DE19832517C2 - Active silencing methods and silencers therefor - Google Patents

Description

The invention relates to a method for active sound attenuation according to the The preamble of claim 1 features specified and a silencer for this.

Such a method is known from US Pat. No. 5,680,337 for using a compensa tion speaker depending on signals from a reference microphone and one Error microphones to generate counter-sound to compensate for primary sound. in this connection the coherence is determined and a coherence filter is optimized filtering non-coherent signal components the compensation capacity of the active attenuator system should be focused on the coherent signal components. such Systems that are also referred to as ANC systems tend to tend to mismatch in speaker position at certain frequencies Instability, especially in this case no compensation for the corresponding frequencies is possible. In the case of pseudo sound, instabilities are also to be expected if especially turbulent pressure fluctuations on the reference microphone and / or errors microphone can be interpreted as sound. In addition, unfavorable microphone positions, at which the level for selected frequencies are too low (pressure node) Lead to instabilities.

Furthermore, from DE 39 08 881 A1 is an electronic noise attenuation system known in which a feedback compensation using the transmission Function provided between a compensation speaker and two microphones is. According to the output signal in the direction of sound propagation the compensation speaker arranged second microphone is an over Support function set up with a time delay, one being a generation Direction of a driver signal to be supplied transfer function according to a given adaptive algorithm taking into account the specific over Carrying function is determined with a time delay. The expert is thus the teaching conveys the time delay, thus the phase information of the transfer to use the compensation function.  

Furthermore, an active silencer is known from DE 196 44 085 A1 by means of a first microphone, also referred to below as a reference microphone To detect sound of a flowing gas and a control unit Activate compensation speakers. As the primary sound source, its signal should be extinguished, for example on an internal combustion engine, blower, Fan, compressor or the like referenced. The compensation Loudspeaker-generated sound waves overlap with the incoming ones according to sound waves detected by the reference microphone such that the amplitudes the sound coming from the primary sound source and the counter-sound of the sound at least approximately cancel each other out. Furthermore, in the channel behind the Compensation speaker another microphone, which is referred to as an error microphone is provided to regulate the compensation loudspeaker. Such active sound attenuation occurs, especially in acoustically narrow channels, under the Requirement for flat waves to be used.

Proceeding from this, the invention is based on the object, the Schwie shown to avoid and a procedure that can be implemented with little effort and propose a silencer for improved effective sound absorption. The Integration in a silencer and / or in an existing system can be easily implemented.

This object is achieved in accordance with that specified in claim 1 Features with regard to the method and according to that specified in claim 6 Characteristics of the sound intended to carry out the method damper.

The method proposed according to the invention and also that for implementation proposed silencers enable with little effort, especially failure adjustments, pseudo sound and unfavorable microphone positions through coherence measurements identify and compensate solutions. For this purpose, the primary Sound source the coherence of the signals between intended microphone positions, the can also be varied, measured. By measuring coherence, those Frequencies or frequency ranges in which the using the reference micro signals and the error microphone measured a predeterminable degree of coherence γ * exceed. With a degree of coherence γ = 0 there is no connection, whereas for a degree of coherence γ = 1 the signals are completely dependent on one another. The signals  which lie above this degree of coherence γ * become further signal treatment Control of the compensation speaker supplied.

In contrast to previous practice, in which the transfer function between speaker 6 and error microphone 12 of the transmission path 25 is formed in order to recognize the frequencies with a low transfer function and to amplify them accordingly, according to the invention, depending on the peak value (SW) of the measured transfer function Frequency ranges which are up to a factor of 30 d / B, in particular 25, preferably 20 d / B below this limit value are filtered out. This prevents inappropriate reinforcement and oversteering of the system. The signals which correspond to the mentioned range of the transfer function are passed on to the further signal treatment for controlling the compensation loudspeaker.

The reference microphone and the error microphone are expediently filters downstream, which are referred to below as a suitability filter (TF) and through the coherence measurement and / or a transmission filter, whose coefficients from the transfer function between the compensation loudspeaker and the error micro fon is determined, defined, only those frequencies being passed which are suitable for active noise reduction in ANC systems. In only frequencies or frequency ranges with good coherence are expedient or treated from the specified range of the transfer function. The Kohä similar to the measurement of the transfer function, the limit measurement is preferred in the System initialization integrated. Corresponding filters are in the signal paths called microphones arranged. Furthermore, in order to make a hearing noise to get combat, in a particularly useful manner, the detected microphone signals A- rated or filtered according to the A curve, with an A filter, below is also referred to as AF. By means of the or the A-filter, which the microphones downstream, an audible lowering of the low frequencies is effected.

