EP1800400A1

EP1800400A1 - Method for stabilizing an adaptive algorithm and device for carrying out said method

Info

Publication number: EP1800400A1
Application number: EP05788936A
Authority: EP
Inventors: Harry Bachmann
Original assignee: Anocsys AG
Current assignee: Anocsys AG
Priority date: 2004-10-12
Filing date: 2005-10-07
Publication date: 2007-06-27
Also published as: US20090052690A1; JP2008516540A; WO2006039826A1

Abstract

The invention relates to a method for estimating an unknown transfer function (H) which comprises an input signal (x) and an actual output signal (y). According to the invention, an estimated output signal (y) is generated in an adaptive process (3) using the input signal (x). An error signal (e) is generated from the actual output signal (y) and the estimated output signal (y) and the error signal (e) is used to improve the adaptive process (3), whereby at least one of the following signal paths is interrupted or opened depending on at least one condition: a signal path carrying the error signal (e), a signal path carrying the input signal (x), or a signal path carrying the estimated output signal (y). The invention allows to substantially stabilize adaptive processes or algorithms. The invention also relates to the use of the method, to a device and to the use of said device.

Description

Method for stabilizing an adaptive algorithm and a device for carrying out the method

The present invention relates to a method for stabilizing an adaptive algorithm according to the preamble of claim 1, an application of the method, a device for carrying out the method and a use of the device.

Noise sources are increasingly perceived as an environmental impact and are considered to reduce the quality of life. However, as noise sources are often unavoidable, noise reduction methods based on the principle of wave cancellation have already been proposed.

The principle of Active Noise Canceling (ANC) is based on the cancellation of sound waves due to interference. These interferences are generated by one or more electro-acoustic transducers, such as loudspeakers. The signal radiated by the electro-acoustic transducers is calculated by means of a suitable algorithm and continuously corrected. The basis for the calculation of the signal to be radiated by the electro-acoustic transducers is the information supplied by one or more sensors. These are on the one hand information about the nature of the signal to be minimized. For example, a microphone can be used for this purpose which detects the noise to be minimized. On the other hand, information about the remaining residual signal is needed. Again, microphones can be used.

The basic principle applied with active noise reduction was Paul Lueg in a patent document from the year 1935 and the disclosure number AT-141 998 B described. This document discloses how noise in a tube can be extinguished by means of

Generation of a signal with opposite phase position.

Further developments led to a number of specific algorithms, such as the LMS (Least Mean Square) and related algorithms such as the FxLMS or the NLMS.

An algorithm for active noise reduction requires information from at least one sensor (for example a microphone), which determines the residual error - also referred to below as an error signal. Depending on the application and algorithm used, another sensor is provided that provides information about the nature of the signal to be minimized. Further, an adaptive noise reduction system requires one or more actuators (for example in the form of loudspeakers) to output the correction signal. The information from the sensors must be converted by an analog / digital converter into a suitable format. After being processed by the algorithm, the signal is reconverted from a digital to analogue converter and sent to the Actuators transmitted. These converters are subject to limitations in both resolution and dynamics.

Many algorithms, in particular the known gradient methods, have some instabilities with uncorrelated input signals. Together with the limitations of the converters, this can lead to uncontrolled behavior of the algorithm for small input signals. This can be expressed in the form of low-frequency noises or as generally unstable behavior of the overall system.

The present invention is therefore based on the object of specifying a method for stabilizing an adaptive algorithm which does not have the above disadvantages.

This object is achieved by the features stated in the characterizing part of claim 1. Advantageous embodiments, an application of the method, a device for carrying out the method and a use of the device are specified in further claims.

A method of stabilizing an adaptive algorithm by which an unknown transfer function having an input signal and an actual output signal is estimated, the method being that a is generated using the input signal by means of an adaptive process such that an error signal is generated from the actual output signal and the estimated output signal and that the adaptive process due to the

Error signal is improved. According to the invention, at least one of the following signal paths is interrupted or opened as a function of at least one condition:

a signal path carrying the error signal; a signal path leading the input signal;

a signal path carrying the estimated output signal.

Thus, for the first time, a method is provided which is particularly suitable for stabilizing adaptive algorithms, since signals which can not be processed by the method according to the invention can be kept away from the adaptive process from the outset, making the system more stable and robust overall.

Another embodiment of the present invention is that a predetermined signal level or a predetermined average signal power of one of the following signals is exceeded or exceeded: - input signal;

estimated output signal;

- error signal.

