CN103648066A - Loudness control active noise control system and method thereof - Google Patents

Abstract

The invention discloses a loudness control active noise control system and a method thereof, wherein the system includes I secondary sources, J microphones, a reference signal sensor, a minimum mean square algorithm module, a sound transmission channel model and a loudness filter, and is characterized in that the reference signal sensor is connected onto the loudness filter and the I secondary sources respectively; the loudness filter is connected onto the sound transmission channel model; the sound transmission channel model is connected onto the minimum mean square algorithm module; and the J microphones are connected onto the minimum mean square algorithm module via the loudness filter. According to the invention, frequency components sensitive to human ears in an error sensor signal can be appropriately amplified, and insensitive components can be attenuated, so that the system is capable of increasing the noise control over sensitive components and auditory subjective comfortableness of active noise control can be improved.

Description

A kind of loudness is controlled Active Noise Control system and method thereof

Technical field

The present invention relates to Active Noise Control technology, particularly a kind of loudness is controlled Active Noise Control system and method thereof.

Background technology

Without source noise, control by acoustical material or acoustic construction and sound wave interaction and consume acoustic energy and noise reduction, high-frequency noise is had to better control effect, but undesirable to low-frequency noise effect, and quality, the volume of the acoustical material adopting (structure) are larger, increased quality, volume and the cost of controlled device.Active Noise Control is that a kind of secondary source of introducing sends that amplitude is identical, the sound wave of single spin-echo is offset former noisy technology.With without source noise, control and to compare, Active Noise Control have advantages of the bandwidth of control, low frequency control effective, lightweight, control with strong points.

The aims of systems of tradition Active Noise Control is that the acoustic pressure quadratic sum of certain or several positions that make that error microphone records is minimum.But acoustic pressure size is not quite identical to sound impression with people's ear, the sound that sound pressure level is less may sound louder more uncomfortable than the larger sound of another sound pressure level, its reason be the sense of hearing form be one with subjectivity, complicated physiology and mental process, sound pressure level only shows that sound, in the power of physics meaning, can not reflect more auditory properties.

The new trend that current noise is controlled is that control theory just turns to from single noise reduction the sense of hearing comfortableness of improving sound, for related personnel provides an acoustic environment that the sense of hearing is comfortable.For the high electronic product of commercialization degree, even can provide the product with corresponding auditory properties for the different consumer groups' sense of hearing preference, improve market attractiveness and the competitiveness of this product.In the world since last century concept that the nineties is introduced sound quality study noise control problem.Sound quality can be compared with sound pressure level from subjectivity and objectivity two impressions of statements people in aspects to sound, to the description of sound more comprehensively, truer, can embody the multidimensional property of sound.

Along with the proposition of sound Evolvement of quality concept and the development of Active Noise Control technology, existing research and inquirement is used for Active Noise Control to improve the sense of hearing comfortableness of noise.

Scheuren etc. are installed on an automobile by Active Noise Control, carry out the objective evaluation of the subjective assessment of sound quality and psychoacoustic parameter, the conclusion obtaining is that sound pressure level and loudness have obviously and reduces, sharpness slightly rises, but obtaining the former car of proportion by subtraction, overall subjective assessment increases (Scheuren J, et al.Active noise control and sound quality design in motor vehicles.SAE 1999-01-1846.).

Gonzaleza etc. have carried out double-channel active noise elimination experiment in the cuboid room of 7.35m * 4.16m * 2.59m, compared the loudness before and after controlling, sharpness, the psychoacoustic parameter such as roughness and tone color, also carried out sense of hearing comfortableness subjective assessment, conclusion be in most cases sense of hearing comfortableness with loudness reduce directly related, the improvement of auditory effect is except having much relations with noise abatement, also there is relation (Gonzaleza A with the spectral characteristics of noise before and after controlling, et al.Sound quality of low-frequency and car engine noises after active noise control.Journal of sound and vibration, 2003, 265:663 – 679.).

But, Scheuren etc. are the Active Noise Control systems that traditional acoustic pressure is minimised as target with the employings such as Gonzaleza, what study is that Active Noise Control is to the comfortable sex problem of the sense of hearing, sound quality model and psychoacoustic parameter are only as an evaluation foundation controlling effect, and itself does not consider auditory properties Active Noise Control system.

The psychoacoustic parameter that objective quantification is described sound quality mainly contains loudness, sharpness, roughness, fluctuation degree and tone, and wherein loudness is basic parameter, its having the greatest impact to sound quality under most situations.

