CN103592022B - Adopt the real-time method for sound field separation that acoustic pressure and particle acceleration are measured - Google Patents

Abstract

The invention discloses a kind of real-time method for sound field separation adopting acoustic pressure and particle acceleration to measure, it is characterized in that arranging two measurement plane H and measurement plane H be parallel to each other between target sound source and interference sound source ₁, the acoustic pressure time-domain signal on synchronous acquisition two measurement planes; Utilize the acoustic pressure time-domain signal on two measurement planes, obtain the particle acceleration time-domain signal on measurement plane H by method of finite difference; Acoustic pressure time domain wavenumber spectrum on recycling measurement plane H and particle acceleration time domain wavenumber spectrum, known time-domain pulse response function, isolate the acoustic pressure time domain wavenumber spectrum of target sound source institute's radiation on measurement plane H separately in real time, and then obtain the acoustic pressure time-domain signal of target sound source institute's radiation on measurement plane H separately.The inventive method is without the need to carrying out any inverting and regularization calculation process, and computing velocity is fast, stability is high; The inventive method possesses the ability of real-time sound field separation, can be used for on-the site analysis target sound source under noise environment time become radiation characteristic.

Description

Adopt the real-time method for sound field separation that acoustic pressure and particle acceleration are measured

Technical field

The present invention relates to noise class field method for sound field separation in Speciality of Physics.

Background technology

In Practical Project, target sound source is positioned at the side of measurement plane, and often there is interference sound source at the opposite side of measurement plane, the sound field that these interference sound sources produce have impact on the Measurement accuracy to target sound source institute radiated sound field, therefore needs to adopt certain method for sound field separation the impact of interference sound source to be separated from measurement result.Up to the present, Chinese scholars has proposed multiple method for sound field separation, these methods can be roughly divided into five classes: one is the sound field separation technique based on spatial fourier transform method (SFT): G.V.Frisk etc. proposed to adopt SFT method indirect inspection sea floor first reflection coefficient in 1980, and the two-sided sound field separation established based on SFT method is theoretical; The reflection coefficient of M.Tamura material when the method that G.V.Frisk etc. proposes is used for measuring oblique incidence by nineteen ninety; M.T.Cheng etc. promote the method that M.Tamura proposes, and establish the two measurement sound field separation formulas under cylindrical coordinates, and for realizing the separation of scattering acoustic field.Two is the method for sound field separation based on spherical wave superposition: 1956, and J.Pachner adopts spherical wave method of superposition to achieve the separation of traveling and standing wave sound field in any wave field; G.Weinreich etc. have done further improvement in 1980 to the method that J.Pachner proposes, and the sound field separation established based on two spherical surface measurement is theoretical; In 2013, the two separation method of sphere sound pressure measurement of employing and the separation method of single sphere acoustic pressure plane vibration speed measurement achieved the identification of sound source in little spatial dimension to Y.Braikia etc. respectively.Three is sound field separation technique of Corpus--based Method optimal near-field acoustical holography (SONAH): J.Hald, on the basis studied SONAH, proposes the optimum sound field separation technique of statistics based on two holographic facet sound pressure measurement; On the basis of the method that F.Jacobsen etc. propose at J.Hald, propose the optimum sound field separation technique of statistics based on acoustic pressure plane vibration speed measurement; E.Fernandez-Grande etc. proposed the optimum sound field separation technique of statistics based on double plane vibration speed measurement in 2011.Four is that the sound field separation technique based on boundary element method (BEM): C.Langrenne etc. proposed a kind of two-sided method for sound field separation based on BEM in 2007; Subsequently, E.G.Williams etc. proposed a kind of method for sound field separation based on BEM and acoustic pressure plane vibration speed measurement (Cauchy data) in 2008.Five is sound field separation technique based on ESM that the sound field separation technique based on equivalent source method (ESM): C.X.Bi proposes, and be applicable to the measuring surface of arbitrary shape, and computational stability is good, computational accuracy is high; E.Fernandez-Grande etc. proposed the isolation technics of isolation technics based on the double plane vibration speed measurement of ESM and one side acoustic pressure plane vibration speed measurement in 2012.But above-mentioned method for sound field separation all only achieves the sound field separation under single-frequency or frequency band.To realize the sound field separation under any time, above-mentioned method for sound field separation will be no longer applicable.For realizing the sound field separation under any time, X.Z.Zhang etc. proposed a kind of time domain method for sound field separation based on biplane sound pressure measurement in 2012.In this separation method, the sound pressure signal of each moment target sound source institute radiation needs to be obtained by uncoiling computing, because this uncoiling calculating process has ill-posedness, needs to adopt svd and regularization to carry out stable solution procedure.And the employing of regularization makes this separation method length consuming time, and separation accuracy depends on the Rational choice of regularization parameter to a great extent.