This results in an advantageous manner that the inventive method controlled itself and in - O ^{• *} -

its structure to the current waveforms and signal strengths automatically adapts.

In yet another embodiment of the present invention, it is contemplated that the condition for

Changing a signal path of the guided in this signal path signal is dependent.

Finally, in a further embodiment it is provided that several signal paths are interrupted or opened simultaneously.

Even if the method according to the invention is particularly suitable for active noise reduction, other applications are by no means excluded. On the contrary: that

Methods according to the invention are particularly suitable for improving the stability and robustness of all adaptive systems.

Furthermore, the subject of the present invention is a device which comprises the following features:

- An adaptive processor unit for determining an estimated output signal, wherein the processor unit an input signal is applied, - means for determining an error signal from an actual output signal and the estimated output signal, wherein the error signal is supplied to the adaptive processor unit, and

a switching unit in at least one of the following signal paths: a signal path carrying the error signal;

a signal path leading the input signal;

a leader in the estimated output

Signal path.

A further embodiment variant comprises means for determining a level or an average power of a signal, wherein these means are operatively connected to at least one switching unit.

In yet another embodiment of the present invention, the means for determining a level or an average power of a signal is operatively connected to the switching unit in the same signal path.

In a further embodiment of the present invention, a plurality of switching units can be activated simultaneously.

In a further embodiment of the present invention, the switching unit has an adjustable switching characteristic.

Finally, in a further embodiment of the present invention, the switching units are operatively connected to each other.

The present invention is based on

Embodiments with reference to drawings further explained below. Show it: 1 shows a variant of an inventive device in a schematic representation,

2 is a simplified block diagram of the embodiment variant shown in FIG. 1, also in a schematic representation,

Fig. 3 is a simplified block diagram of a switching unit used in Figs. 1 and 2 and

4 shows a signal curve for illustrating a possible mode of operation of a switching unit according to FIG. 3.

Fig. 1 shows an embodiment of an inventive device for the reduction of noise. It is a so-called Adaptive Noise Canceller (ANC) system that eliminates or at least reduces a noise in a room R using the principle of signal cancellation.

Central unit of such an adaptive noise reduction system is an adaptive

Processor unit 3, which is operatively connected to an external microphone unit 1, wherein the addition "externally" indicates a arranged outside the space R microphone unit. Thus, a noise source which is usually present outside the room R can be better detected. Of Furthermore, two internal microphone units 5 and two loudspeaker units 7 are provided in the space R, all of which are operatively connected to the adaptive processor unit 3. As can be seen from FIG. 1, a switching unit 2, 4, 6 is provided in each case between one of the microphone units 1, 5 and the adaptive processor unit 3 or respectively between one of the loudspeaker units 7 and the adaptive processor unit 3, which makes it possible to control the respective one Signal path to interrupt.

In the adaptive processor unit 3, due to the signal recorded with the microphone unit 1, a reduction signal is fed via the loudspeakers 7 into the room R, so that an interference signal passing through the walls or windows into the room R is deleted by signal cancellation or reduction in the room R. or reduced. So that this can be achieved successfully under changing conditions, an error signal is picked up with the aid of the microphone units 5 and fed back to the adaptive processor unit 3, so that in the adaptive processor unit 3 the calculations of the reduction signal can be improved and, as a consequence, optimum signal cancellation or signal reduction can be achieved.

It is expressly understood that any number of microphone units 1, 5 and speaker units 7 are conceivable without the principle according to the invention being abandoned. Also, instead of the microphone units 1, 5 and / or the speaker units 7, other converter units are conceivable. 2 shows a block diagram in a simplified equivalent circuit diagram of the embodiment variant of the invention according to FIG. 1. According to the invention, the signal paths can be interrupted, which is accomplished with the aid of switching units 2, 4, 6. A switching unit 2, 4, 6 in this case denotes a switch which is controlled by the size of a parameter. The size of the parameter corresponds, for example, to the average input power or the input level. The switching unit 2, which is connected between the microphone unit 1 and the adaptive processor unit 3, is used to interrupt the signal path to the adaptive processor unit 3 as soon as, for example, the noise recorded with the microphone unit 1 is below a predetermined one

