A method and device for passively locating the direction of a noise source

Abstract

The present invention discloses a method and device for passively locating the direction of a noise source, including: obtaining the noise source signal collected by each microphone in a microphone array, calculating the sound pressure level of the noise source signal collected by each microphone; screening out each noise source signal whose sound pressure level is greater than a preset threshold as a first signal, and according to the respective first signals of each group of microphone pairs, using a generalized cross-correlation phase transformation method and combining a far-field model, determining the angle between each first signal and the corresponding microphone pair, and converting the angle into a global angle with a preset layout of the microphone array as a reference; using a consensus detection method based on angle differences to exclude the mirror angles in each global angle, and obtaining the true angle of each microphone pair; fusing the true angles of multiple groups of microphone pairs, and weighted calculation to obtain the position of the noise source. The present invention realizes a simpler passive positioning of the noise direction of a single sound source.

Description

Technical Field

[0001] The invention relates to the field of noise source orientation, and in particular to a method and device for passively locating the direction of a noise source. Background Art

[0002] In order to facilitate noise control and law enforcement by government management departments, it is necessary to accurately determine the location or direction of the noise source. In the noise source direction estimation technology, it is a common method to estimate the direction of the sound source using the wave delay detected by different microphone array elements. This method is mainly divided into two steps: the first step is to estimate the time difference between the sound source signal arriving at each microphone; the second step is to estimate the location of the sound source using the time difference obtained in the previous step and the geometric relationship between the sound source and the microphone array.

[0003] In order to eliminate the interference of environmental noise, echo and other factors, the existing technology requires a large amount of data and hardware support to build complex sound source estimation feature classifiers and algorithms. However, although the existing technology improves the sound source estimation to a certain extent, it also increases the computational complexity and reduces the computational efficiency of sound source localization. Therefore, there is currently a lack of a simple and efficient passive sound source direction estimation technology. Summary of the invention

[0004] The present invention provides a method and device for passively locating the direction of a noise source, thereby realizing simpler passive positioning of the noise direction of a single sound source.

[0005] In order to solve the above technical problems, an embodiment of the present invention provides a method for passively locating the direction of a noise source, comprising:

[0006] Acquire the noise source signal collected by each microphone in the microphone array, and determine the time domain sampling value of the instantaneous sound pressure in each noise source signal; wherein one of the microphones corresponds to one of the noise source signals;

[0007] Calculating the sound pressure level of the noise source signal collected by each microphone according to the time domain sampling value of the instantaneous sound pressure in each noise source signal;

[0008] Screening out the noise source signals whose sound pressure levels are greater than a preset threshold as the first signal, and determining the cross-correlation using a generalized cross-correlation phase transformation method according to the first signals of each microphone pair, and taking the maximum cross-correlation value as the first time delay of each microphone pair;

[0009] Determine the angle between each first signal and the corresponding microphone pair according to the first time delay of each group of microphone pairs in combination with a far-field model, and convert the angle into a global angle with a preset layout of the microphone array as a reference;

[0010] A consensus detection method based on angle difference is used to exclude the mirror angles in each of the global angles to obtain the true angle of each microphone pair;

[0011] The real angles of multiple microphone pairs are integrated and weighted to obtain the position of the noise source.

[0012] It can be understood that, compared with the prior art, the method provided by the present invention calculates the sound pressure level by collecting the time domain sampling value of the instantaneous sound pressure in the noise source signal, screens out the noise source signal with a sound pressure level greater than the standard value, and reduces unnecessary direction calculations. In the process of direction estimation, the time difference from the noise source to each pair of microphones is determined by calculating the generalized cross-correlation phase transformation function. The function itself has a certain anti-interference ability, so it can accurately measure the delay of each group of microphone pairs. There is no need to consider the sound wave propagation model. Only the receiving phase difference of the microphone pair and the simple geometric relationship of the propagation path are needed to give the direction positioning of the noise source. It has excellent effects in simple application scenarios and does not require high hardware equipment. The weighted calculation method is used to fuse the real angles of multiple microphones, which improves the accuracy of noise source direction estimation. In addition, the weighted method can be used for three-dimensional noise source direction estimation, which is simple and efficient to calculate and has a wider range of applications.

[0013] Furthermore, the sound pressure level of the noise source signal collected by each microphone is calculated according to the time domain sampling value of the instantaneous sound pressure in each noise source signal, and the specific steps include:

[0014] Calculating the effective sound pressure of each noise source signal according to the time domain sampling value of the instantaneous sound pressure in each noise source signal;

[0015] The sound pressure levels of the noise source signals collected by the microphones are calculated according to the benchmark reference sound pressure and the effective sound pressure of the noise source signals.

[0016] It is understandable that the method provided by the present invention needs to obtain the time domain sampling value of the instantaneous sound pressure of the microphone array. Among them, the calculation of the sound pressure level is to compare with the standard values ​​of environmental noise in various areas of the city during the day and at night. If it exceeds the standard value of different measurement scenarios, it will be judged as noise, and then enter the directional calculation process, otherwise the microphone array will continue to collect noise signals. The noise source signal that meets the specifications is screened out by the sound pressure level, reducing unnecessary direction calculations.

[0017] Further, the method of determining the cross-correlation using a generalized cross-correlation phase transformation method according to the first signal of each microphone pair, and taking the maximum cross-correlation value as the first time delay of each microphone pair, specifically includes:

[0018] Performing Fourier transform on the first signal of each group of microphones to obtain a first observation signal power spectrum and a second observation signal power spectrum, and calculating a cross power spectrum and a phase transformation weighting function between the first observation signal power spectrum and the second observation signal power spectrum;

[0019] Performing inverse Fourier transformation on the cross power spectrum in combination with the phase transformation weighting function to obtain a generalized cross correlation phase transformation weighting function;

[0020] The maximum value of the mutual correlation in the generalized mutual correlation phase transformation weighting function is calculated, and the maximum value of the correlation is used as the first time delay of each group of microphone pairs.