Summary of the invention

For realizing the sound field separation under any time, the invention provides a kind of computing velocity is fast, stability is high employing acoustic pressure and the real-time method for sound field separation that particle acceleration is measured, to obtaining the ability of real-time sound field separation, so for on-the site analysis target sound source under noise environment time become radiation characteristic.

The technical scheme that technical solution problem of the present invention adopts is:

The feature of the real-time method for sound field separation that the present invention adopts acoustic pressure and particle acceleration to measure is carried out as follows:

Step a, at target sound source M _owith interference sound source M _dbetween arrange the measurement plane H and measurement plane H that are parallel to each other ₁; Described target sound source M _owith interference sound source M _dall radiation-curable linear acoustic field signal arbitrarily; At described measurement plane H and measurement plane H ₁above be evenly distributed with M examination network point respectively, measurement plane H and measurement plane H ₁on examination network size identical with examination network point position; With target sound source M _ocenter be that true origin sets up cartesian coordinate system, the xoy coordinate plane in described cartesian coordinate system is parallel to measurement plane H, and z-axis direction is perpendicular to measurement plane H; Examination network point coordinate on measurement plane H is (x, y, z _h), measurement plane H ₁on examination network point coordinate be (x, y, z _h1); Each examination network point (x, y, z on synchronous acquisition measurement plane H _h) acoustic pressure time-domain signal p (x, y, the z at place _h, t) with measurement plane H ₁upper each examination network point (x, y, z _h1) acoustic pressure time-domain signal p (x, y, the z at place _h1, t);

Step b, to carry out finite difference according to formula (1) and obtain each examination network point (x, y, z on measurement plane H _h) particle acceleration time-domain signal a (x, y, the z at place _h, t):

$a(x,y,z_H,t)=-\frac{p(x,y,z_{H1},t)-p(x,y,z_H,t)}{{\mathrm{\ρ}}_0(z_{H1}-z_H)}---\left(1\right)$

In formula (1), t is the time; ρ ₀for Media density;

Step c, for acoustic pressure time-domain signal p (x, y, the z on measurement plane H _h, t) carry out two-dimensional space Fourier transform according to formula (2) and obtain acoustic pressure time domain wavenumber spectrum P (k _x, k _y, z _h, t), to particle acceleration time-domain signal a (x, y, z on measurement plane H _h, t) carry out two-dimensional space Fourier transform according to formula (3) and obtain particle acceleration time domain wavenumber spectrum A (k _x, k _y, z _h, t),

$P(k_x,k_y,z_H,t)={\∫}_{-\∞}^{\∞}{\∫}_{-\∞}^{\∞}p(x,y,z_H,t)e^{j(k_{x}x+k_{y}y)}\mathrm{dxdy}---\left(2\right)$

$A(k_x,k_y,z_H,t)={\∫}_{-\∞}^{\∞}{\∫}_{-\∞}^{\∞}a(x,y,z_H,t)e^{j(k_{x}x+k_{y}y)}\mathrm{dxdy}---\left(3\right)$

In formula (2) and formula (3), j represents imaginary unit; k _x, k _ybe respectively the wave number in x, y direction;

Steps d, structure acoustic pressure time domain wavenumber spectrum P (k _x, k _y, z _h, t), particle acceleration time domain wavenumber spectrum A (k _x, k _y, z _h, t), known time-domain pulse response function h (k _x, k _y, 0, t) with target sound source M _othe acoustic pressure time domain wavenumber spectrum P of institute's radiation on measurement plane H separately _o(k _x, k _y, z _h, the relation t) is such as formula shown in (4):

P _o(k _x,k _y,z _H,t)=0.5[P(k _x,k _y,z _H,t)+A(k _x,k _y,z _H,t)*h(k _x,k _y,0,t)] (4)

In formula (4), " * " represents the convolution algorithm of two functions of time;

By discrete for the time t in formula (4) be t _n=(n-1) Δ t, wherein Δ t is sampling time interval, n=1,2 ..., N, N are total number of sample points, and the discrete form of formula (4) is such as formula shown in (5):

$P_o(k_x,k_y,z_H,t_n)=0.5[P(k_x,k_y,z_H,t_n)+\underset{i=1}{\overset{n}{\mathrm{\Σ}}}A(k_x,k_y,z_H,t_i)h(k_x,k_y,0,t_{n-i+1})]---\left(5\right)$

Step e, for described acoustic pressure time domain wavenumber spectrum P _o(k _x, k _y, z _h, t _n) carry out two-dimensional space Fourier inversion according to formula (6), inscribe target sound source M when isolating each _othe acoustic pressure time-domain signal p of institute's radiation on measurement plane H separately _o(x, y, z _h, t _n):

