CN1629610A - Method and equipment for detecting and positioning correlated or uncorrelated noise sources - Google Patents

Abstract

This invention relates to a method for detecting and locationing noise sources. Every noise source generates related electric signals to be tested by sound wave or vibration sensors and each sensor emits related electric signals along with time variance, including: fetching the electric signals along with time variance emitted by sensors, each signals Si(t) emitted by the sensor is the sum generated by the noise source, amplifying and filtering the electric signals along with time variance to digitalize the electric signals, compute the function and minimize the function to determine the noise source direction.

Description

The method and apparatus whether relevant noise source is surveyed and located to no matter

Technical field

The present invention relates to utilize sensor noise source is under normal conditions surveyed and to be located, sensor wherein is suitable for the characteristic of noise source.

The present invention relates to noise source is surveyed and located and present method narrow or the broadband emission spectrum, noise source wherein is positioned at one, two or three-dimensional or can select related space each other.

The present invention particularly points out is having best application to following field, promptly to selectively positioning with echo and from for example noise source of vehicle, steamer, aircraft or weapon.

Background technology

In many application scenarios, need and to locate noise source with relative accurate way, thereby take measures to suppress this noise source.In the prior art, known have many solutions to come noise source is carried out the acoustics location.Main known solution is to utilize the signal that acquisition sensor the is sent technology that is mutually related.

The defective that these technology have is, and is responsive especially to the interaction noise that occurs under the environment of survey sensor.What must consider in addition, is that these technology have constituted the adhoc approach that is used for each application scenario under consideration.

Technology in widespread use comprises and has big quantity sensor the antenna of (hundreds of) and carry out large computer system that light beam forms with the aiming assigned direction, thereby increased signal to noise ratio (S/N ratio).This method is not with regard to noise source quantity and anyly between them may association make any hypothesis in advance, and what caused like this finding the solution loses.

Summary of the invention

Therefore, during very little and known or overestimate, just need general method when noise source quantity to surveying and locate in the spatial noise source.

Utilization of the present invention is suitable for the sensor of noise source characteristic, noise source is surveyed and is located and seek to address that need by proposing.This method has lower realization cost.

In order to reach this purpose, method of the present invention comprises:

Get the time dependent electric signal that sends by sensor, each the signal s that sends by sensor _i(t) be the S that transmits by noise source _jAnd;

. the time dependent electric signal of being got is amplified and filters;

With described electric signal digitizing

. with following formula computing function f:

$f(n_1,...n_j,...,n_m)=\frac{\det (<T_k\left(\mathrm{\&omega;}\right),T_1^{*}\left(\mathrm{\&omega;}\right)>k,1=0\mathrm{toM})}{\det (<T_k\left(\mathrm{\&omega;}\right),T_1^{*}\left(\mathrm{\&omega;}\right)>k,1=0\mathrm{toM})}$

Wherein

${\left(T_k\left(\mathrm{\&omega;}\right)\right)}_i=e^{\mathrm{J\&omega;}\frac{<n_k,c_i>}{c}}$

＜.. be scalar product;

..c _iBe the vector between sensor center of gravity and the sensor i position;

..n _jFor determine the unit vector of direction at sensor center of gravity and noise source;

..T ₀=s; And

..c=speed of sound; And

. to the vector of M function f is minimized with respect to j=1, determine noise source direction n _j

Description of drawings

With reference to accompanying drawing, other features of each side as can be known from description given below, wherein accompanying drawing shows the embodiment of the invention and the implementation procedure as non-limiting example.

Fig. 1 is the synoptic diagram of detection method principle of the present invention.

Fig. 2 is the detailed features synoptic diagram of the inventive method.

Embodiment

As can be seen from Figure 1, the inventive method comprises noise source X ₁, X ₂..., X _j..., X _mPosition, wherein j changes in 1 to M scope, and noise source is distributed in the space, and each noise source is all launched corresponding signal S _j, wherein j changes in 1 to M scope.Method of the present invention comprises utilizes sound wave or vibration transducer Y ₁, Y ₂..., Y _i..., Y _NTo noise source X _jPosition, wherein i changes in 1 to N scope, and each sensor sends corresponding time dependent electric signal s ₁, s ₂..., s _i..., s _N

This method comprises gets the time dependent electric signal s that each sensor sends _i(t) and noise source X _jThe signal S that sends _jWith typical value.On the coefficient basis of various noise sources, the signal s that receives at N sensor _i(t) be expressed as follows:

$s_i\left(t\right)=\underset{j=1}{\overset{M}{\mathrm{\&Sigma;}}}{A}_{\mathrm{ij}}{S}_j(t-\frac{r_{\mathrm{ij}}}{c})$

Wherein i=1 is to N, and r _IjBe noise source X _jWith sensor Y _jBetween distance, and C is the speed of sound in the medium around.