[0021] It can be understood that the method provided by the present invention determines the time difference between two microphones by calculating the generalized cross-correlation phase transformation function, and this function itself has a certain anti-interference ability, so it can accurately measure the time delay of each microphone pair.

[0022] Furthermore, the first delay of each group of microphone pairs is combined with a far-field model to determine the angle between each first signal and the corresponding microphone pair, and convert the angle into a global angle with reference to a preset layout of the microphone array, specifically including:

[0023] According to the far-field model, the angle between each first signal and the corresponding microphone pair is determined in combination with the first delay of each group of microphone pairs and the microphone array element spacing of the preset layout. The specific formula is:

[0024] Wherein, β is the angle between the first signal and the corresponding microphone pair, is the first delay of each group of microphone pairs, c is the speed of sound, and d is the spacing between microphone array elements;

[0025] According to the preset layout of the microphone array, the angle between each microphone pair and the microphone array coordinate system is added to and subtracted from the angle between each first signal and each microphone pair to obtain the global angle with reference to the preset layout of the microphone array.

[0026] It can be understood that the method provided by the present invention does not need to consider the sound wave propagation model. It only needs the receiving phase difference of the microphone pair and the simple geometric relationship of the propagation path to give the direction positioning of the noise source. It has excellent effects in simple application scenarios and does not have high requirements on hardware equipment.

[0027] Furthermore, the consensus detection method based on angle difference is used to exclude the mirror angles in the global angles to obtain the true angles of each microphone pair, which specifically includes:

[0028] The global angles of the first group of microphone pairs are used as the reference group, the global angles of the second group of microphone pairs are used as the verification group, and the global angles of the remaining groups of microphone pairs are used as the comparison groups respectively;

[0029] Calculate the absolute value of the difference between the reference group and the verification group, and use the global angle of the reference group and the global angle of the verification group corresponding to the minimum absolute value as the true angle of the reference group and the true angle of the verification group;

[0030] The absolute values ​​of the differences between the true angle of the reference group and each of the comparison groups are calculated respectively, and the global angle of the comparison group corresponding to the minimum absolute value is used as the true angle of the current comparison group.

[0031] It is understandable that the method provided by the present invention detects the real angle by using a consensus mechanism to exclude the mirror angle. Since the actual placement positions of each group of microphones are different, their mirror angles will also be different, but the real angles are relatively close, so by comparing the comparison group with the reference group one by one, finding the angle with a smaller angle between the comparison group and the reference group, the real angle set can be obtained.

[0032] Furthermore, the fusing of the real angles of the multiple microphone pairs and weighted calculation to obtain the noise source position specifically includes:

[0033] The normalized weight of the angle between each of the first signals and the corresponding microphone is calculated, and the specific formula is:

[0034] Among them, β i is the angle between the i-th first signal and the corresponding microphone pair, and S is the number of the microphone pairs;

[0035] The true angles and the normalized weights of multiple microphone pairs are calculated to obtain the noise source position. The specific formula is:

[0036] Among them, θ i is the true angle of the i-th microphone pair, and α is the position of the noise source.

[0037] It is understandable that the method provided by the present invention can resist the interference of environmental noise by fusing the real angles of multiple microphones, thereby improving the accuracy of noise source direction estimation. In addition, the weighted method can be used for three-dimensional noise source direction estimation, with simple and efficient calculation and wider application range.

[0038] Accordingly, the present invention also provides a device for passively locating the direction of a noise source, the device comprising:

[0039] A data acquisition module, used to obtain the noise source signal collected by each microphone in the microphone array, and determine the time domain sampling value of the instantaneous sound pressure in each noise source signal; wherein one of the microphones corresponds to one of the noise source signals;

[0040] A sound pressure level calculation module, used to calculate the sound pressure level of the noise source signal collected by each microphone according to the time domain sampling value of the instantaneous sound pressure in each noise source signal;

[0041] A time delay calculation module, used to screen out the noise source signals whose sound pressure levels are greater than a preset threshold as the first signal, and determine the cross-correlation using a generalized cross-correlation phase transformation method according to the first signals of each microphone pair, and use the maximum value of the cross-correlation as the first time delay of each microphone pair;

[0042] a direction determination module, configured to determine the angle between each of the first signals and the corresponding microphone pair according to the first time delay of each group of microphone pairs in combination with a far-field model, and convert the angle into a global angle with reference to a preset layout of the microphone array;

[0043] An image elimination module is used to eliminate the image angles in each of the global angles by using a consensus detection method based on angle differences to obtain the true angle of each microphone pair;

[0044] The direction enhancement module is used to fuse the real angles of multiple microphone pairs and obtain the noise source position through weighted calculation.

[0045] It can be understood that, compared with the prior art, the device provided by the present invention first pre-processes the noise source signal, calculates the sound pressure level according to the time domain sampling value of the instantaneous sound pressure in the noise source signal, and screens out the noise source signal with a sound pressure level greater than the standard value, thereby reducing unnecessary direction calculations. A generalized cross-correlation phase transformation method with certain resistance to environmental interference is used to determine the time difference from the noise source to each group of microphone pairs. As a passive positioning method for a single noise source, the device provided by the present invention does not need to consider the sound wave propagation model. It only needs the receiving phase difference of the microphone pair and the geometric relationship of the propagation path possessed by the far-field model to give the direction positioning of the noise source. It has excellent effects in simple application scenarios and does not require high hardware equipment. The angle estimation results of multiple groups of microphone pairs can be comprehensively utilized to improve the accuracy of noise source direction estimation. In addition, the device can be used for three-dimensional noise source direction positioning, with simple and efficient calculations and a wider range of applications.

