A method for identifying the sound source of a ducted fan

Abstract

This invention provides a method for identifying the sound source of a ducted fan. First, a series of microphone arrays are set along the axial direction of the duct to monitor the sound pressure signal. Second, the measured sound pressure signal is transformed to the position of a virtual microphone array that rotates synchronously with the sound source. Next, the transformed sound pressure signal is processed using beamforming, and then focused to obtain the sound source signal. Finally, the sound source distribution is identified based on the sound source signal. This method for identifying the sound source of a ducted fan, by transforming the sound pressure signal measured by a fixed microphone array into the position of a virtual microphone array that rotates synchronously with the sound source, cancels the Doppler effect caused by the rotation of the sound source, thereby reducing the complexity of solving the delay time problem during sound source focusing and facilitating the localization and identification of rotating sound sources within ducts.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of sound source identification, and particularly relates to a ducted fan sound source identification method. BACKGROUND

[0002] The conventional beamforming (CB) algorithm has been widely applied due to its intuitive, fast calculation, high robustness and other advantages, but its shortcomings are that the spatial resolution is limited, the sidelobe level is high, and especially in the high frequency region. In order to overcome the shortcomings of the conventional beamforming algorithm, in recent decades, people have developed a variety of advanced microphone array data processing methods to adapt to various different test requirements, such as orthogonal beamforming algorithm, CLEAN algorithm, DAMAS algorithm, wavenumber beamforming algorithm, robust adaptive beamforming (RAB) algorithm, generalized inverse beamforming algorithm, etc., among which the CLEAN-SC deconvolution method and the DAMAS deconvolution method have been widely applied in aeronautical aerodynamic acoustics and become standard deconvolution algorithms.

[0003] It is an important and potential development direction to apply the conventional beamforming method to the identification and positioning of rotating sound sources, and the sound source identification and positioning method of the ducted fan is still in the primary stage and needs further algorithm development. The current research mainly converts the time domain beamforming method to the rotating coordinate system to identify and test the rotating sound source, and the application research on the sound source positioning of the ducted fan is very limited. The research achievements of the rotating focusing in the free field are also very limited in the development and research of the engine fan sound source identification and positioning. SUMMARY

[0004] In view of the deficiencies in the prior art, the application provides a ducted fan sound source identification method to solve the problem of ducted fan sound source identification.

[0005] The application achieves the above technical object through the following technical means.

[0006] A ducted fan sound source identification method comprises the following steps:

[0007] Step one, a plurality of circles of microphone arrays are arranged along the axial direction of the pipeline to be tested to monitor the sound pressure signals;

[0008] Step two, the measured sound pressure signals are transformed to the virtual microphone array position which rotates synchronously with the sound source;

[0009] Step three, the transformed sound pressure signal is processed based on beamforming method, and then the sound source signal is obtained through focusing processing, and finally the sound source distribution is identified according to the sound source signal.

[0010] Further, in step three, the focusing vector is constructed by the pipe sound modal transfer function, and the signal is focused by using the constructed focusing vector, and the transfer function is:

[0011]

[0012] Where y is the sound source position, z and r are the axial coordinate and radial coordinate of any receiving point in the pipe respectively, ω is the angular frequency of sound pressure, is the axial wave number of (m, n) mode, m is the circumferential modal order, n is the radial modal order, β is the polar angle of the normal vector of the dipole sound source on the ducted fan blade r s , θ s , z s are the radial coordinate, angular coordinate and axial coordinate of the receiving point in the sound source area respectively, Ψ mn is the shape function of (m, n) order modal wave, i is the parameter of Green function, is the normalization constant, K mn is the eigenvalue of the pipe eigenvalue function.

[0013] Further, before step three, the sound pressure signal is denoised to remove the noise in the sound pressure signal unrelated to the duct blade rotation.