$p_o(x,y,z_H,t_n)=\frac{1}{{\left(2\mathrm{\π}\right)}^2\mathrm{}}{\∫}_{-\∞}^{\∞}{\∫}_{-\∞}^{\∞}{P}_o(k_x,k_y,z_H,t_n)e^{-j(k_{x}x+k_{y}y)}{\mathrm{dk}}_x{\mathrm{dk}}_y---\left(6\right).$

The feature of the real-time method for sound field separation that the present invention adopts acoustic pressure and particle acceleration to measure also is:

Described measurement plane H and measurement plane H ₁acoustic pressure time-domain signal p (x, y, the z of upper each examination network point _h, t) with p (x, y, z _h1, be t) adopt sound pressure sensor array snapshot to measure to obtain.

Described interference sound source M _dfor noise source, reflection sources or scattering source.

Theoretical model:

At the target sound source M of any linear acoustic field signal of radiation _owith the interference sound source M of any linear acoustic field signal of radiation _dbetween arrange two measurement plane H and measurement plane H be parallel to each other respectively ₁.According to Euler formula, particle acceleration time-domain signal a (x, y, z on measurement plane H _h, acoustic pressure time-domain signal p (x, y, z t) and measured by measurement plane H _h, there is the relation be directly proportional in partial derivative t), shown in (7):

$a(x,y,z_H,t)=-\frac{1}{{\mathrm{\ρ}}_0}\frac{\∂p(x,y,z_H,t)}{\∂z}---\left(7\right)$

In formula (7), ρ ₀for Media density.

Utilize acoustic pressure time-domain signal p (x, y, the z on measurement plane H _h, t) with measurement plane H ₁on acoustic pressure time-domain signal p (x, y, z _h1, t), carry out acoustic pressure time-domain signal p (x, y, the z on finite difference acquisition measurement plane H according to formula (8) _h, partial derivative t):

$\frac{\∂p(x,y,z_H,t)}{\∂z}=\frac{p(x,y,z_H,t)-p(x,y,z_{H1},t)}{z_H-z_{H1}}---\left(8\right)$

Association type (7) and (8) obtain particle acceleration time-domain signal a (x, y, z on measurement plane H _h, t), that is:

$a(x,y,z_H,t)=-\frac{p(x,y,z_H,t)-p(x,y,z_{H1},t)}{{\mathrm{\ρ}}_0(z_H-z_{H1})}---\left(9\right)$

According to the superposition principle of sound wave, acoustic pressure time-domain signal p (x, y, z measured on measurement plane H _h, t) equal target sound source M _othe acoustic pressure time-domain signal p of institute's radiation on measurement plane H _o(x, y, z _h, t) with interference sound source M _dat the acoustic pressure time-domain signal p of measurement plane H institute radiation _d(x, y, z _h, t) sum, shown in (10):

p(x,y,z _H,t)=p _o(x,y,z _H,t)+p _d(x,y,z _H,t) (10)

Because particle acceleration is vector, particle acceleration time-domain signal a (x, y, z that measurement plane H obtains _h, t) should target sound source M be equaled _othe particle acceleration time-domain signal a of institute's radiation on measurement plane H _o(x, y, z _h, t) with interference sound source M _dat the particle acceleration time-domain signal a of measurement plane H institute radiation _d(x, y, z _h, difference t), shown in (11):

a(x,y,z _H,t)=a _o(x,y,z _H,t)-a _d(x,y,z _H,t) (11)

Defined function f (x, y, z, t) about the two-dimensional space Fourier transform of x, y is:

$F(k_x,k_y,z,t)={\∫}_{-\∞}^{\∞}{\∫}_{-\∞}^{\∞}f(x,y,z,t)e^{j(k_{x}x+k_{y}y)}\mathrm{dxdy}---\left(12\right)$

In formula (12), F (k _x, k _y, z, t) and represent the time domain wavenumber spectrum of f (x, y, z, t).

Carry out can obtaining about the two-dimensional space Fourier transform of x, y to formula (10) and formula (11) respectively:

P(k _x,k _y,z _H,t)=P _o(k _x,k _y,z _H,t)+P _d(k _x,k _y,z _H,t) (13)

A(k _x,k _y,z _H,t)=A _o(k _x,k _y,z _H,t)-A _d(k _x,k _y,z _H,t) (14)

From acoustic pressure and the particle acceleration relation at time domain wavenumber domain:

P _o(k _x,k _y,z _H,t)=A _o(k _x,k _y,z _H,t)*h(k _x,k _y,0,t)(15)

P _d(k _x,k _y,z _H,t)=A _d(k _x,k _y,z _H,t)*h(k _x,k _y,0,t) (16)

In formula (15) and formula (16), h (k _x, k _y, 0, be t) the time-domain pulse response function between acoustic pressure and particle acceleration, its expression formula is:

$h(k_x,k_y,0,t)={\mathrm{\ρ}}_0{\mathrm{cJ}}_0\left(\sqrt{c^2(k_x^2+k_y^2)t}\right)H\left(t\right)---\left(17\right)$

In formula (17), c is the velocity of sound, J ₀for exponent number is the Bessel function of the first kind of 0, H (t) is Heaviside function.