[0046] Furthermore, the direction determination module includes an angle calculation submodule and a global angle conversion submodule:

[0047] The angle calculation submodule is used to determine the angle between each first signal and the corresponding microphone pair according to the far field model, combined with the first delay of each group of microphone pairs and the microphone array element spacing of the preset layout. The specific formula is:

[0048] Wherein, β is the angle between the first signal and the corresponding microphone pair, is the first delay of each group of microphone pairs, c is the speed of sound, and d is the spacing between microphone array elements;

[0049] The global angle conversion submodule is used to add and subtract the angle between each first signal and each microphone pair to the angle between each microphone pair and the microphone array coordinate system according to the preset layout of the microphone array, so as to obtain the global angle with reference to the preset layout of the microphone array.

[0050] It can be understood that the device provided by the present invention does not need to consider the sound wave propagation model. It only needs the receiving phase difference of the microphone pair and the simple geometric relationship of the propagation path to give the direction positioning of the noise source. It has excellent effects in simple application scenarios and does not have high requirements for hardware equipment.

[0051] Furthermore, the mirror image exclusion module includes a grouping submodule, a first analysis submodule and a second analysis submodule:

[0052] The grouping submodule is used to use the global angles of the first group of microphone pairs as a reference group, the global angles of the second group of microphone pairs as a verification group, and the global angles of the remaining groups of microphone pairs as comparison groups;

[0053] The first analysis submodule is used to calculate the absolute value of the difference between the reference group and the verification group, and use the reference group global angle and the verification group global angle corresponding to the minimum absolute value as the true angle of the reference group and the true angle of the verification group;

[0054] The second analysis submodule calculates the absolute value of the difference between the true angle of the reference group and each of the comparison groups, and uses the global angle of the comparison group corresponding to the minimum absolute value as the true angle of the current comparison group.

[0055] It is understandable that the device provided by the present invention detects the real angle by using a consensus mechanism to exclude the mirror angle. Since the actual placement positions of each group of microphones are different, their mirror angles will also be different, but the real angles are relatively close, so by comparing the comparison group with the reference group one by one, finding the angle with a smaller angle between the comparison group and the reference group, the real angle set can be obtained.

[0056] Furthermore, the direction enhancement module includes a normalized weight calculation submodule and a weighted calculation submodule:

[0057] The normalized weight calculation submodule is used to calculate the normalized weight of the angle between each of the first signals and the corresponding microphone. The specific formula is:

[0058] Among them, β i is the angle between the i-th first signal and the corresponding microphone pair, and S is the number of the microphone pairs;

[0059] The weighted calculation submodule is used to calculate the true angles and the normalized weights of multiple microphone pairs to obtain the noise source position. The specific formula is:

[0060] Among them, θ i is the true angle of the i-th microphone pair, and α is the position of the noise source.

[0061] It is understandable that the device provided by the present invention can resist the interference of environmental noise by fusing the real angles of multiple microphones, thereby improving the accuracy of noise source direction estimation. In addition, the weighted method can be used for three-dimensional noise source direction estimation, with simple and efficient calculation and wider application range. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 : A schematic flow chart of the steps of a method for passively locating the direction of a noise source provided by an embodiment of the present invention;

[0063] Figure 2 : A schematic diagram of the direction of arrival sensing principle of a microphone array provided in an embodiment of the present invention;

[0064] Figure 3 : A schematic diagram of the direction estimation mirror and angle conversion principle provided by an embodiment of the present invention;

[0065] Figure 4 : A schematic diagram of a flow chart of a noise source direction positioning method based on an arrival delay method provided by an embodiment of the present invention;

[0066] Figure 5 : A schematic diagram of a two-dimensional microphone array provided in an embodiment of the present invention;

[0067] Figure 6 : A schematic diagram of a three-dimensional microphone array provided in an embodiment of the present invention;

[0068] Figure 7 : A curve diagram showing the variation of the angle error with the position of the noise source in a three-dimensional scene provided by an embodiment of the present invention;

[0069] Figure 8 : A curve diagram showing the variation of angle estimation error with signal-to-noise ratio in a three-dimensional scene provided by an embodiment of the present invention;

[0070] Fig. 9 : A simulation running time curve diagram of the method for passively locating the direction of a noise source in a three-dimensional scene provided by an embodiment of the present invention;

[0071] Fig.10 : A schematic diagram of the structure of a device for passively locating the direction of a noise source provided by an embodiment of the present invention;

[0072] Fig.11 : A schematic diagram of the internal structure of a direction determination module of a device for passively locating the direction of a noise source provided by an embodiment of the present invention;

[0073] Fig.12 : A schematic diagram of the internal structure of a device for passively locating the direction of a noise source excluding a mirror image module provided by an embodiment of the present invention;

[0074] Fig.13 : A schematic diagram of the internal structure of a direction enhancement module of a device for passively locating the direction of a noise source provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0075] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0076] Embodiment 1

[0077] Please refer to Figure 1 , is a schematic flow chart of a method for passively locating the direction of a noise source provided by an embodiment of the present invention, including the following steps S101-S106, and the specific methods of each step are as follows:

[0078] S101: Acquire a noise source signal collected by each microphone in a microphone array, and determine a time domain sampling value of an instantaneous sound pressure in each noise source signal.

[0079] It should be noted that the microphone array includes but is not limited to measurement microphones or MEMS microphones; the microphone array form includes but is not limited to circular arrays, elliptical arrays, grid arrays, linear arrays and other microphone arrays specially designed for specific application scenarios.

[0080] S102: Calculate the sound pressure level of the noise source signal collected by each microphone according to the time domain sampling value of the instantaneous sound pressure in each noise source signal.