[0014] Further, in step two, the sound pressure signal is processed according to the following formula:

[0015]

[0016] Where P Ω，m (t) is the transformed sound pressure signal;

[0017] S1 and S u are weighting factors, which are obtained by the following formula:

[0018]

[0019] S1(t)=1-S u (t)

[0020] Where, is the angle value of the test time sequence, α is the angle interval of the microphone, α = 2π / N c , N c is the number of microphones per circle;

[0021] And are the lower and upper indicators of the sound pressure signal actually measured by the microphone, which are respectively obtained by transforming the modal form m1(m, t) and m u (m, t) of the sound pressure signal measured by the microphone, and the calculation formulas of m1(m, t) and m u (m, t) are as follows:

[0022]

[0023]

[0024] Wherein m is the circumferential modal order of the sound pressure signal measured by the microphone.

[0025] Further, the microphone array is 4 circles, and 8 microphones are uniformly arranged in each circle.

[0026] Further, the microphone array is 2 circles, and 16 microphones are uniformly arranged in each circle.

[0027] The beneficial effects of the present application are:

[0028] (1) The present application provides a ducted fan sound source identification method, wherein the sound pressure signal tested by the fixedly installed microphone array is converted into the position of the virtual microphone array rotating synchronously with the sound source, so as to offset the Doppler effect generated by the rotation of the sound source, and to reduce the complexity in solving the delay time of the sound source focusing, which is helpful for the positioning and identification of the rotating sound source in the pipeline.

[0029] (2) In the ducted fan sound source identification method of the present application, the sound source identification in the free field is expanded to the sound source identification in the pipeline by the acoustic modal focusing vector, so as to realize the ducted fan sound source identification and positioning. BRIEF DESCRIPTION OF DRAWINGS

[0030] Figure 1 It is a schematic diagram of the microphone array of the present application;

[0031] Figure 2 It is a sound source diagram measured by setting 4 circles of microphone array of the present application;

[0032] Figure 3 It is a sound source diagram measured by setting 2 circles of microphone array of the present application. DETAILED DESCRIPTION

[0033] The embodiments of the present application will be described in detail below, and the embodiments described below are exemplary and are intended to explain the present application, and cannot be understood as a limitation of the present application.

[0034] The application provides a ducted fan sound source identification method, which is a sound source identification method in a duct based on adaptive virtual rotation.

[0035] The specific theoretical derivation process of the application is as follows:

[0036] S1, a preset duct is analyzed to obtain the broadband noise generated by the rotation of a rotor:

[0037] S11, an infinitely long duct with uniform axial flow inside is preset, wherein the convection wave equation is:

[0038]

[0039] wherein is a stream derivative related to the average axial flow velocity, c is the sound speed, M is the Mach number, z is the derivative of the independent variable, is the Hamiltonian operator, p is the sound pressure, and t is the parameter in the derivative.

[0040] S12, the broadband noise generated by the rotation of the rotor is represented as a dipole sound source with specific spatial and frequency related characteristics on the surface of the rotor blade. Therefore, the sound source distribution is represented as f(y, tau) = f(r s , theta s - omega t, z s , tau), which represents the rotation of the sound source in the theta direction along the axial direction of the duct at an angular frequency omega, wherein y is the sound source position, tau is the sound emission time at the sound source, r s , theta s , z s are the coordinates of the receiving points in the sound source area. In the absence of a static source, the main broadband noise source is located on the rotor, which is generated by the turbulent boundary layer and the trailing edge interference of the rotor blade.

[0041] S2, the modal wave propagation condition is analyzed:

[0042] S21, set wherein is a unit vector perpendicular to the surface S of the blade (wherein S represents the area integral region), note that f(y, tau) is different from f(y, tau); the time sequence pressure signal p(x, t) received by any receiving point x = (r, theta, z) in the duct (r is the radial coordinate of the point x in the duct, theta is the circumferential direction angle coordinate, and z is the axial coordinate) can be calculated by the Green function solution of the wave equation in the duct:

[0043]

[0044] Where f(y, τ) is the intensity of the dipole sound source distributed on the surface of the rotor blade, the sound source position is a function of the emission time τ, G(x, t|y, τ) is the Green's function, and S(y) is the area integral region.