Summary of the invention

For problems of the prior art, the object of the present invention is to provide a kind of noise elimination effect that can strengthen the responsive frequency range of people's ear, control Active Noise Control with traditional acoustic pressure and compare, the loudness with better subjective auditory effect is controlled Active Noise Control system and method thereof.

In order to achieve the above object, the present invention is by the following technical solutions: a kind of loudness is controlled Active Noise Control system, comprise I secondary source, J microphone, reference signal transducer, least mean square algorithm module, transaudient channel pattern and loudness filter, it is characterized in that described reference signal transducer is connected on loudness filter and I secondary source, described loudness filter is connected on transaudient channel pattern, described transaudient channel pattern is connected in least mean square algorithm module, a described J microphone is connected in least mean square algorithm module by loudness filter.

The frequency characteristic of described loudness filter has equal loudness contour and is inverted shape, and it is benchmark that described loudness filter be take the amplitude of 1000Hz sound, the formula of the relative amplitude β at frequency f place wherein, L _nfor the loudness level of selected equal loudness contour, L _fbe respectively the sound pressure level at this curve upper frequency f and 1000Hz place, p with L1000 _ffor L _fcorresponding acoustic pressure, p ₁₀₀₀for L ₁₀₀₀corresponding acoustic pressure, p ₀for reference sound pressure, p ₀=2 * 10-5Pa.

Loudness is controlled an Active Noise Control method, and step comprises:

1) by reference to signal transducer detection noise source, obtain reference signal x (n), wherein n is that time series performs step 2 afterwards);

2) described reference signal x (n) is through loudness filter filtering, and filtered signal is divided into again I * J road, and each signal is respectively through corresponding transaudient channel pattern module

wherein i is 1 to I integer, and j is 1 to J integer,

be i secondary source to the transaudient channel pattern of j microphone, x (n) through loudness filter and

after filtering, obtain raij (n), these signals are divided into again after I group, every group of J, input respectively I least mean square algorithm module; Simultaneously, described J microphone detects and obtains J acoustic pressure error signal, each acoustic pressure error signal obtains J loudness error signal through a loudness filter filtering, and described J loudness error signal inputted respectively I least mean square algorithm module, performs step afterwards 3);

3) described least mean square algorithm module is according to weight coefficient iterative algorithm

i controller weight coefficient of difference real-time update, wherein, μ is convergence coefficient, W _iand W (n+1) _i(n) be respectively n+1 and n i controller weight coefficient constantly, e _aj(n) be j loudness error signal, perform step afterwards 4);

4) the reference signal x described in step 1) (n) is divided into I road, and every road x (n) signal, through I controller weight coefficient resume module, obtains I secondary source control signal, performs step afterwards 5);

5) described I secondary source of I secondary source control signal input, sends respective secondary sound abating noises source by microphone.

Adopt technique scheme, the present invention has following beneficial effect: people's ear sensitive frequency composition in can appropriate fault in enlargement transducing signal, and the insensitive composition of decaying, make system strengthen the noise elimination effect to sensitive composition, can improve the subjective comfortableness of the sense of hearing of Active Noise Control.

Accompanying drawing explanation

Fig. 1 is that the loudness of an I of the present invention secondary source and J microphone is controlled Active Noise Control structure chart.

Fig. 2 is equal loudness contour (ISO226-2003).

Fig. 3 is the spectra of sound pressure before and after controlling.

Fig. 4 is the loudness spectrum before and after controlling.

Embodiment

Below in conjunction with Figure of description and specific embodiment, the present invention is further illustrated.

As shown in Figure 1, a kind of loudness is controlled Active Noise Control system, comprise I secondary source, a J microphone, reference signal transducer, least mean square algorithm module, transaudient channel pattern and loudness filter, it is characterized in that described reference signal transducer is connected on loudness filter and I secondary source, described loudness filter is connected on transaudient channel pattern, described transaudient channel pattern is connected in least mean square algorithm module, and a described J microphone is connected in least mean square algorithm module by loudness filter.

In figure:

C1, C2, CI is respectively the 1st, 2nd and I secondary source;

M1, M2, MJ is respectively the 1st, 2nd and J microphone;

R is reference signal transducer;

X (n) is n reference signal constantly;

W1, W2, WI is respectively the 1st, 2nd and I secondary source controller weight coefficient;

LMS is least mean square algorithm module;

Y1 (n), y2 (n), yI (n) is respectively the 1st, 2nd and I secondary source control signal;

A is loudness filter;

E1 (n), e2 (n), eJ (n) is respectively the 1st, 2nd and J the acoustic pressure error signal that microphone records;

Eaj (n), eaj (n), eaj (n) is respectively the 1st, 2nd and J loudness error signal;

(i=1 wherein, 2 ..., I; J=1,2 ..., be J) that i secondary source is to the transaudient channel pattern of j microphone;

Raij (n) (i=1 wherein, 2 ..., I; J=1,2 ..., signal J) obtaining through filtering for x (n).