[0081] In this embodiment, the sound pressure level of the noise source signals collected by each microphone is calculated according to the time domain sampling value of the instantaneous sound pressure in each noise source signal. The specific steps include: calculating the effective sound pressure of each noise source signal according to the time domain sampling value of the instantaneous sound pressure in each noise source signal; calculating the sound pressure level of the noise source signals collected by each microphone according to the benchmark reference sound pressure and the effective sound pressure of each noise source signal.

[0082] It is understandable that the method provided by the present invention needs to obtain the time domain sampling value of the instantaneous sound pressure of the microphone array. Among them, the calculation of the sound pressure level is to compare with the standard values ​​of environmental noise in various areas of the city during the day and at night. If it exceeds the standard value of different measurement scenarios, it will be judged as noise, and then enter the directional calculation process, otherwise the microphone array will continue to collect noise signals. The noise source signal that meets the specifications is screened out by the sound pressure level, reducing unnecessary direction calculations.

[0083] S103: Filter out the noise source signals whose sound pressure levels are greater than a preset threshold as the first signal, and determine the cross-correlation using a generalized cross-correlation phase transformation method according to the first signals of each microphone pair, and use the maximum cross-correlation value as the first time delay of each microphone pair.

[0084] In this embodiment, the cross-correlation is determined based on the first signal of each microphone pair using a generalized cross-correlation phase transform method, and the maximum cross-correlation value is used as the first time delay of each microphone pair, specifically including: performing Fourier transform on the first signal of each microphone pair to obtain a first observation signal power spectrum and a second observation signal power spectrum, and calculating the cross-power spectrum and phase transform weighting function of the first observation signal power spectrum and the second observation signal power spectrum; performing inverse Fourier transform on the cross-power spectrum in combination with the phase transform weighting function to obtain a generalized cross-correlation phase transform weighting function; calculating the maximum cross-correlation value in the generalized cross-correlation phase transform weighting function, and using the maximum cross-correlation value as the first time delay of each microphone pair.

[0085] It can be understood that the method provided by the present invention determines the time difference between two microphones by calculating the generalized cross-correlation phase transformation function, and this function itself has a certain anti-interference ability, so it can accurately measure the time delay of each microphone pair.

[0086] S104: Determine the angle between each first signal and the corresponding microphone pair according to the first time delay of each group of microphone pairs in combination with a far-field model, and convert the angle into a global angle with reference to a preset layout of the microphone array.

[0087] In this embodiment, please refer to Figure 2, is a schematic diagram of the direction of arrival sensing principle of a microphone array provided in an embodiment of the present invention, showing the geometric relationship between the noise source and each group of microphone pairs. Figure 3 Where x(n)∈[-1,1] is the first signal of each microphone pair, β is the angle between the first signal and the corresponding microphone pair, and d is the microphone array element spacing. The angle between each first signal and the corresponding microphone pair is determined based on the first time delay of each group of microphone pairs in combination with the far-field model, and the angle is converted into a global angle with the preset layout of the microphone array as a reference, specifically including: according to the far-field model, combined with the first time delay of each group of microphone pairs and the preset layout of the microphone array element spacing, the angle between each first signal and the corresponding microphone pair is determined, and the specific formula is: Wherein, β is the angle between the first signal and the corresponding microphone pair, is the first delay of each group of microphone pairs, c is the speed of sound, and d is the microphone array element spacing.

[0088] In this embodiment, please refer to Figure 3 , is a schematic diagram of the direction estimation mirror and angle conversion principle provided by an embodiment of the present invention, wherein φ i ∈[0,π] represents the angle of the i-th microphone pair relative to the microphone array coordinate system. Since a group of microphone pairs can only detect the angle of the noise source relative to the current microphone pair, it is impossible to determine the left and right directions. Therefore, according to the preset layout of the microphone array, the angle between each group of microphone pairs and the microphone array coordinate system is added and subtracted respectively, to obtain the global angle with reference to the preset layout of the microphone array.

[0089] It can be understood that the method provided by the present invention does not need to consider the sound wave propagation model. It only needs the receiving phase difference of the microphone pair and the simple geometric relationship of the propagation path to give the direction positioning of the noise source. It has excellent effects in simple application scenarios and does not have high requirements on hardware equipment.

[0090] S105: Using a consensus detection method based on angle differences to exclude mirror angles in the global angles, and obtaining the true angles of each microphone pair.

[0091] In this embodiment, the consensus detection method based on angle difference is used to exclude the mirror angles in each global angle to obtain the true angle of each microphone pair, specifically including: taking the global angles of the first group of microphone pairs as the reference group, the global angles of the second group of microphone pairs as the verification group, and the global angles of the remaining groups of microphone pairs as comparison groups; calculating the absolute value of the difference between the reference group and the verification group, and taking the global angle of the reference group and the global angle of the verification group corresponding to the minimum absolute value as the true angle of the reference group and the true angle of the verification group; respectively calculating the absolute value of the difference between the true angle of the reference group and each of the comparison groups, and taking the global angle of the comparison group corresponding to the minimum absolute value as the true angle of the current comparison group.

[0092] It is understandable that the method provided by the present invention detects the real angle by using a consensus mechanism to exclude the mirror angle. Since the actual placement positions of each group of microphones are different, their mirror angles will also be different, but the real angles are relatively close, so by comparing the comparison group with the reference group one by one, finding the angle with a smaller angle between the comparison group and the reference group, the real angle set can be obtained.

[0093] S106: The real angles of the multiple microphone pairs are integrated and weighted calculation is performed to obtain the position of the noise source.

