and calculate the value of the sound propagation vector G;

[0102] Model building module 62: used to measure the output of the acoustic sensor array in various scenarios during the wind tunnel test signal collection process, perform multi-layer model design, and build an array-signal multi-layer model;

[0103] Model calculation module 64: used to perform iterative calculations on the array-signal multi-layer model to estimate the state value of each parameter;

[0104] The noise elimination module 66 is used to perform iterative operations on the sound source signal collected by the sound sensor array, and repeat the iterative operations for each grid point to eliminate the noise in the sound source signal and realize the sound source positioning.

[0105] The embodiment of the present invention is a system embodiment corresponding to the above-mentioned method embodiment. The specific operations of each module can be understood by referring to the description of the method embodiment, which will not be repeated here.

[0106] Device Example 1

[0107] An embodiment of the present invention provides an electronic device, such as Figure 7 As shown, it includes: a memory 70, a processor 72, and a computer program stored in the memory 70 and executable on the processor 72. When the computer program is executed by the processor 72, the following method steps are implemented:

[0108] S1. Collecting sound source signals through the acoustic sensor array, dividing the acoustic sensor array observation plane into several grid points, and calculating the value of the sound propagation vector G;

[0109] S2. Measure the output of the acoustic sensor array in each scenario during the wind tunnel test signal acquisition process, design a multi-layer model, and construct an array-signal multi-layer model;

[0110] S3. Iterate the array-signal multilayer model to estimate the state value of each parameter;

[0111] S4. Perform an iterative operation on the sound source signal collected by the sound sensor array, and repeat the iterative operation for each grid point to eliminate noise in the sound source signal and realize sound source positioning.

[0112] Device Example 2

[0113] An embodiment of the present invention provides a computer-readable storage medium, on which a program for implementing information transmission is stored. When the program is executed by the processor 72, the following method steps are implemented:

[0114] S1. Collecting sound source signals through the acoustic sensor array, dividing the acoustic sensor array observation plane into several grid points, and calculating the value of the sound propagation vector G;

[0115] S2. Measure the output of the acoustic sensor array in each scenario during the wind tunnel test signal acquisition process, design a multi-layer model, and construct an array-signal multi-layer model;

[0116] S3. Iterate the array-signal multilayer model to estimate the state value of each parameter;

[0117] S4. Perform an iterative operation on the sound source signal collected by the sound sensor array, and repeat the iterative operation for each grid point to eliminate noise in the sound source signal and realize sound source positioning.

[0118] The computer-readable storage medium in this embodiment includes, but is not limited to, ROM, RAM, magnetic disk or optical disk, etc.

[0119] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for intelligently eliminating signal noise in acoustic measurement, characterized in that: include: S1. Collect the sound source signal through the acoustic sensor array, divide the observation plane of the acoustic sensor array into several grid points, and calculate the value of the sound propagation vector G; S2. Measure the output of the acoustic sensor array in each scenario during the wind tunnel test signal acquisition process, design a multi-layer model, and construct an array-signal multi-layer model; S3. Iterate the array-signal multilayer model to estimate the state value of each parameter; S4. Perform an iterative operation on the sound source signal collected by the sound sensor array, and repeat the iterative operation for each grid point to eliminate noise in the sound source signal and realize sound source positioning.

2. The method according to claim 1, characterized in that The acoustic sensor array is used to collect the sound source signal, and the observation plane of the acoustic sensor array is divided into a number of grid points. The value of the sound propagation vector G is calculated specifically including: The sound source signal collected by the sound sensor is divided into two parts of length T according to the time series. B The data block is obtained by the output signal y of the acoustic sensor in the time domain n ,n=1,2,...,N, where n represents the number of the data block; The observation plane of the acoustic sensor array is divided into a number of grid points, the value of the acoustic propagation vector G between each array element in the array and the grid point of the observation plane is calculated, and the output value of the acoustic sensor array is expressed in the frequency domain as a propagation vector.

3. The method according to claim 2, characterized in that The acoustic sensor array adopts an odd-number-arm spiral layout and a seven-arm 63-element array.

4. The method according to claim 3, characterized in that The noise includes channel noise, coherent background noise and incoherent background noise.

5. The method according to claim 4, characterized in that The output of the acoustic sensor array in each scenario of the wind tunnel test signal acquisition process is measured, a multi-layer model is designed, and an array-signal multi-layer model is constructed, which includes: Determine the scenarios to measure, including: The first scenario: measuring the background noise incoherent with the sound source and obtaining the incoherent background noise The output is mixed with the channel noise E Scenario 2: Measuring background noise that is incoherent with the sound source Coherent background noise And the mixed output of channel noise E If there is no Then this step degenerates to the first scenario; Scenario 3: Measuring signal X S , incoherent background noise Coherent background noise and the mixture of channel noise E A "multi-layer" model is designed for three scenarios in the signal acquisition process, including: The "first layer" includes incoherent background noise and channel noise E, where the channel noise E is processed according to Gaussian white noise, and the array output is expressed as: The "second layer" includes incoherent background noise Coherent background noise Channel noise E, array output is expressed as: In particular, if there is no coherent background noise Then the second layer is the same as the first layer; "Layer 3" includes signal X S , incoherent background noise Coherent background noise Channel noise E, array output is expressed as: Considering the steady-state phase difference φ0 and the phase difference disturbance φ', a state space model about φ is established on the kth data block, and the coherent background noise phase difference φ=φ0+φ'; On the kth data block, the "multi-layer" acquisition model is rewritten as a state-space expression.

6. The method according to claim 5, characterized in that The array-signal multilayer model is iterated to estimate the state value of each parameter: The array-signal multilayer model is iterated using Kalman filtering and strong tracking filtering: The sound source on the k-th data block is recursively estimated by Kalman filtering, and the coherent background noise, incoherent background noise and channel noise on the k-th data block are removed at the same time; The change of the phase difference φ of the coherent background noise is tracked by strong tracking filtering.

7. The method according to claim 6, characterized in that The sound source signal collected by the sound sensor array is iterated, and the iterative operation is repeated for each grid point to eliminate the noise in the sound source signal and realize the sound source localization. The specific method is: For each grid point on the observation plane of the acoustic sensor array, Kalman filtering and strong tracking filtering are used to perform iterative operations in the presence of channel noise to remove incoherent background noise and coherent background noise, thereby realizing the distribution and positioning of the sound source on the observation plane.

8. A system for intelligently eliminating signal noise in acoustic measurement, characterized in that: include: Signal processing module: used to collect sound source signals through the sound sensor array, divide the observation plane of the sound sensor array into a number of grid points, and calculate the value of the sound propagation vector G; Model building module: used to measure the output of the acoustic sensor array in various scenarios during the wind tunnel test signal acquisition process, perform multi-layer model design, and build an array-signal multi-layer model; Model calculation module: used to iterate the array-signal multi-layer model and estimate the state value of each parameter; Noise elimination module: used to perform iterative operations on the sound source signal collected by the sound sensor array, and repeat the iterative operation for each grid point to eliminate the noise in the sound source signal and realize sound source positioning.

9. An electronic device, characterized in that: include: processor; as well as, A memory arranged to store computer executable instructions, which, when executed, cause the processor to implement the steps of the method for intelligently eliminating signal noise in acoustic measurement according to any one of claims 1 to 7.

10. A storage medium, characterized in that: Used to store computer executable instructions, which, when executed, implement the steps of the method for intelligently eliminating signal noise in acoustic measurement according to any one of claims 1 to 7.

Images