The frequency characteristic of described loudness filter has equal loudness contour and is inverted shape, and it is benchmark that described loudness filter be take the amplitude of 1000Hz sound, the formula of the relative amplitude β at frequency f place

wherein, LN is the loudness level of selected equal loudness contour, and Lf and L1000 are respectively the sound pressure level at this curve upper frequency f and 1000Hz place, and pf is the acoustic pressure that Lf is corresponding, and p1000 is the acoustic pressure that L1000 is corresponding, and p0 is reference sound pressure, p0=2 * 10-5Pa.

As shown in Figure 2, equal loudness contour is obtained by a large amount of audiology experimental data statistical analyses, and the corresponding sound pressure level point of each frequency pure tone that loudness level is equated links up and forms equal loudness contour.

According to the approximate auditory system frequency characteristic that obtains of equal loudness contour, and according to its design loudness filter.Each sound that frequency is different for loudness level is identical, corresponding sound pressure level is also different, and the sound pressure level that certain frequency sound is corresponding is higher, and people's ear is more insensitive to this frequency.The characteristics design of loudness filter is become to have certain equal loudness contour and be inverted shape, pass through so this filter filtering, insensitive frequency content has decay to a certain degree, and sensitive composition is amplified, and sound pressure signal obtains loudness signal approx through this filter filtering effect.

A selected equal loudness contour that loudness level is LN, this curve upper frequency is that the sound pressure level at f and 1000Hz place is respectively

L _f＝20lg(p _f/p ₀) （1）

L ₁₀₀₀＝L _N＝20lg(p ₁₀₀₀/p ₀) （2）

In formula, pf is the acoustic pressure that Lf is corresponding, and p1000 is the acoustic pressure that L1000 is corresponding, and p0 is reference sound pressure, p0=2 * 10-5Pa.

By formula (1) and formula (2), obtained

$p_f/p_{1000}={10}^{\frac{L_f-L_s}{20}}---\left(3\right)$

For auditory system frequency characteristic being designed to have the inversion shape of equal loudness contour, the amplitude of 1000Hz sound of take is benchmark, the relative amplitude β at pull-in frequency f place

$\mathrm{\β}=p_{1000}/p_f={10}^{-\frac{L_f-L_N}{20}}---\left(4\right)$

There are many curves by class of 10phon in equal loudness contour, by equal loudness contour and formula (4), can obtain a plurality of loudness filter characteristics.Each equal loudness contour shape is different, filter characteristic shape is also different, show that the size of β peak value and the frequency range occurring thereof are variant, the β peak value of the filter characteristic that loudness level is low is less, and appear at higher frequency band, otherwise the higher filter of loudness level, β peak value is large and appear at compared with low-frequency range.Through the effect of loudness filter, the sound composition of the frequency range that β value is less is attenuated, and the composition of the larger frequency range of β value is amplified largely.While selecting loudness filter, the filter that different controlled noises are selected is also different, and principle is make β peak value frequency range and to total loudness, contribute maximum frequency range consistent in controlled noise.During practical operation, can attempt a plurality of filters, a selected best filter of effect.

Loudness filter in the present invention adopts FIR filter, and its coefficient obtains by the design of " fir2 " function in MATLAB according to loudness filter characteristic.

Setting controller weight coefficient length is L, defines i secondary source controller weight coefficient vector Wi (n) and reference signal vector X (n)

W _i(n)＝[w _i1(n),w _i2(n),...,w _iL(n)] ^T (5)

X(n)＝[x(n),x(n-1),...,x(n-L+1)] ^T (6)

I secondary sound source input signal is expressed as

$y_i\left(n\right)=W_i{\left(n\right)}^{T}X\left(n\right)=\underset{l=0}{\overset{L}{\mathrm{\Σ}}}{w}_{\mathrm{il}}\left(n\right)x(n-l+1)---\left(7\right)$

J microphone acoustic pressure error signal e j (n) is expressed as

$e_j\left(n\right)=d_j\left(n\right)+\underset{i=1}{\overset{I}{\mathrm{\Σ}}}{y}_i\left(n\right)*h_{S_{\mathrm{ij}}}\left(n\right)---\left(8\right)$

In formula, * represents convolution, and dj (n) arrives the signal of j microphone for noise source signal through primary channel,

be that i secondary sound source is to the unit impulse response of j microphone.