Formula (16) is substituted in formula (13) and obtains:

P(k _x,k _y,z _H,t)=P _o(k _x,k _y,z _H,t)+A _d(k _x,k _y,z _H,t)*h(k _x,k _y,0,t) (18)

By formula (14) both sides respectively with h (k _x, k _y, 0, t) do convolution and obtain:

A(k _x,k _y,z _H,t)*h(k _x,k _y,0,t)=A _o(k _x,k _y,z _H,t)*h(k _x,k _y,0,t)-A _d(k _x,k _y,z _H,t)*h(k _x,k _y,0,t) (19)

Formula (15) is substituted in formula (19) and obtains:

A(k _x,k _y,z _H,t)*h(k _x,k _y,0,t)=P _o(k _x,k _y,z _H,t)-A _d(k _x,k _y,z _H,t)*h(k _x,k _y,0,t) (20)

Formula (18) is added can obtains further with formula (20):

P _o(k _x,k _y,z _H,t)=0.5[P(k _x,k _y,z _H,t)+A(k _x,k _y,z _H,t)*h(k _x,k _y,0,t)] (21)

In formula (21), P (k _x, k _y, z _h, t) with A (k _x, k _y, z _h, t) can respectively by the acoustic pressure time-domain signal p (x, y, the z that record _h, the particle acceleration time-domain signal a (x, y, the z that t) and by finite difference obtain _h, t) carry out two-dimensional space Fourier transform and obtain, h (k _x, k _y, 0, t) be known function.Target sound source M is inscribed when utilizing formula (21) can isolate each in real time according to the following procedure _othe acoustic pressure time domain wavenumber spectrum P of institute's radiation on measurement plane H separately _o(k _x, k _y, z _h, t):

By discrete for the time t in formula (21) be t _n=(n-1) Δ t, wherein Δ t is sampling time interval, n=1,2 ..., N, N are total number of sample points.

When getting n=1, separation obtains:

$P_o(k_x,k_y,z_H,t_1)=0.5[P(k_x,k_y,z_H,t_1)+\underset{i=1}{\overset{1}{\mathrm{\Σ}}}A(k_x,k_y,z_H,t_1)h(k_x,k_y,0,t_{1-i+1})]---\left(22\right)$

When getting n=2, separation obtains:

$P_o(k_x,k_y,z_H,t_2)=0.5[P(k_x,k_y,z_H,t_2)+\underset{i=1}{\overset{2}{\mathrm{\Σ}}}A(k_x,k_y,z_H,t_i)h(k_x,k_y,0,t_{2-i+1})]---\left(23\right)$

……

When getting n=N, separation obtains:

$P_o(k_x,k_y,z_H,t_N)=0.5[P(k_x,k_y,z_H,t_N)+\underset{i=1}{\overset{N}{\mathrm{\Σ}}}A(k_x,k_y,z_H,t_i)h(k_x,k_y,0,t_{N-i+1})]---\left(24\right)$

From the detachment process of formula (22) to formula (24), t be isolated _nthe P in moment _o(k _x, k _y, z _h, t _n), only need to utilize t _np (the k in moment _x, k _y, z _h, t _n) and t _i(i=1,2 ..., the n) A (k in moment _x, k _y, z _h, t _i), and P (k _x, k _y, z _h, t _n) and A (k _x, k _y, z _h, t _i) by the acoustic pressure time-domain signal p (x, y, the z that record _h, the particle acceleration time-domain signal a (x, y, the z that t) and by finite difference obtain _h, t) carry out two-dimensional space Fourier transform and obtain.Therefore, once record and descend measurement plane H and measurement plane H sometime ₁on acoustic pressure time-domain signal, the acoustic pressure time domain wavenumber spectrum in this moment can be isolated, realize the real-time separation of sound field.