Furthermore, it has proven to be useful to avoid a possible mismatch of the Speaker position by attaching at least two speakers to prevent different positions along the canal. So in a be Special embodiment of the invention two speakers in a row in the channel Improvement of the radiation is appropriate. It should be ensured that not for both positions of the loudspeakers the condition c / f = 4.l / n (n = 1, 3, 5, 7...) apply  may. If the channel is divided into at least two subchannels, the speakers become preferably also arranged offset along the channel, the Fehleri Krofone be arranged in particular at the end of the channel division. Can too the microphones in the case of a divided channel with a high-frequency absorbing lining be covered in the manner of a windbreak.

Further developments and special refinements of the invention are in the sub claims and the further description given. The invention will follow explained in more detail with reference to the drawing and it shows:

Fig. 1 is a block diagram of a single channel adaptive control with feedback compensation and secondary route adaptation for a "Filtered-X-algorith mus"

Fig. 2 shows a typical graph of coherence between the reference and error microphone,

Figure 3 shows a typical transfer function compensating Speaker / fon error mic.,

Fig. 4 schematically shows a channel having two offset in a channel mounted speakers,

Fig. 5 shows schematically the speaker position in a divided channel,

Fig. 6 shows schematically a divided channel with high-frequency absorbent clothing.

Fig. 1 shows the block diagram of a single-channel adaptive control with feedback compensation and secondary link adaptation for the "Filtered-X algorithm". The given sound field, which is indicated by the primary sound 2 , a primary source is canceled by a secondary field corresponding to the schematically indicated secondary sound 4 , which is generated by a compensation loudspeaker 6 , so that the resulting sound pressure 5 assumes a minimum. To control the speaker 6 a coherent with the field size to be deleted reference signal is required. In the channel 8 in the direction of propagation of the primary sound 2 in front of the speaker 6, a reference microphone 10 and behind the speaker 6 an error microphone 12 is provided. A loudspeaker 6 is controlled by a reference signal which is coherent with the field size to be erased and which is fed to a unit 14 connected downstream of the reference microphone 10 for signal conditioning. Unit 14 contains an amplifier, an A / D converter and anti-aliasing low-pass filters. Such a unit 16 for signal conditioning is also connected downstream of the error microphone 12 . A corresponding unit 18 for signal conditioning is connected to the loudspeaker 6 , containing a D / A converter and an amplifier.

The signal conditioning units 14 , 16 of the reference microphone 10 and of the error microphone 12 are each followed by a suitability filter 20 , 22 designated as TF. Using the filters 20 , 22 , those signal components or frequency ranges are filtered out according to the invention whose coherence falls below a predeterminable value or degree of coherence γ *, so that only frequency ranges with good coherence are taken into account in the further signal processing. In addition, those frequencies were filtered out at which the transfer function from the loudspeaker to the error microphone had strong drops. For this purpose, according to the invention, coherence measurements for specifying and setting the filters 20 , 22 are carried out. The coefficients of these filters 20 , 22 are specified as a function of the determined coherence and / or transfer function. This is particularly the case with system initialization.

The amplitudes of the microphone signals are preferably evaluated in accordance with the A-filter AF in order to reduce the low frequencies to suit the hearing. This A filter expediently forms a unit with the suitability filter. The suitability filters 20 , 22 are designed or set such that only those frequency ranges are considered in the further signal processing for which there is good coherence or for which the coherence exceeds a predeterminable degree of coherence. It has proven to be expedient to specify a degree of coherence of at least 0.5, preferably at least 0.7.