[0094] In this embodiment, the real angles of the fused multiple microphone pairs are weightedly calculated to obtain the noise source position, which specifically includes: calculating the normalized weight of the angle between each first signal and the corresponding microphone, and the specific formula is: Among them, β i is the angle between the i-th first signal and the corresponding microphone pair, and S is the number of the microphone pairs; the true angles and the normalized weights of multiple groups of microphone pairs are calculated to obtain the noise source position, and the specific formula is: Among them, θ i is the true angle of the i-th microphone pair, and α is the position of the noise source.

[0095] It is understandable that the method provided by the present invention can resist the interference of environmental noise by fusing the real angles of multiple microphones, thereby improving the accuracy of noise source direction estimation. In addition, the weighted method can be used for three-dimensional noise source direction estimation, with simple and efficient calculation and wider application range.

[0096] As a preferred option, please refer to Figure 4 , which is a flow chart of a noise source direction positioning method based on the arrival delay method provided by an embodiment of the present invention. The method for passively positioning the direction of a noise source provided by the present invention mainly includes two processes: preprocessing and direction estimation. The sound signal is collected by each microphone in the microphone array and transmitted to the cloud server to complete the sound source direction positioning.

[0097] As a preferred solution, the noise source signal collected by each microphone in the microphone array is obtained, and the time domain sampling value of the instantaneous sound pressure in each noise source signal is determined.

[0098] Specifically, the effective sound pressure of each noise source signal is calculated according to the time domain sampling value of the instantaneous sound pressure in each noise source signal. The specific formula is: Among them, p e ∈R is the effective sound pressure, the unit is Pa, x(n)∈[-1,1] is the time domain sampling value of the instantaneous sound pressure, n=1,…,N represents the time domain sampling number, N∈N + is the number of samples sampled in the time domain.

[0099] Specifically, the sound pressure level of the noise source signal collected by each microphone is calculated according to the reference sound pressure and the effective sound pressure of each noise source signal. The specific formula is: Where L∈R is the sound pressure level, its unit is dB, p REF ∈R is the reference sound pressure, which is generally 2×10 5 Pa.

[0100] Specifically, the noise source signals whose sound pressure levels are greater than a preset threshold are selected as the first signal, and each group of microphones performs Fourier transform on the respective first signals to obtain a first observation signal power spectrum and a second observation signal power spectrum, and calculate the cross power spectrum and phase transformation weighting function of the first observation signal power spectrum and the second observation signal power spectrum, and the cross power spectrum The specific calculation formula is:

[0101] in,[·] * is a conjugate operator; Y1(f)∈C is the first observed signal power spectrum, and Y2(f) is the second observed signal power spectrum;

[0102] Phase shift weighting function θ PHAT The specific calculation formula of (f)∈R is:

[0103] The cross power spectrum is combined with the phase transformation weighting function to perform inverse Fourier transformation to obtain a generalized cross-correlation phase transformation weighting function The specific calculation formula is:

[0104] Among them, F -1 [·] is the inverse Fourier transform, j is the complex number sign;

[0105] The maximum value of the cross-correlation in the generalized cross-correlation phase transform weighting function is calculated, and the specific formula is: The maximum value of the cross-correlation is used as the first time delay of each group of microphone pairs.

[0106] Specifically, based on the first time delay of each group of microphone pairs and combined with the far-field model, the angle between each of the first signals and the corresponding microphone pair is determined, and the angle is converted into a global angle with a preset layout of the microphone array as a reference; a consensus detection method based on angle difference is used to exclude the mirror angles in each of the global angles to obtain the true angle of each microphone pair; the true angles of multiple groups of microphone pairs are fused, and the noise source position is obtained by weighted calculation.

[0107] As a preferred solution, this embodiment is applied to outdoor actual scene related experiments. Figure 5 , is a schematic diagram of a two-dimensional microphone array provided in an embodiment of the present invention, in which the four microphones in the microphone array are arranged in a uniform circular array with a radius of 0.5m. The noise source is placed 5m, 7m and 10m in front of the positive semi-axis of the y-axis of the center of the array. When the noise source signal is collected at each collection point, the collection time is 10 seconds, and the outdoor temperature is about 18°C ​​when the noise source signal is collected. The equipment models used are shown in Table 1.

[0108] equipment microphone Sound Card model KM-2 Measurement Microphone RME Fireface UFX II

[0109] Table 1 Equipment Model

[0110] Table 2 shows the direction estimation error at three different noise source distances. When the noise source distance increases within the microphone's sensing range, the signal propagates closer to the far field, and the direction estimation error decreases, which confirms the feasibility of this method.

[0111] Distance from noise source (m) 5 7 10 Direction estimation error (degrees) 2.17 1.91 1.79

[0112] Table 2 Two-dimensional direction estimation results

[0113] As a preferred solution, this embodiment is applied to software simulation experiments of three-dimensional scenes. Figure 6 , is a schematic diagram of a three-dimensional microphone array provided by an embodiment of the present invention, wherein eight microphones in the microphone array form a three-dimensional rectangular array. The two-dimensional position of the noise source is placed 50m in front of the center of the array on the y-axis, and the height varies with the simulation experiment.

[0114] As a preferred solution, please refer to the software simulation results of the present invention applied to three-dimensional scenes. Figure 7 and Figure 8, respectively, are the angle error versus noise source position curves and the angle estimation error versus signal-to-noise ratio curves in a three-dimensional scene provided by an embodiment of the present invention. It can be seen that when the signal-to-noise ratio is 0, when the vertical height of the sound source from the array plane increases, the azimuth estimation error gradually increases, while the elevation estimation error is not greatly affected. When the vertical height of the sound source from the array plane is fixed at 50 meters, when the signal-to-noise ratio is reduced from -30dB to -20dB and -6dB respectively, both the azimuth estimation error and the elevation estimation error can converge.