Formula (7) substitution (8) has

$e_j\left(n\right)=d_j\left(n\right)+\underset{i=1}{\overset{I}{\mathrm{\Σ}}}{R_{\mathrm{ij}}}^T\left(n\right)W_i\left(n\right)---\left(9\right)$

In formula,

$R_{\mathrm{ij}}\left(n\right)=X\left(n\right)*h_{S_{\mathrm{ij}}}\left(n\right)---\left(10\right)$

Loudness error signal e aj (n) is expressed as

$e_{\mathrm{aj}}\left(n\right)=e_j\left(n\right)*h_a\left(n\right)=d_{\mathrm{aj}}\left(n\right)+\underset{i=1}{\overset{I}{\mathrm{\Σ}}}{R_{\mathrm{aij}}}^T\left(n\right)W_i\left(n\right)---\left(11\right)$

Ha in formula (n) is loudness filter impulse response,

d _aj(n)＝d _j(n)*h _a(n) (12)

R _aij(n)＝R _ij(n)*h _a(n) (13)

The control target that loudness is controlled Active Noise Control system be J position loudness quadratic sum

Little, target function is

$I_{\mathrm{LD}}\left(n\right)=\underset{j=1}{\overset{J}{\mathrm{\Σ}}}E\left[e_{\mathrm{aj}}^2\left(n\right)\right]---\left(14\right)$

ILD is asked to minimum, by steepest descent method, obtained the iterative algorithm of i controller weight coefficient

$W_i(n+1)=W_i\left(n\right)-\mathrm{\μ}\underset{j=1}{\overset{J}{\mathrm{\Σ}}}{e}_{\mathrm{aj}}\left(n\right)R_{\mathrm{aij}}\left(n\right)---\left(15\right)$

In formula, μ is convergence coefficient.

Loudness is controlled an Active Noise Control method, it is characterized in that step comprises:

1) by reference to signal transducer detection noise source, obtain reference signal x (n), wherein n is that time series performs step 2 afterwards);

2) described reference signal x (n) is through loudness filter filtering, and filtered signal is divided into again I * J road, and each signal is respectively through corresponding transaudient channel pattern module

wherein i is 1 to I integer, and j is 1 to J integer, be i secondary source to the transaudient channel pattern of j microphone, x (n) through loudness filter and

after filtering, obtain raij (n), these signals are divided into again after I group, every group of J, input respectively I least mean square algorithm module; Simultaneously, described J microphone detects and obtains J acoustic pressure error signal, each acoustic pressure error signal obtains J loudness error signal through a loudness filter filtering, and described J loudness error signal inputted respectively I least mean square algorithm module, performs step afterwards 3);

3) described least mean square algorithm module is according to weight coefficient iterative algorithm

i controller weight coefficient of difference real-time update, wherein, μ is convergence coefficient, Wi (n+1) and Wi (n) are respectively n+1 and n i controller weight coefficient constantly, eaj (n) is j loudness error signal, performs step afterwards 4);

4) the reference signal x described in step 1) (n) is divided into I road, and every road x (n) signal, through I controller weight coefficient resume module, obtains I secondary source control signal, performs step afterwards 5);

5) described I secondary source of I secondary source control signal input, sends respective secondary sound abating noises source by microphone.

The Active Noise Control emulation of take in the cuboid enclosure space of 1.5m * 2.5m * 2m is example, introduces implementation procedure of the present invention.

Emulation adopts single channel Active Noise Control system, comprises 11 secondary source of Zao Sheng Yuan ﹑ and 1 control point, and position is (0.1,0.1,0.1) m, (0.2,2.3,1.8) m and (0.5,1.7,1.1) m in space respectively.

Reference signal x (n) is taken as 6 and take sine that 30Hz is fundamental frequency and the stack of random noise, the frequency of sinusoidal noise from 30Hz to 330Hz, frequency interval 60Hz, and amplitude reduces gradually from low frequency to high frequency.

Convergence coefficient μ gets 0.001, and controller weight coefficient gets 30, and the transfer function of transaudient passage is calculated according to cuboid Space Theory mode model, and with high order FIR filter simulation, loudness, according to Zwicker method, is calculated with Matlab language.

10～100phon equal loudness contour and formula (4) in 2 designed 10 loudness frequency characteristic of filter with reference to the accompanying drawings, according to these characteristics, with " fir2 " function design in MATLAB, obtain the coefficient of FIR loudness filter, these 10 loudness filters are respectively used to system.