To isolated time domain wavenumber spectrum P _o(k _x, k _y, z _h, t _n) carry out two-dimensional space Fourier inversion according to formula (25), inscribe target sound source M when final acquisition is each _othe acoustic pressure time-domain signal p of institute's radiation on measurement plane H separately _o(x, y, z _h, t _n)

$P_o(x,y,z_H,t_n)=\frac{1}{{\left(2\mathrm{\π}\right)}^2}{\∫}_{-\∞}^{\∞}{\∫}_{-\∞}^{\∞}{P}_o(k_x,k_y,z_H,t_n)e^{-j(k_{x}x+k_{y}y)}d{k}_{x}d{k}_y---\left(25\right)$

Compared with the prior art, beneficial effect of the present invention is embodied in:

1, the detachment process of the inventive method is a simple forward solution procedure, does not need to carry out any inverting and regularization calculation process, and thus computing velocity is fast, stability is high.

2, the inventive method possesses the ability of real-time sound field separation, be thus more suitable for on-the site analysis target sound source under noise environment time become radiation characteristic.

Accompanying drawing explanation

Fig. 1 is the real-time method for sound field separation schematic diagram that the present invention adopts acoustic pressure and particle acceleration measurement;

The experiment sound source that Fig. 2 (a) is the inventive method, measurement plane location distribution plan;

The experiment measuring planar mesh schematic diagram that Fig. 2 (b) is the inventive method;

The experiment measuring point A place acoustic pressure time-domain signal figure that Fig. 3 (a) is the inventive method;

The experiment measuring point B place acoustic pressure time-domain signal figure that Fig. 3 (b) is the inventive method;

The experiment measuring point C place acoustic pressure time-domain signal figure that Fig. 3 (c) is the inventive method;

Fig. 4 (a) tests the numeric distribution figure of the phase valuation factor Ep obtained for the inventive method;

Fig. 4 (b) tests the numeric distribution figure of the amplitude evaluation points Ea obtained for the inventive method;

Fig. 5 (a) is the inventive method t=9.38ms moment target sound source M _othe measurement sonic pressure field p of institute's radiation on measurement plane H separately _m;

Fig. 5 (b) is the inventive method t=10.78ms moment target sound source M _othe measurement sonic pressure field p of institute's radiation on measurement plane H separately _m;

The compound voice that Fig. 5 (c) is the inventive method t=9.38ms moment is had a meeting, an audience, etc. well under one's control p _c;

The compound voice that Fig. 5 (d) is the inventive method t=10.78ms moment is had a meeting, an audience, etc. well under one's control p _c;

Fig. 5 (e) adopts the isolated sonic pressure field p of the inventive method for the t=9.38ms moment _e;

Fig. 5 (f) adopts the isolated sonic pressure field p of the inventive method for the t=10.78ms moment _e.

Embodiment

See Fig. 1, the real-time method for sound field separation that the present embodiment adopts acoustic pressure and particle acceleration to measure isolates target sound source M as follows _othe acoustic pressure time-domain signal of institute's radiation on measurement plane H separately:

Step a, at target sound source M _owith interference sound source M _dbetween arrange the measurement plane H and measurement plane H that are parallel to each other ₁; Described target sound source M _owith interference sound source M _dall radiation-curable linear acoustic field signal arbitrarily; At described measurement plane H and measurement plane H ₁above be evenly distributed with M examination network point respectively, measurement plane H and measurement plane H ₁on examination network size identical with examination network point position; With target sound source M _ocenter be that true origin sets up cartesian coordinate system, the xoy coordinate plane in described cartesian coordinate system is parallel to measurement plane H, and z-axis direction is perpendicular to measurement plane H; Examination network point coordinate on measurement plane H is (x, y, z _h), measurement plane H ₁on examination network point coordinate be (x, y, z _h1); Each examination network point (x, y, z on synchronous acquisition measurement plane H _h) acoustic pressure time-domain signal p (x, y, the z at place _h, t) with measurement plane H ₁upper each examination network point (x, y, z _h1) acoustic pressure time-domain signal p (x, y, the z at place _h1, t).

Step b, to carry out finite difference according to formula (26) and obtain each examination network point (x, y, z on measurement plane H _h) particle acceleration time-domain signal a (x, y, the z at place _h, t):

$a(x,y,z_H,t)=-\frac{p(x,y,z_{H1},t)-p(x,y,z_H,t)}{{\mathrm{\ρ}}_0(z_{H1}-z_H)}---\left(26\right)$

In formula (26), t is the time; ρ ₀for Media density.