Item A _IjExpression is expressed as follows owing to propagating the decay that causes with the sensor sensing coefficient:

A _ij＝B _iC(r _ij)

Wherein i=1 is to N, and j=1 is to M, wherein B _iBe sensor Y _iSensitivity coefficient, and C (r _Ij) be because at distance r _IjThe decay that last propagation causes.

Sensor Y _iBe connected with corresponding electronic unit (not shown), wherein electronic unit is used for its signal that picks up is amplified and low-pass filtering.Sensor is preferably on modulus and the phase place and mates, thus their susceptibility unanimity.Like this, arrive N, B for i=1 _i=G.

Preferably, use aforesaid sensor antenna for convenience, sensor Y _iClose toward each other.Therefore, for noise source at a distance, apart from r _IjBe distance r _jProgression, i.e. sensor and noise source X _jDistance between the center of gravity.Like this, has only C (r _Ij)=C (r _j) time, decay just becomes apart from r _jFunction, wherein, i=1 to N and j=1 to M.

Derive from above:

A _ij＝G.C(r _j)＝a(r _j)

Wherein i=1 is to N, and j=1 is to M, and:

$s_i\left(t\right)=\underset{j=1}{\overset{M}{\mathrm{\&Sigma;}}}a\left(r_j\right)S_j(t-\frac{r_{\mathrm{ij}}}{c})$

Wherein i=1 is to N.

Because noise source X _jAmplitude the unknown, to S _jIn item a (r _j) quadrature, below formula can be write as:

$s_i\left(t\right)=\underset{j=1}{\overset{M}{\mathrm{\&Sigma;}}}{S}_j(t-\frac{r_{\mathrm{ij}}}{c})$

Wherein i=1 is to N.

Utilize Fourier transform, signal s _i(t) expression formula becomes:

Wherein i=1 is to N.

Wherein

With

Be respectively the Fourier transform of s and S, and ω is an angular frequency.

This first formula (1) is the signal that receives and apart from r _Ij, promptly with noise source X _jThe position connect.

As can be seen from Figure 2, can express other and consider relevant relation with how much, wherein said how much considerations make apart from r _IjWith unit vector n _jConnect, this vector has been determined by sensor and has been produced signal S _jThe direction that limits of noise source center of gravity.The position origin autobiography sensor Y of sensor _iThe vector C of position and centre of gravity place thereof _iLimit.To r _IjThe first order further limit, in providing:

(2)r _ij≈r _i-<n _j，c _i>

Wherein i=1 to N and j=1 to M, simultaneously＜.. be scalar product.

Like this, by r _IjReplace with the approximate expression that provides in (2), and to this phase term integration:

$e^{-\mathrm{J\&omega;}\frac{r_j}{c}}$

This formula only depends on The noise source X of magnitude _j, formula (1) can be write as:

Wherein i=1 is to N.

This relation also can be expressed as matrix and vector form:

Wherein, vector T _jI coordinate be:

${\left(T_j\right)}_i=e^{-\mathrm{J\&omega;}\frac{<n_j,c_i>}{c}}$

Wherein i=1 is to N.

Perhaps in fact:

(5)s _i(ω)＝T.S(ω)

Wherein T=has the matrix of general term:

$T_{\mathrm{ij}}=e^{-\mathrm{J\&omega;}\frac{<n_j,c_i>}{c}}$

When having the noise of interpolation to exist, formula (4) becomes:

Wherein B is the noise vector that depends on ω.

Method of the present invention comprises to be determined by vector n _jThe noise source X that limits _jDirection, wherein j=1 is to M.

As noise source X _jBe arbitrarily the time, be association or dereferenced, then at sensor Y _iThe possibility that exists of Gaussian noise provide by following formula:

$b.e^{-a\&Integral;{\left|B\right|}^2.\mathrm{d\&omega;}}$

Wherein a and b are variable.