[0115] As a preferred option, please refer to Fig. 9 , which is a simulation running time curve of the method for passively locating the direction of the noise source in a three-dimensional scene provided by an embodiment of the present invention. The curve graph shows the computational efficiency of the method: the average computational time for completing 30 simulation experiments is 0.6872s.

[0116] The method provided by the present invention calculates the sound pressure level by collecting the time domain sampling value of the instantaneous sound pressure in the noise source signal, screens out the noise source signal with a sound pressure level greater than the standard value, and reduces unnecessary direction calculations. In the process of direction estimation, the time difference from the noise source to each pair of microphones is determined by calculating the generalized cross-correlation phase transformation function. The function itself has a certain anti-interference ability, so it can accurately measure the delay of each group of microphone pairs. There is no need to consider the sound wave propagation model. Only the receiving phase difference of the microphone pair and the simple geometric relationship of the propagation path are needed to give the direction positioning of the noise source. It has excellent effects in simple application scenarios and does not require high hardware equipment. The weighted calculation method is used to fuse the real angles of multiple microphones, thereby improving the accuracy of noise source direction estimation. In addition, the weighted method can be used for three-dimensional noise source direction estimation, which is simple and efficient to calculate and has a wider range of applications.

[0117] Embodiment 2

[0118] Please refer to Fig.10 , which is a structural schematic diagram of a device for passively locating the direction of a noise source provided in an embodiment of the present invention, including a data acquisition module 201, a sound pressure level calculation module 202, a delay calculation module 203, a direction determination module 204, an image elimination module 205, and a direction enhancement module 206.

[0119] The data acquisition module 201 is used to obtain the noise source signal collected by each microphone in the microphone array, and determine the time domain sampling value of the instantaneous sound pressure in each noise source signal; wherein one microphone corresponds to one noise source signal;

[0120] The sound pressure level calculation module 202 is used to calculate the sound pressure level of the noise source signal collected by each microphone according to the time domain sampling value of the instantaneous sound pressure in each noise source signal;

[0121] The time delay calculation module 203 is used to screen out the noise source signals whose sound pressure levels are greater than a preset threshold as the first signal, and determine the cross-correlation using a generalized cross-correlation phase transformation method according to the first signals of each microphone pair, and use the maximum cross-correlation value as the first time delay of each microphone pair;

[0122] The direction determination module 204 is used to determine the angle between each first signal and the corresponding microphone pair according to the first time delay of each group of microphone pairs in combination with a far-field model, and convert the angle into a global angle with a preset layout of the microphone array as a reference;

[0123] The mirror image exclusion module 205 is used to exclude the mirror image angles in each of the global angles by using a consensus detection method based on angle differences to obtain the true angle of each microphone pair;

[0124] The direction enhancement module 206 is used to fuse the real angles of multiple microphone pairs and perform weighted calculation to obtain the noise source position.

[0125] In this embodiment, the sound pressure level calculation module 202 calculates the effective sound pressure of each noise source signal based on the time domain sampling value of the instantaneous sound pressure in each noise source signal; and calculates the sound pressure level of the noise source signal collected by each microphone based on the benchmark reference sound pressure and the effective sound pressure of each noise source signal.

[0126] It is understandable that the device provided by the present invention needs to obtain the time domain sampling value of the instantaneous sound pressure of the microphone array. The calculation of the sound pressure level is to compare with the standard values ​​of the ambient noise in various areas of the city during the day and at night. If it exceeds the standard value of different measurement scenarios, it is judged as noise, and then the directional calculation process is entered. Otherwise, the microphone array will continue to collect noise signals. The noise source signals that meet the specifications are screened out by the sound pressure level, reducing unnecessary direction calculations.

[0127] In this embodiment, the time delay calculation module 203 performs Fourier transform on the first signal of each microphone pair to obtain a first observation signal power spectrum and a second observation signal power spectrum, and calculates the cross-power spectrum and the phase transformation weighting function of the first observation signal power spectrum and the second observation signal power spectrum; performs inverse Fourier transform on the cross-power spectrum in combination with the phase transformation weighting function to obtain a generalized cross-correlation phase transformation weighting function; calculates the maximum cross-correlation value in the generalized cross-correlation phase transformation weighting function, and uses the maximum cross-correlation value as the first time delay of each microphone pair.

[0128] It can be understood that the device provided by the present invention determines the time difference between two microphones by calculating the generalized cross-correlation phase transformation function, and this function itself has a certain anti-interference ability, so it can accurately measure the time delay of each microphone pair.

[0129] In this embodiment, please refer to Fig.11 , is a schematic diagram of the internal structure of a direction determination module of a device for passively locating the direction of a noise source provided in an embodiment of the present invention, wherein the direction determination module 204 includes an angle calculation submodule 2041 and a global angle conversion submodule 2042.

[0130] The angle calculation submodule 2041 is used to determine the angle between each first signal and the corresponding microphone pair according to the far field model, combined with the first delay of each group of microphone pairs and the microphone array element spacing of the preset layout. The specific formula is: Wherein, β is the angle between the first signal and the corresponding microphone pair, is the first delay of each group of microphone pairs, c is the speed of sound, and d is the microphone array element spacing.

[0131] The global angle conversion submodule 2042 is used to add and subtract the angle between each microphone pair and the microphone array coordinate system to the angle between each microphone pair and the microphone array coordinate system according to the preset layout of the microphone array, so as to obtain the global angle with reference to the preset layout of the microphone array.

[0132] It can be understood that the device provided by the present invention does not need to consider the sound wave propagation model. It only needs the receiving phase difference of the microphone pair and the simple geometric relationship of the propagation path to give the direction positioning of the noise source. It has excellent effects in simple application scenarios and does not have high requirements for hardware equipment.