Step c, for acoustic pressure time-domain signal p (x, y, the z on measurement plane H _h, t) carry out two-dimensional space Fourier transform according to formula (27) and obtain acoustic pressure time domain wavenumber spectrum P (k _x, k _y, z _h, t), to particle acceleration time-domain signal a (x, y, z on measurement plane H _h, t) carry out two-dimensional space Fourier transform according to formula (28) and obtain particle acceleration time domain wavenumber spectrum A (k _x, k _y, z _h, t),

$P(k_x,k_y,z_H,t)={\∫}_{-\∞}^{\∞}{\∫}_{-\∞}^{\∞}p(x,y,z_H,t)e^{j(k_{x}x+k_{y}y)}\mathrm{dxdy}---\left(27\right)$

$A(k_x,k_y,z_H,t)={\∫}_{-\∞}^{\∞}{\∫}_{-\∞}^{\∞}a(x,y,z_H,t)e^{j(k_{x}x+k_{y}y)}\mathrm{dxdy}---\left(28\right)$

In formula (27) and formula (28), j represents imaginary unit; k _x, k _ybe respectively the wave number in x, y direction.

Steps d, structure acoustic pressure time domain wavenumber spectrum P (k _x, k _y, z _h, t), particle acceleration time domain wavenumber spectrum A (k _x, k _y, z _h, t), known time-domain pulse response function h (k _x, k _y, 0, t) with target sound source M _othe acoustic pressure time domain wavenumber spectrum P of institute's radiation on measurement plane H separately _o(k _x, k _y, z _h, the relation t) is such as formula shown in (29):

P _o(k _x,k _y,z _H,t)=0.5[P(k _x,k _y,z _H,t)+A(k _x,k _y,z _H,t)*h(k _x,k _y,0,t)] (29)

In formula (29), " * " represents the convolution algorithm of two functions of time.

By discrete for the time t in formula (29) be t _n=(n-1) Δ t, wherein Δ t is sampling time interval, n=1,2 ..., N, N are total number of sample points.The discrete form of formula (29) is such as formula shown in (30)

$P_o(k_x,k_y,z_H,t_n)=0.5[P(k_x,k_y,z_H,t_n)+\underset{i=1}{\overset{n}{\mathrm{\Σ}}}A(k_x,k_y,z_H,t_i)h(k_x,k_y,0,t_{n-i+1})]---\left(30\right)$

Step e, for described acoustic pressure time domain wavenumber spectrum P _o(k _x, k _y, z _h, t _n) carry out two-dimensional space Fourier inversion according to formula (31), inscribe target sound source M when isolating each _othe acoustic pressure time-domain signal p of institute's radiation on measurement plane H separately _o(x, y, z _h, t _n)

$p_o(x,y,z_H,t_n)=\frac{1}{{\left(2\mathrm{\π}\right)}^2}{\∫}_{-\∞}^{\∞}{\∫}_{-\∞}^{\∞}{P}_o(k_x,k_y,z_H,t_n)e^{-j(k_{x}x+k_{y}y)}{\mathrm{dk}}_x{\mathrm{dk}}_y---\left(31\right)$

Measurement plane H and measurement plane H ₁acoustic pressure time-domain signal p (x, y, the z of upper each examination network point _h, t) with p (x, y, z _h1, be t) adopt sound pressure sensor array snapshot to measure to obtain.Interference sound source M _dfor noise source, reflection sources or scattering source.

The inspection of method:

Arrange that in the side of measurement plane H and measurement plane H1 audio amplifier S1 and audio amplifier S3 is as target sound source M _o, arrange that in other side audio amplifier S2 is as interference sound source M _d.Adopt the inventive method by target sound source M on measurement plane H _othe acoustic pressure time-domain signal of independent institute radiation is separated, and compares with its measurement acoustic pressure time-domain signal.

In this experiment, target sound source M _o(comprising audio amplifier S1 and audio amplifier S3) and interference sound source M _d(audio amplifier S2) all radiation cosine modulation signal, its expression formula is:

$s\left(t\right)=\cos \left(2\mathrm{\π}{f}_{0}t\right)e^{[-{10}^6{(t-0.005)}^2/2]}---\left(32\right)$

In formula (32), frequency f ₀=800Hz.

Position relationship between measurement plane H and H1 and sound source is see Fig. 2 (a), and measurement plane H is positioned in the plane of z=0m, measurement plane H ₁be positioned in the plane of z=-0.02m.The cone center of audio amplifier S1 and audio amplifier S3 lays respectively at (0.1m, 0.45m, 0.145m) and (0.55m, 0.35m, 0.14m) place, and the cone of audio amplifier S2 is centrally located at (0.4m, 0.45m ,-0.14m) place.The size of measurement plane H and H1 is all 0.7m × 0.7m, and it distributes all equably 15 × 15 measurement points, see Fig. 2 (b).Time-domain signal sample frequency is 25600Hz, and sampling number is 512.