Carry out the Active Noise Control emulation that loudness is controlled and traditional acoustic pressure is controlled.When loudness is controlled emulation, find the control better effects if of employing 70 and 80phon filter, wherein adopt the control result of 70phon loudness filter and adopt acoustic pressure to control to the results are shown in accompanying drawing 3 and accompanying drawing 4.

With do not control noise and compare, acoustic pressure after loudness is controlled has reduced 7.6dB, loudness has reduced 41.2%, and the acoustic pressure of acoustic pressure after controlling to reduce value be 11.2dB, than loudness, control large, but it is poor that the effect that loudness reduces is controlled than loudness, only reduces 19.8%, this explanation loudness is controlled and can be obtained better subjective auditory effect.

From accompanying drawing 3 (a) and 4 (a), the acoustic pressure of the following composition of 100Hz of not controlling noise is higher, the acoustic pressure of frequency content subsequently reduces gradually, consider people's ear to the sensitivity of following (0～1bark) sound of 100Hz not as 150～300Hz(1～3bark) sensitivity of sound is large, although the acoustic pressure of the following composition of 100Hz is higher, but the contribution that it is done total loudness is also little, and the contribution of the sound of 150～300Hz composition is much bigger.In comparison diagram 3 and 4, the acoustic pressure of each noise and loudness are known, acoustic pressure is controlled the following composition decay of 100Hz more, and loudness control is more to the decay of 150～300Hz composition, its reason is: the control target that acoustic pressure is controlled is acoustic pressure, the following composition of 100Hz is because acoustic pressure is compared with high and obtained more by system attenuation, and loudness is controlled owing to there being the effect of loudness filter, the following composition of 100Hz of controlled noise is reduced, 150～300Hz composition is exaggerated, correspondingly system has strengthened 150～300Hz composition attenuation degree, because the following composition of 100Hz is larger to total pressure contribution, thereby total acoustic pressure decrease that acoustic pressure is controlled is greater than loudness control, and 150～300Hz composition is larger to total loudness contribution, the loudness decrease that loudness is controlled is controlled more than acoustic pressure, be that loudness control can obtain better subjective auditory effect.

Claims (3)

1. a loudness is controlled Active Noise Control system, comprise I secondary source, a J microphone, reference signal transducer, least mean square algorithm module, transaudient channel pattern and loudness filter, it is characterized in that described reference signal transducer is connected on loudness filter and I secondary source, described loudness filter is connected on transaudient channel pattern, described transaudient channel pattern is connected in least mean square algorithm module, and a described J microphone is connected in least mean square algorithm module by loudness filter.

2. a kind of loudness according to claim 1 is controlled Active Noise Control system, the frequency characteristic that it is characterized in that described loudness filter has equal loudness contour inversion shape, it is benchmark that described loudness filter be take the amplitude of 1000Hz sound, the formula of the relative amplitude β at frequency f place

wherein, L _nfor the loudness level of selected equal loudness contour, L _fbe respectively the sound pressure level at this curve upper frequency f and 1000Hz place, p with L1000 _ffor L _fcorresponding acoustic pressure, p ₁₀₀₀for L ₁₀₀₀corresponding acoustic pressure, p ₀for reference sound pressure, p ₀=2 * 10-5Pa.

3. loudness is controlled an Active Noise Control method, it is characterized in that step comprises:

1) by reference to signal transducer detection noise source, obtain reference signal x (n), wherein n is that time series performs step 2 afterwards);

2) described reference signal x (n) is through loudness filter filtering, and filtered signal is divided into again I * J road, and each signal is respectively through corresponding transaudient channel pattern module

wherein i is 1 to I integer, and j is 1 to J integer, be i secondary source to the transaudient channel pattern of j microphone, x (n) through loudness filter and

after filtering, obtain raij (n), these signals are divided into again after I group, every group of J, input respectively I least mean square algorithm module; Simultaneously, described J microphone detects and obtains J acoustic pressure error signal, each acoustic pressure error signal obtains J loudness error signal through a loudness filter filtering, and described J loudness error signal inputted respectively I least mean square algorithm module, performs step afterwards 3);

3) described least mean square algorithm module is according to weight coefficient iterative algorithm

i controller weight coefficient of difference real-time update, wherein, μ is convergence coefficient, W _iand W (n+1) _i(n) be respectively n+1 and n i controller weight coefficient constantly, e _aj(n) be j loudness error signal, perform step afterwards 4);

4) the reference signal x described in step 1) (n) is divided into I road, and every road x (n) signal, through I controller weight coefficient resume module, obtains I secondary source control signal, performs step afterwards 5);

5) described I secondary source of I secondary source control signal input, sends respective secondary sound abating noises source by microphone.

Images