A control system, indicated by dash-dotted line 23 , is briefly explained below, which, like the SK units 14 , 16, is previously known [Guicking, D .: Active noise and vibration reduction, HDT specialist event, Essen, June 28-29, 1995]. In order to free the microphone signal of the reference microphone 10 from those parts which result from acoustic feedback 24 of the loudspeaker 6 , the transfer function of the acoustic feedback path is simulated by a further filter 26 , which is referred to as a feedback compensation filter RKF, the output signal of which is at the addition point 28 is subtracted from the microphone signal. The input signal x (t) of the downstream main filter 30 thus contains no components originating from the loudspeaker 6 . In the event of incomplete compensation, the error microphone 12 picks up an error signal e (t) which, in the manner shown, serves to track the adaptive main filter 30 . The adaptation of the main filter 30 is carried out in such a way that it simulates the acoustic transmission function including the complex frequency responses of the reference microphone 10 and the compensation loudspeaker 6 . The main filter 30 is preferably designed as a transversal filter, FIR filter for short, in particular with the Filtered-X-LMS algorithm according to Widrow and Hoff as an adaptation algorithm. The control takes place accordingly to the product e (t) .x (t), the change in the filter coefficients of the main filter being carried out according to a stochastic gradient method in such a way that x (t) and e (t) are decorrelated as best as possible, that is to say that e (t ) if possible no longer contains any shares of x (t). In addition, the decorrelation, which is particularly detrimental to a stochastic primary noise, is reduced by the secondary path 32 in accordance with HLE (ω) in that x (t) is pre-filtered via the simulation 32 before multiplication by e (t). In this "secondary path detection", a noise generator 34, designated RG, serves as an auxiliary signal source and the additional adaptation circuit. The coefficients for the RKF filter 26 and the prefilter 36 are expediently determined once in an initialization phase, in which in particular the microphone amplifiers are also controlled, copied into the corresponding preferably digital filters and subsequently kept constant.

A plurality of coherence measurements are expediently carried out with different positions of the error microphone 12 in the channel 8 , in particular in different, axially offset positions. The optimal coherence determined here for a specific microphone position is used while maintaining this microphone position for setting the suitability filter. For the coherence measurement, the primary sound originating from the primary sound source is recorded with both the reference microphone 10 and the error microphone 12 and corresponding signals are fed to the KH unit 38 for determining the coherence, in particular via the units 14 , 16 for signal conditioning. In a transmission filter, which is part of the suitability filter 22 , the frequencies or frequency ranges whose values are up to a certain factor (20-30) below the peak value SW were also filtered. The control and regulation of the compensation loudspeaker is carried out depending on the coherence determined in this way, preferably taking into account the predeterminable degree of coherence or depending on the transfer function. In the embodiment shown here, the two suitability filters 20 , 22 are defined in such a way that only predetermined frequencies or frequency ranges are passed which are suitable for active sound attenuation. It should be noted, however, that regardless of the specific design of the control and regulation of the compensation speaker according to the invention not the entire frequency spectrum of the primary sound is taken into account, but only those frequencies or frequency ranges with coherence greater than the degree of coherence γ *.

Fig. 2 shows a typical diagram of the coherence between reference and error microphone. The degree of coherence γ, which by definition assumes values between 0 and 1, is plotted against the frequency in a frequency range from 0 to 400 Hz. It should be noted that usually the active sound attenuation for frequencies in the order of magnitude up to 500 Hz is realized. In higher frequency ranges, active sound attenuation or experience has shown that ANC systems are practically not useful. In a preferred manner, the channel for damping higher frequencies is therefore provided with an absorbent lining. As explained above, only those frequency ranges are taken into account in the control system for controlling the compensation loudspeaker, for which the degree of coherence γ * is greater than 0.5, in particular greater than 0.7.

Fig. 3 shows a typical diagram of a transfer function from the compensation speaker to the error microphone. The transfer function is again plotted on a logarithmic scale for frequencies from 0 to 400 Hz. As can be seen, the transfer function has a peak value SW. Analogous to the coherence measurement, only those values are filtered out on the basis of the transfer function for which the transfer function has strong drops. For this purpose, a limit value GW is specified which is a predetermined factor below the peak value SW. A factor of up to 30, in particular up to 25 and preferably up to 20, has proven to be particularly expedient. Analogous to the above statements on coherence measurement, only the frequencies or frequency ranges are considered based on the transfer function that are below the peak value of the transfer function in the proposed range. The coefficients of the suitability filters 20 , 22 explained with reference to FIG. 1 are specified accordingly. Thus, the frequencies or frequency ranges are filtered out according to the invention at which there are sharp drops or decreases in the transfer function from loudspeakers to the error microphone.

Fig. 4 shows a special embodiment of the invention with a further compensation speaker 7 , which is arranged in the channel 8 behind the compensation speaker 6 . The two loudspeakers 6 , 7 arranged one behind the other are driven together as explained above with reference to FIG. 1. This improves the radiation, if the speaker positions are selected so that the two clauses 6 , 7 do not simultaneously meet the conditions c / f = 4.l / n, where n = 1, 3, 5, 7. , , and l means the respective distance of the speakers from the end of the channel.