[0133] In this embodiment, please refer to Fig.12 , which is a schematic diagram of the internal structure of an image exclusion module of a device for passively locating the direction of a noise source provided in an embodiment of the present invention. The image exclusion module 205 includes a grouping submodule 2051, a first analysis submodule 2052 and a second analysis submodule 2053.

[0134] The grouping submodule 2051 is used to use the global angles of the first group of microphone pairs as a reference group, the global angles of the second group of microphone pairs as a verification group, and the global angles of the remaining groups of microphone pairs as comparison groups.

[0135] The first analysis submodule 2052 is used to calculate the absolute value of the difference between the reference group and the verification group, and use the reference group global angle and the verification group global angle corresponding to the minimum absolute value as the true angle of the reference group and the true angle of the verification group.

[0136] The second analysis submodule calculates the absolute value of the difference between the true angle of the reference group and each of the comparison groups, and uses the global angle of the comparison group corresponding to the minimum absolute value as the true angle of the current comparison group.

[0137] It is understandable that the device provided by the present invention detects the real angle by using a consensus mechanism to exclude the mirror angle. Since the actual placement positions of each group of microphones are different, their mirror angles will also be different, but the real angles are relatively close, so by comparing the comparison group with the reference group one by one, finding the angle with a smaller angle between the comparison group and the reference group, the real angle set can be obtained.

[0138] In this embodiment, please refer to Fig.13 , which is a schematic diagram of the internal structure of a direction enhancement module of a device for passively locating the direction of a noise source provided in an embodiment of the present invention. The direction enhancement module 206 includes a normalized weight calculation submodule 2061 and a weighted calculation submodule 2062.

[0139] The normalized weight calculation submodule 2061 is used to calculate the normalized weight of the angle between each first signal and the corresponding microphone. The specific formula is: Among them, β i is the angle between the ith first signal and the corresponding microphone pair, and S is the number of the microphone pairs.

[0140] The weighted calculation submodule 2062 is used to calculate the true angles and the normalized weights of multiple microphone pairs to obtain the noise source position. The specific formula is: Among them, θ i is the true angle of the i-th microphone pair, and α is the position of the noise source.

[0141] It is understandable that the device provided by the present invention can resist the interference of environmental noise by fusing the real angles of multiple microphones, thereby improving the accuracy of noise source direction estimation. In addition, the weighted method can be used for three-dimensional noise source direction estimation, with simple and efficient calculation and wider application range.

[0142] The device provided by the present invention first pre-processes the noise source signal, calculates the sound pressure level according to the time domain sampling value of the instantaneous sound pressure in the noise source signal, and screens out the noise source signal with a sound pressure level greater than the standard value, thereby reducing unnecessary direction calculations. A generalized cross-correlation phase transformation method with a certain degree of resistance to environmental interference is used to determine the time difference from the noise source to each group of microphone pairs. As a passive positioning method for a single noise source, the device provided by the present invention does not need to consider the sound wave propagation model. It only needs the receiving phase difference of the microphone pair and the geometric relationship of the propagation path possessed by the far-field model to give the direction positioning of the noise source. It has excellent effects in simple application scenarios and does not require high hardware equipment. The angle estimation results of multiple groups of microphone pairs can be comprehensively utilized to improve the accuracy of noise source direction estimation. In addition, the device can be used for three-dimensional noise source direction positioning, with simple and efficient calculations and a wider range of applications.

[0143] The specific embodiments described above further illustrate the purpose, technical solutions and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. It is particularly pointed out that for those skilled in the art, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for passively locating the direction of a noise source, characterized in that: include: Acquire the noise source signal collected by each microphone in the microphone array, and determine the time domain sampling value of the instantaneous sound pressure in each noise source signal; wherein one microphone corresponds to one noise source signal; wherein the microphone array is a two-dimensional uniform circular array or a three-dimensional rectangular array; Calculating the sound pressure level of the noise source signal collected by each microphone according to the time domain sampling value of the instantaneous sound pressure in each noise source signal; Screening out the noise source signals whose sound pressure levels are greater than a preset threshold as the first signal, and determining the cross-correlation using a generalized cross-correlation phase transformation method according to the first signals of each microphone pair, and taking the maximum cross-correlation value as the first time delay of each microphone pair; Determine the angle between each first signal and the corresponding microphone pair according to the first time delay of each group of microphone pairs in combination with a far-field model, and convert the angle into a global angle with a preset layout of the microphone array as a reference; A consensus detection method based on angle difference is adopted to exclude the mirror angles in each global angle to obtain the true angle of each microphone pair; the global angles of the first group of microphone pairs are used as the reference group, the global angles of the second group of microphone pairs are used as the verification group, and the global angles of the remaining groups of microphone pairs are used as comparison groups respectively; the absolute value of the difference between the reference group and the verification group is calculated, and the global angle of the reference group and the global angle of the verification group corresponding to the minimum absolute value are used as the true angle of the reference group and the true angle of the verification group; the absolute value of the difference between the true angle of the reference group and each comparison group is calculated respectively, and the global angle of the comparison group corresponding to the minimum absolute value is used as the true angle of the current comparison group; wherein each group of microphone pairs at different placement positions produces different mirror angles; The real angles of multiple microphone pairs are integrated and weighted to obtain the position of the noise source.

2. The method for passively locating the direction of a noise source according to claim 1, characterized in that: The specific steps of calculating the sound pressure level of the noise source signal collected by each microphone according to the time domain sampling value of the instantaneous sound pressure in each noise source signal include: Calculating the effective sound pressure of each noise source signal according to the time domain sampling value of the instantaneous sound pressure in each noise source signal; The sound pressure levels of the noise source signals collected by the microphones are calculated according to the benchmark reference sound pressure and the effective sound pressure of the noise source signals.