For the sound field separation effect of inspection the inventive method in time domain, measurement plane H have chosen three measurement points, i.e. measurement point A, measurement point B and measurement point C, its position is respectively A (0.1m, 0.45m, 0m), B (0.4m, 0.45m, 0m) with C (0.55m, 0.35m, 0m), wherein measurement point A and measurement point C faces target sound source M respectively _oaudio amplifier S1 and audio amplifier S3, measurement point B face interference sound source M _daudio amplifier S2.Fig. 3 (a), Fig. 3 (b) and Fig. 3 (c) corresponding measurement point A, measurement point B and measurement point C respectively, in figure, curve a represents target sound source M _othe measurement acoustic pressure time-domain signal of independent institute radiation, in figure, curve b represents and comprises target sound source M _owith interference sound source M _dthe mixing acoustic pressure time-domain signal of institute's radiation, in figure, curve c represents the isolated acoustic pressure time-domain signal of employing the inventive method, and the curve a in comparison diagram and curve b can find out, interference sound source M _dto target sound source M _oon measurement plane H, the acoustic pressure time-domain signal of institute's radiation causes larger interference; Comparison curves a and curve c can find out, adopts the inventive method can eliminate interference sound source M well _dimpact, thus isolate target sound source M _othe acoustic pressure time-domain signal of institute's radiation on measurement plane H separately.

In order to evaluate the separating effect of the inventive method more objectively, define two evaluation points at this, their expression formula is respectively:

$\mathrm{Ep}(x,y,z)=\frac{<p_m(x,y,z,t)p_e(x,y,z,t)>}{\sqrt{<p_m^2(x,y,z,t)><p_e^2(x,y,z,t)>}}---\left(33\right)$

$\mathrm{Ea}(x,y,z)=\frac{|\sqrt{<p_m^2(x,y,z,t)>}-\sqrt{<p_e^2(x,y,z,t)>}|}{\sqrt{<p_m^2(x,y,z,t)>}}---\left(34\right)$

In formula (33) and formula (34), <> represents and averages, and subscript " m " represents the measurement sound pressure level of the independent radiation of target sound source, and subscript " e " represents and is separated sound pressure level.Evaluation points Ep is used to the measurement sound pressure level weighing target sound source independent radiation and the phase error be separated between sound pressure level, when Ep value the closer to 1 time, phase error is less.Evaluation points Ea is used to the measurement sound pressure level weighing target sound source independent radiation and the amplitude error be separated between sound pressure level, when Ea value the closer to 0 time, amplitude error is less.Utilization formula (33) and (34) calculate Ep and Ea of each measurement point on measurement plane H respectively.Fig. 4 (a) represents the Ep value of all measurement points, and in figure, value of contour is 0.9.Fig. 4 (b) represents the Ea value of all measurement points, and in figure, value of contour is 0.1.See Fig. 4 (a) and Fig. 4 (b), in most of measurement point, no matter be phase place or amplitude, measure sound pressure level and be separated sound pressure level all in consistent manner better, just lower slightly at measurement plane edge degree of agreement.

For inspection the inventive method is in the sound field separation effect of spatial domain, have chosen two moment t=9.38ms and t=10.78ms.Fig. 5 (a) and Fig. 5 (b) is respectively 9.38ms and 10.78ms moment target sound source M _othe measurement sonic pressure field p of institute's radiation on measurement plane H separately _m, the compound voice that Fig. 5 (c) and Fig. 5 (d) are respectively 9.38ms and the 10.78ms moment is had a meeting, an audience, etc. well under one's control p _c(comprise target sound source M _owith interference sound source M _dthe sonic pressure field of institute's radiation), Fig. 5 (e) and Fig. 5 (f) are respectively 9.38ms and the 10.78ms moment and adopt the isolated sonic pressure field p of the inventive method _e.Comparison diagram 5 (a) and Fig. 5 (c), Fig. 5 (b) can find out with Fig. 5 (d), interference sound source M _dto target sound source M _othe sonic pressure field of institute's radiation on measurement plane H causes serious interference separately; Comparison diagram 5 (a) and Fig. 5 (e), Fig. 5 (b) can find out with Fig. 5 (f), adopt the inventive method can eliminate interference sound source M preferably _dimpact, thus obtain target sound source M _othe sonic pressure field of institute's radiation on measurement plane H separately.