[0045] For an infinitely long, hard-walled pipe with a uniform axial airflow inside, its Green's function is the result of a superposition of infinitely many modes:

[0046]

[0047] Where i is the parameter of the Green's function, m + m - m is the circumferential modal order, and n is the radial modal order. It is the normalization constant, and ω is the sound pressure angular frequency. K is the axial wavenumber of the (m, n) mode. mn It is the eigenvalue of the pipe's eigenvalue function; Ψ mn It is the shape function of the (m, n) order modal wave. Meaning and Ψ mn Similarly, the superscript asterisk is only used to distinguish between the function variables r and r. s ).

[0048] Ψ mn The expression is:

[0049]

[0050] J m It is a Bessel function of the first kind of order m, σ mn It is the nth fixed value of the m-th order Bessel function, and a is the pipe radius.

[0051] Normalization constant The expression is:

[0052]

[0053] S22, Substituting the Green's function from formula (2) of S21 into the convection wave equation, we obtain the axial wave number of the modal wave:

[0054]

[0055]

[0056] Where k is the eigenvalue of the pipeline eigenvalue function. mn If K is a real number, and the modal wave can propagate without attenuation within the pipe, then at the dimensionless pipe frequency, K mn Written as:

[0057]

[0058] When the frequency is lower than the cutoff frequency, the modal wave decays exponentially along the propagation direction.

[0059] S3, Define the transfer function of the modal wave:

[0060] S31, the normal vector of the dipole sound source on the blade. for:

[0061]

[0062] in yes The coordinates are β, where β is the polar angle.

[0063] Hamiltonian operator In cylindrical coordinates:

[0064]

[0065] in G is a component, and G is the Green's function.

[0066] The above Non-zero part and They are respectively:

[0067]

[0068]

[0069] Substituting the above two expressions into formula (1) of S21 for the time-series pressure signal p(x,t), we get:

[0070]

[0071] g m Let g be the transfer function of the m-th order modal wave propagating from a unit intensity dipole sound source at position y to point x = (r, 0, z). m (y, z, r, ω) simplifies to:

[0072]

[0073] S4, as Figure 1 As shown, several rings of microphone arrays are set on the outside of the pipe along the pipe axis, with each ring of microphones arranged at equal intervals along the circumference of the pipe.

[0074] S41 uses the set microphone array to monitor the sound pressure signal in the pipe.

[0075] S42 performs virtual rotation processing on the sound pressure signal directly measured by the microphone, converting the original sound pressure signal to the position of a virtual microphone array that rotates synchronously with the sound source. This stabilizes the relative spatial position of the microphone array measuring points and the sound source point, canceling the Doppler effect caused by the sound source rotation, and ultimately simplifying the identification and localization of rotating sound sources within the pipe. Specifically:

[0076] set up To test the angle values ​​of the time series, N c If the number of microphones is a uniformly arranged circle, then the angular interval α of the microphones is α = 2π / N. c Therefore, the time-series signal of the virtual microphone can be iterated using the following interpolation method:

[0077]

[0078]

[0079] Where m1(m,t) is the lower index of the actual microphone sound pressure signal mode, m u (m, t) is the upper index of the actual microphone sound pressure signal mode. The weighting coefficients in the interpolation calculation are obtained by the following formula:

[0080]

[0081] S1(t)=1-S u (t)

[0082] Where S u S(t) and S1(t) are weighting factors.

[0083] Therefore, the sound pressure data of the virtual rotating array in the rotating coordinate system is:

[0084]

[0085] Among them, P Ω，m (t) is the sound pressure signal at the virtual rotating array position obtained by the required rotation processing, S1 and S u It is a weighting factor. and These are the lower and upper indices of the sound pressure signal measured by the actual microphone, respectively, and are denoted by m1(m,t) and m... u Obtained by (m, t) transformation.