3. The method for passively locating the direction of a noise source according to claim 1, characterized in that: The method of determining the cross-correlation using a generalized cross-correlation phase transformation method according to the first signals of each microphone pair, and taking the maximum cross-correlation value as the first time delay of each microphone pair, specifically includes: Performing Fourier transform on the first signal of each group of microphones to obtain a first observation signal power spectrum and a second observation signal power spectrum, and calculating a cross power spectrum and a phase transformation weighting function between the first observation signal power spectrum and the second observation signal power spectrum; Performing inverse Fourier transformation on the cross power spectrum in combination with the phase transformation weighting function to obtain a generalized cross correlation phase transformation weighting function; The maximum value of the mutual correlation in the generalized mutual correlation phase transform weighting function is calculated, and the maximum value of the mutual correlation is used as the first time delay of each group of microphone pairs.

4. The method for passively locating the direction of a noise source according to claim 1, characterized in that: The determining, based on the first time delay of each group of microphone pairs and in combination with a far-field model, an angle between each first signal and a corresponding microphone pair, and converting the angle into a global angle with a preset layout of the microphone array as a reference, specifically includes: According to the far-field model, the angle between each first signal and the corresponding microphone pair is determined in combination with the first delay of each group of microphone pairs and the microphone array element spacing of the preset layout. The specific formula is: Wherein, β is the angle between the first signal and the corresponding microphone pair, is the first delay of each group of microphone pairs, c is the speed of sound, and d is the spacing between microphone array elements; According to the preset layout of the microphone array, the angle between each microphone pair and the microphone array coordinate system is added to and subtracted from the angle between each first signal and each microphone pair to obtain the global angle with reference to the preset layout of the microphone array.

5. The method for passively locating the direction of a noise source according to claim 1, characterized in that: The fusing the real angles of the multiple microphone pairs and obtaining the noise source position by weighted calculation specifically includes: The normalized weight of the angle between each of the first signals and the corresponding microphone is calculated, and the specific formula is: Among them, β i is the angle between the i-th first signal and the corresponding microphone pair, and S is the number of the microphone pairs; The true angles and the normalized weights of multiple microphone pairs are calculated to obtain the noise source position. The specific formula is: Among them, θ i is the true angle of the i-th microphone pair, and α is the position of the noise source.

6. A device for passively locating the direction of a noise source, characterized in that: include: A data acquisition module, used to obtain the noise source signal collected by each microphone in the microphone array, and determine the time domain sampling value of the instantaneous sound pressure in each noise source signal; wherein one microphone corresponds to one noise source signal; wherein the microphone array is a two-dimensional uniform circular array or a three-dimensional rectangular array; A sound pressure level calculation module, used to calculate the sound pressure level of the noise source signal collected by each microphone according to the time domain sampling value of the instantaneous sound pressure in each noise source signal; A time delay calculation module, used to screen out the noise source signals whose sound pressure levels are greater than a preset threshold as the first signal, and determine the cross-correlation using a generalized cross-correlation phase transformation method according to the first signals of each microphone pair, and use the maximum value of the cross-correlation as the first time delay of each microphone pair; a direction determination module, configured to determine the angle between each of the first signals and the corresponding microphone pair according to the first time delay of each group of microphone pairs in combination with a far-field model, and convert the angle into a global angle with reference to a preset layout of the microphone array; A mirror image exclusion module is used to exclude the mirror image angles in each of the global angles by using a consensus detection method based on angle differences to obtain the true angle of each microphone pair; the mirror image exclusion module includes a grouping submodule, a first analysis submodule and a second analysis submodule; the grouping submodule is used to use the global angles of the first group of microphone pairs as a reference group, the global angles of the second group of microphone pairs as a verification group, and the global angles of the remaining groups of microphone pairs as comparison groups; the first analysis submodule is used to calculate the absolute value of the difference between the reference group and the verification group, and use the global angle of the reference group and the global angle of the verification group corresponding to the minimum absolute value as the true angle of the reference group and the true angle of the verification group; the second analysis submodule calculates the absolute value of the difference between the true angle of the reference group and each of the comparison groups, and uses the global angle of the comparison group corresponding to the minimum absolute value as the true angle of the current comparison group; wherein each group of microphone pairs at different placement positions produces different mirror image angles; The direction enhancement module is used to fuse the real angles of multiple microphone pairs and obtain the noise source position through weighted calculation.

7. The device for passively locating the direction of a noise source according to claim 6, characterized in that: The direction determination module includes an angle calculation submodule and a global angle conversion submodule: The angle calculation submodule is used to determine the angle between each first signal and the corresponding microphone pair according to the far field model, combined with the first delay of each group of microphone pairs and the microphone array element spacing of the preset layout. The specific formula is: Wherein, β is the angle between the first signal and the corresponding microphone pair, is the first delay of each group of microphone pairs, c is the speed of sound, and d is the spacing between microphone array elements; The global angle conversion submodule is used to add and subtract the angle between each first signal and each microphone pair to the angle between each microphone pair and the microphone array coordinate system according to the preset layout of the microphone array, so as to obtain the global angle with reference to the preset layout of the microphone array.

8. The device for passively locating the direction of a noise source as claimed in claim 6, characterized in that: The direction enhancement module includes a normalized weight calculation submodule and a weighted calculation submodule: The normalized weight calculation submodule is used to calculate the normalized weight of the angle between each of the first signals and the corresponding microphone. The specific formula is: Among them, β i is the angle between the i-th first signal and the corresponding microphone pair, and S is the number of the microphone pairs; The weighted calculation submodule is used to calculate the true angles and the normalized weights of multiple microphone pairs to obtain the noise source position. The specific formula is: Among them, θ i is the true angle of the i-th microphone pair, and α is the position of the noise source.

Images