Like this, according to formula (6), the most probable position of noise source is for making the minimized position of following magnitude:

In other words, s arrives perpendicular to vector T _jThe projection that produces the direction of lineoid is necessary for minimum norm.Wherein j=1 is to M.

This has just formed at vector s and T _jThe parallelepipedon height of last foundation square, described height H is a volume and the ratio of floorage S, that is:

$h=\frac{V}{S}$

Numerical value V and S are expressed as the function of Gramm matrix determinant, wherein element (k l) is made of scalar product:

<T _K，T _l ^*>

T wherein _l ^*For being T _lThe scalar of the value of gripping altogether.

Like this,

S ²=det (＜T _K, T _l ^*K, l=1 is to M)

V ²=det (＜T _K, T _l ^*K, l=1 is to M)

Wherein suppose T ₀=s.

So:

${\left|\right|B\left|\right|}^2=\frac{V^2}{{\mathrm{<figure-callout\; id="2"\; label="S"\; filenames="A20031012129500121.png"\; state="\{\{state\}\}">S</figure-callout>}}^2}$

Perhaps in fact:

f(n ₁，...，n _j，...，n _M)＝‖B‖ ²

This is trivector n _jFunction, this vector only depends on two angle θ _jAnd φ _j, promptly height and orientation depend on angular frequency simultaneously.Any existing relevant noise source s _jSpectrum knowledge also can adopt.For example:

. for the narrow-band noise source, the employing measure makes following function f ₁Minimize:

$\left(7\right)f_1=\underset{k}{\mathrm{\&Sigma;}}{\left|\right|B\left({\mathrm{\&omega;}}_k\right)\left|\right|}^2$

ω wherein _KBe the angular frequency of being concerned about; And

. for the broad-band noise source, the function f relevant below taking measures to make with the relation section ₂Minimize:

(8)f ₂＝∫‖B‖ ²dω

Actually, do not calculate ‖ B ‖ ², and can adopt time one-level best approach, this method comprises denominator s ²Replace with 1.Then as can be seen, suppose the direction n that separates for distinct _jStudy, this separate with obtain by exact method those are approaching.

When at least two noise sources overlapped, denominator can be ignored.This denominator is used for separating in elimination interference when a plurality of directions are consistent.

When the inferior one-level best approach comprised that broadband is handled, it also comprises made following function f ₃Minimize:

(9) f ₃=∫ det (＜T _K, T _l ^*K, l=1 is to M) d ω

This magnitude can be expressed becomes signal s _iAnd s _jCorrelation function r _IjLinear combination, wherein these signals are taken from itself and also are the point of the linear combination that postpones:

$\frac{<n_j,c_i>}{c}$

These correlation functions calculate divided by speed of sound, only are used for postponing, and wherein this delay has the magnitude identical with antenna size.So it is better than calculating on frequency domain on the Fourier transform basis usually to calculate in time domain.

Therefore said method comprises:

. on the antenna that comprises a plurality of sensors (two to ten, best two to five), pick up sound or vibration information;

. amplify to the received signal and filter, thereby limit its frequency spectrum and make sensor coupling in phase place and gain;

. to signal digitalized; And

. utilize following two computing method a) and b) make vector s perpendicular to vowing T _jProjection minimum on the direction:

A) obtain signal s _iFourier transform;

Utilize above-mentioned definite expression formula, computing function f for Gramm matrix determinant ₁, f ₂One of; And

Make function f according to locating noise source quantity ₁, f ₂One of minimize;

B) utilize and to comprise and make function f ₃Minimized shortcut calculation;

Compute associations function r _IjAnd

According to noise source quantity the linear combination of correlation function is minimized.

In case the work of minimizing is finished, just determined the direction n of noise source _jPreferably, also can obtain noise source X again _jFeature.

If N=M, if many sensors are promptly arranged as noise source, then system (5) can be reversed usually.

If N 〉=M then multiply by by on the left side ^tT ^*And derive a square system.Promptly grip transposed matrix T altogether by multiply by.So system (5) just becomes:

^tT ^*.s(ω)＝ ^tT ^*.T.S(ω)

Promptly

(10)S(ω)＝( ^tT ^*.T) ^-1. ^tT ^*.s(ω)

According to formula (10), can calculate signal s _jThereby, draw noise source X _jFeature.