[0086] S43 suppresses background noise by removing noise from the sound pressure signal that is unrelated to the rotation of the duct blades. This minimizes the interference of background noise on sound source identification during the subsequent construction of the cross-spectral matrix, thereby enhancing the accuracy of sound source identification.

[0087] S44 uses beamforming to process the denoised sound pressure signal, and then uses the pipe acoustic mode transfer function g obtained in S3. m A focusing vector (y, z, r, ω) is constructed, and the signal is focused using this vector, converting it into a sound source signal. Finally, the distribution of sound sources is identified based on the sound source signal.

[0088] II. Effect Test

[0089] A four-ring microphone array was set up, with eight microphones evenly distributed in each ring, and the test was conducted accordingly. Figure 2 The image shows the sound source imaging results at 2000Hz, 2500Hz, 3150Hz, 4000Hz, 5000Hz, and 6300Hz. The noise generated by the fan blades inside the pipe can be clearly seen in the image, especially at 2000Hz and 3150Hz, where it can be clearly seen that the noise source of the blades is mainly concentrated at the blade tip.

[0090] Set up a two-ring microphone array, with 16 microphones evenly distributed in each ring. Figure 3 The image shows the sound source imaging results at 3150Hz, 4000Hz, 5000Hz, and 6300Hz. The sound source distribution of the fan blades inside the duct can be clearly seen from the image, with the dominant sound source located in the blade tip region.

[0091] This invention is not limited to the above-described embodiments. Any obvious improvements, substitutions, or modifications that can be made by those skilled in the art without departing from the essence of this invention are within the scope of protection of this invention.

Claims

1. A ducted fan sound source identification method, characterized in that: step one, a plurality of microphone arrays are arranged along the axial direction of the pipeline to be tested to monitor the sound pressure signal; step two, the measured sound pressure signal is transformed to the virtual microphone array position rotating synchronously with the sound source; step three, the transformed sound pressure signal is processed based on the beamforming method, and then the sound source signal is obtained through focusing processing, and finally the sound source distribution is identified according to the sound source signal; wherein: a focusing vector is constructed by a pipeline acoustic mode transfer function, and the signal is focused by using the constructed focusing vector, and the transfer function is: wherein, m is the circumferential mode order of the sound pressure signal measured by the microphone. Before step three, the sound pressure signal is denoised to remove the noise in the sound pressure signal that is irrelevant to the rotation of the duct blade. In step two, the sound pressure signal is processed according to the following formula: Wherein m is the circumferential mode order of the sound pressure signal measured by the microphone. The microphone array is 4 circles, and 8 microphones are uniformly arranged in each circle. where y is the sound source location, z and r are the axial and radial coordinates of an arbitrary receiving point inside the duct, respectively, and ω is the angular frequency of the sound pressure, is the axial wavenumber of the (m, n) mode, m is the circumferential mode order, n is the radial mode order, β is the polar angle of the normal vector of the dipole sound source on the ducted fan blade r s , θ s , z s are the radial, angular, and axial coordinates of the sound source area region receiving point, respectively, Ψ mn is the shape function of the (m, n) order mode wave, i is the parameter of the Green's function, is the normalization constant, K mn is the eigenvalue of the duct eigenvalue function.

2. The bypass fan sound source identification method of claim 1, wherein: The microphone array is 2 circles, and 16 microphones are uniformly arranged in each circle.

3. The bypass fan sound source identification method of claim 1, wherein: ​ where P Ω,m (t) is the transformed sound pressure signal; S1 and S u are weighting factors, both derived from the following equation: S1(t) = 1 - S u (t) wherein, for testing the angle value of time series, α is the angle interval of microphone, α = 2π / N c , N c is the number of microphone for each circle; and are lower and upper indicators of the sound pressure signal actually measured by the microphone, both of which are transformed from their respective modal forms m1(m, t) and m u (m, t), respectively, m1(m, t) and m u (m, t) are calculated by the following formulas: ​ 4. The bypass fan sound source identification method of claim 1, wherein: ​ 5. The bypass fan sound source identification method of claim 1, wherein: ​

Images