FIG. 5 shows an embodiment according to which the channel 8 is divided into two sub-channels 40 , 41 . The compensation speaker 6 is assigned to the sub-channel 40 and the second speaker 7 is assigned to the other sub-channel 41 in the direction of the longitudinal extent of the channel 8 . It can be seen that with such a divided channel with offset speakers 6 , 7 an improved radiation is guaranteed. The two error microphones 12 , 13 are arranged at the end of the channel division 42 .

Finally, an embodiment similar to FIG. 5 is shown in FIG. 6, but the channel 8 is provided on the inside with a high-frequency absorbing lining 44 . Both the reference microphones 10 , 11 and the error microphones 12 , 13 are at least partially covered by means of the lining 44 in the manner of a windbreak, so that interference signals due to turbulent pressure fluctuations or the like are largely avoided.

reference numeral

2

primary sound

4

secondary sound

5

resulting sound pressure

6

.

7

Canceling loudspeaker

8th

channel

10

.

11

reference microphone

12

.

13

error microphone

14

.

16

.

18

Signal conditioning unit

20

.

22

suitability filters

23

line

24

acoustic feedback

25

Transmission path between compensation speakers and error microphone

26

RKF filter

28

addition site

30

main filter

32

Simulation of the secondary line Hle (ω)

34

noise generator

36

prefilter

38

KH unit / unit for determining coherence

40

.

41

subchannel

42

Channel subdivision

44

lining

Claims (9)

1. A method for active sound attenuation in a channel through which a medium can flow, wherein by means of a compensation loudspeaker in response to signals from a reference microphone and an error microphone, counter-sound is generated to compensate for primary sound, and the frequencies or frequency ranges are determined by coherence measurement , in which the signals measured by means of the reference microphone and the error microphone and exceeding a predeterminable degree of coherence are taken into account in the further signal treatment,
characterized in that the frequencies or frequency ranges are determined from the transfer function between the compensation speaker and the error microphone, in which the amplitude of the transfer function lies in a range between its peak value and an underlying predetermined limit value, the frequencies or frequency ranges determined in this way at the further signal handling are taken into account,
and that the frequencies or frequency ranges in which the amplitude of the transfer function is below the predetermined limit are filtered out. 2. The method according to claim 1, characterized in that the factor by which the limit value is below the peak value at up to 30, in particular up to 25, is preferably up to 20. 3. The method according to claim 1 or 2, characterized in that the coherence measurement or the measurement of the transfer function for different positions of the reference microphone ( 10 ) and or the error microphone ( 12 ) are carried out. 4. The method according to claim 1 or 2, characterized in that the coherence measurement or the measurement of the transfer function in the system initialization is integrated. 5. The method according to any one of claims 1 to 3, characterized in that depending on the coherence measurement or the transfer function, coefficients for at least one of the reference microphone ( 10 ) and / or the error microphone ( 12 ) after switched filter ( 20 , 22 ) are determined and entered become. 6. Muffler for performing the method according to one of claims 1 to 5, characterized in that the reference microphone ( 10 ) and / or the error microphone ( 12 ) is followed by a filter ( 20 , 22 ) whose coefficient corresponds to that by the coherence measurement and values resulting from the transfer function are set. 7. Silencer according to claim 6, characterized in that in the channel ( 8 ) in addition to the compensation speaker ( 6 ) a further compensation speaker ( 7 ) is arranged, the speakers ( 6 , 7 ) in the longitudinal direction of the channel ( 8 ) are arranged one behind the other. 8. Silencer according to claim 6 or 7, characterized in that in the channel ( 8 ), which is divided into at least two sub-channels ( 40 , 41 ), the respective sub-channel ( 40 , 41 ) assigned compensation speakers ( 6 , 7 ) are arranged offset to one another in the longitudinal direction of the channel ( 8 ) and / or that the error microphones ( 12 , 13 ) assigned to the respective sub-channel ( 40 , 41 ) are arranged at the end of the channel subdivision ( 42 ), which, depending on the coherence measurements, also in the longitudinal direction can be arranged offset. 9. Silencer according to one of claims 6 to 8, characterized in that the in particular divided channel ( 8 ) is provided with a high-frequency absorbing clothing and / or that the lining ( 42 ) the reference microphones ( 10 , 11 ) and / or Error microphones ( 12 , 13 ) at least partially covers.