Describing below provides the implementation method of utilizing N sensor to survey a noise source (M=1).

Here it is:

Promptly

For wide band signal source, just a minimized problem is arranged:

Promptly the incidence relation between the signal of measuring is write as γ _Kl:

$\&Integral;{\left|\right|B\left(\mathrm{\&omega;}\right)\left|\right|}^2\mathrm{d\&omega;}=\frac{1}{N}[(N-1).\&Integral;{\left|\right|S\left(\mathrm{\&omega;}\right)\left|\right|}^2\mathrm{dt}-\underset{k\&NotEqual;1}{\underset{k,l}{\mathrm{\&Sigma;}}}{{\mathrm{\&gamma;}}_{\mathrm{kl}}}^{\left(\frac{<n,c_1-c_k>}{c}\right)}]$

So derive and minimize by changing n:

$\underset{k\&NotEqual;1}{\underset{k,l}{\mathrm{\&Sigma;}}}{{\mathrm{\&gamma;}}_{\mathrm{kl}}}^{\left(\frac{<n,c_1-c_k>}{c}\right)}$

Wherein n only depends on N _PParameter, and N _PEqual the size of negative 1 unit in space.

Claims (5)

1. method that noise source is surveyed and located, wherein each noise source is launched corresponding signal S _j, j=1 utilizes sound wave or vibration transducer to survey to M, and wherein each sensor sends corresponding time dependent electric signal s _i, i is positioned at 1 to N scope, and this method comprises:

Get the time dependent electric signal that sends by sensor, each the signal s that sends by sensor _i(t) be the S that transmits by noise source _jAnd;

The time dependent electric signal of being got is amplified and filters;

With following formula computing function f:

$f(n_1,...n_j,...,n_m)=\frac{\det (\&lang;T_k\left(\mathrm{\&omega;}\right),T_1^{*}\left(\mathrm{\&omega;}\right)\&rang;k,1=0\mathrm{toM})}{\det (\&lang;T_k\left(\mathrm{\&omega;}\right),T_1^{*}\left(\mathrm{\&omega;}\right)\&rang;k,1=0\mathrm{toM})}$

Wherein

${\left(T_k\left(\mathrm{\&omega;}\right)\right)}_i=e^{\mathrm{J\&omega;}\frac{\&lang;n_k,c_i\&rang;}{c}}$

＜.. be scalar product;

c _iBe the vector between sensor center of gravity and the sensor i position;

n _jFor determine the unit vector of direction at sensor center of gravity and noise source;

T ₀=s; And

The c=speed of sound; And

To the vector of M function f is minimized with respect to j=1, determine noise source direction n _j

2. according to the method for claim 1, it is characterized in that minimize in order to make the function f when noise source is the narrow-band source, this method comprises:

To the signal s that sends by sensor _i(t) carry out Fourier transform;

Utilize the expression formula of general term determinant of a matrix

<T _k(ω)，T _l ^*(ω)>

Come computing function:

$f_1=\underset{k}{\mathrm{\&Sigma;}}{\left|\right|B{\left(\mathrm{\&omega;}\right)}_k\left|\right|}^2$

Behind the noise source of having selected quantification, make function f ₁Minimize, to determine to select the direction n of noise source _j

3. according to the detection method of claim 1, it is characterized in that minimize in order to make the function f when noise source is broadband source, this method comprises:

To the signal s that sends by sensor _i(t) carry out Fourier transform;

Utilize the expression formula of general term determinant of a matrix

<T _k(ω)，T _l ^*(ω)>

Come computing function:

f ₂＝∫‖B‖ ²dω

After the quantity of having selected noise source to determine, make function f ₁Minimize, to determine to select the direction n of noise source _j

4. according to the detection method of claim 1, it is characterized in that minimize in order to make function f, this method comprises:

The expression formula of function f is simplified, so that following function f ₃Minimize:

f ₃=∫ det (＜T _K, T _l ^*K, l=1 is to M) d ω

Signal calculated s _iAnd s _jFunction gamma is associated _IjAnd

Behind the noise source of having selected quantification, make function f ₃Minimize.

5. according to the detection method of claim 1, it is characterized in that after minimization process finished, this method comprises calculated the noise source vector:

S(ω)＝( ^tT ^*.T) ^-1. ^tT ^*.s(ω)

Thereby draw the noise source feature.

Images