Claims (3)

1. the real-time method for sound field separation adopting acoustic pressure and particle acceleration to measure, is characterized in that carrying out as follows:

Step a, at target sound source M _owith interference sound source M _dbetween arrange the measurement plane H and measurement plane H that are parallel to each other ₁; Described target sound source M _owith interference sound source M _dall can radiation arbitrarily linear acoustic field signal; At described measurement plane H and measurement plane H ₁above be evenly distributed with M examination network point respectively, measurement plane H and measurement plane H ₁on examination network size identical with examination network point position; With target sound source M _ocenter be that true origin sets up cartesian coordinate system, the xoy coordinate plane in described cartesian coordinate system is parallel to measurement plane H, and z-axis direction is perpendicular to measurement plane H; Examination network point coordinate on measurement plane H is (x, y, z _h), measurement plane H ₁on examination network point coordinate be (x, y, z _h1); Each examination network point (x, y, z on synchronous acquisition measurement plane H _h) acoustic pressure time-domain signal p (x, y, the z at place _h, t) with measurement plane H ₁upper each examination network point (x, y, z _h1) acoustic pressure time-domain signal p (x, y, the z at place _h1, t);

Step b, to carry out finite difference according to formula (1) and obtain each examination network point (x, y, z on measurement plane H _h) particle acceleration time-domain signal a (x, y, the z at place _h, t):

$a(x,y,z_H,t)=-\frac{p(x,y,z_{H1},t)-p(x,y,z_H,t)}{{\mathrm{\&rho;}}_0(z_{H1}-z_H)}---\left(1\right)$

In formula (1), t is the time; ρ ₀for Media density;

Step c, for acoustic pressure time-domain signal p (x, y, the z on measurement plane H _h, t) carry out two-dimensional space Fourier transform according to formula (2) and obtain acoustic pressure time domain wavenumber spectrum P (k _x, k _y, z _h, t), to particle acceleration time-domain signal a (x, y, z on measurement plane H _h, t) carry out two-dimensional space Fourier transform according to formula (3) and obtain particle acceleration time domain wavenumber spectrum A (k _x, k _y, z _h, t),

$P(k_x,k_y,z_H,t)={\&Integral;}_{-\&infin;}^{\&infin;}{\&Integral;}_{-\&infin;}^{\&infin;}p(x,y,z_H,t)e^{j(k_{x}x+k_{y}y)}\mathrm{dxdy}---\left(2\right)$

$A(k_x,k_y,z_H,t)={\&Integral;}_{-\&infin;}^{\&infin;}{\&Integral;}_{-\&infin;}^{\&infin;}a(x,y,z_H,t)e^{j(k_{x}x+k_{y}y)}\mathrm{dxdy}---\left(3\right)$

In formula (2) and formula (3), j represents imaginary unit; k _x, k _ybe respectively the wave number in x, y direction;

Steps d, structure acoustic pressure time domain wavenumber spectrum P (k _x, k _y, z _h, t), particle acceleration time domain wavenumber spectrum A (k _x, k _y, z _h, t), known time-domain pulse response function h (k _x, k _y, 0, t) with target sound source M _othe acoustic pressure time domain wavenumber spectrum P of institute's radiation on measurement plane H separately _o(k _x, k _y, z _h, the relation t) is such as formula shown in (4):

P _o(k _x,k _y,z _H,t)＝0.5[P(k _x,k _y,z _H,t)+A(k _x,k _y,z _H,t)*h(k _x,k _y,0,t)] (4)

In formula (4), " * " represents the convolution algorithm of two functions of time;

By discrete for the time t in formula (4) be t _n=(n-1) Δ t, wherein Δ t is sampling time interval, n=1,2 ..., N, N are total number of sample points, and the discrete form of formula (4) is such as formula shown in (5):

$P_o(k_x,k_y,z_H,t_n)=0.5[P(k_x,k_y,z_H,t_n)+\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}A(k_x,k_y,z_H,t_i)h(k_x,k_y,0,t_{n-i+1})]---\left(5\right)$

Step e, for described acoustic pressure time domain wavenumber spectrum P _o(k _x, k _y, z _h, t _n) carry out two-dimensional space Fourier inversion according to formula (6), inscribe target sound source M when isolating each _othe acoustic pressure time-domain signal p of institute's radiation on measurement plane H separately _o(x, y, z _h, t _n):

$p_o(x,y,z_H,t_n)=\frac{1}{{\left(2\mathrm{\&pi;}\right)}^2}{\&Integral;}_{-\&infin;}^{\&infin;}{\&Integral;}_{-\&infin;}^{\&infin;}{P}_o(k_x,k_y,z_H,t_n)e^{-j(k_{x}x+k_{y}y)}d{k}_{x}d{k}_y---\left(6\right).$

2. the real-time method for sound field separation of employing acoustic pressure according to claim 1 and particle acceleration measurement, is characterized in that: described measurement plane H and measurement plane H ₁acoustic pressure time-domain signal p (x, y, the z of upper each examination network point _h, t) with p (x, y, z _h1, be t) adopt sound pressure sensor array snapshot to measure to obtain.

3. the real-time method for sound field separation of employing acoustic pressure according to claim 1 and particle acceleration measurement, is characterized in that: described interference sound source M _dfor noise source, reflection sources or scattering source.