Signal level is. The adaptive processor unit 3 no longer receives any noise as a consequence of the signal path interruption, as a result of which the adaptive processor unit 3 no longer outputs a reduction signal to the loudspeaker unit 7. This configuration makes sense, for example, when a noise below a certain volume need not be actively minimized because the walls or windows already sufficiently attenuate the noise. In addition - as a further embodiment of the invention - with the switching unit 6, between the adaptive

Processor unit 3 and the speaker unit 7 is connected, the signal path to be controlled by a signal generator for generating the reduction signal to the speaker units 7. The speaker units 7 so no signal is applied as soon as the switching unit 6 the Signal path to the speaker unit 7 interrupts. In this way, any background noises, which are generated in the speaker units 7 on their own, suppressed. This also contributes significantly to the stability of the overall system, since these background noise would otherwise be detected by the sensors - ie the microphone units 5 - for detecting the residual noise. The additional switching unit 4, which can also be activated in response to the noise, can interrupt the signal path leading the error signal. This causes the adaptive processor unit 3 to receive no error signal from the corresponding microphone unit 5, and thus the adaptive processor unit 3 assumes that the system is optimally adapted to the current situation and, consequently, that it is not necessary to further optimize.

The applications of each of the mentioned embodiments with a switching unit as well as a combination of two or more of the mentioned applications results in a substantial increase in the stability of the overall system, since quiet and thus difficult to process input signals are eliminated. Thus any shortcomings in the analog / digital and digital / analog converters or other components required in digital systems lose their influence.

However, it is also possible to define a limit value for the residual noise in room R; when falling below this limit, the error signal ε is turned off, ie the Signal path from the internal microphone units 5 to the adaptive processor unit 3 is interrupted. This is interpreted by the adaptive processor unit 3 such that no error signal ε is present. This also implies that silent and thus difficult to process signals are excluded from the calculations. Depending on the muting of the error signal ε, both the external microphone unit 1, which is used to record noises, and one or more loudspeaker units 7 can be controlled. such

Configurations make sense, for example, if the adaptive processor unit 3 should only work if a specific limit value in the space R is exceeded.

As has already been pointed out, a switching unit 2, 4, 6 functions like a switch whose state is determined by the input quantity. It is not important for the function of a switching unit 2, 4, 6 whether the input signal detects the level, the average power or another magnitude.

The basic mode of operation of a switching unit 2, 4, 6 will be explained below with reference to FIGS. 3 and 4 on.

Fig. 3 shows one of the switching units 2, 4, 6, the one

Input signal 9 is acted upon. In the switching unit 2, 4, 6, an output signal 10 is generated, resulting for example from the input / output waveform according to FIG. 4. In this case, an arbitrary characteristic curve can be provided, in particular based on the function of a so-called AGC (Automatic Gain Control) unit conceivable that the signal can also experience a gain or a weakening. In the graph shown in Fig. 4, the abscissa represents the level of the input signal 9 and the ordinate represents the level of the respective one

Output signal 10 applied. From the course shown, the following can be derived: As soon as the input signal 9 of the switching unit 2, 4, 6 falls below a predetermined threshold value, the output signal 10 of the switching unit 2, 4, 6 is switched to zero, i. the signal path through the switching unit 2, 4, 6 is interrupted. The output signal 10 thus has no value when the input signal 9 is below the predetermined threshold. On the other hand, the output signal 10 corresponds to the input signal 9 when the input signal 9 exceeds the predetermined threshold.

A further embodiment - as can also be seen from Fig. 4 - is that the

Switching unit 2, 4, 6 has a control input 12. Via this control input 12, the switching unit 2, 4, 6 can be controlled, wherein a control signal is generated for example in the adaptive processor unit 3 or in another switching unit 2, 4, 6.

Yet another embodiment of the present invention is that the switching unit 2, 4, 6 has a control output 13, via which also other switching units 2, 4, 6 can be controlled. It is also conceivable the state of the switching unit 2, 4, 6 is transmitted via the control output 13 to the adaptive processor unit 3 or another arithmetic unit for further processing.

The control output 13 and the control input 12 can also be used inverted. In this way, for example, a switching unit 2, 4, 6 is then opened (i.e., the signal path through the switching unit 2, 4, β is not interrupted) when the signal path is interrupted by another switching unit 2, 4, 6. It is provided, for example, that the time it takes for the switching unit 2, 4, β after exceeding the predetermined threshold value interrupts the signal path, can be set. Likewise, the time required by the switching unit 2, 4, 6 to reopen the signal path after falling below the predetermined threshold value can be set. This has the advantage that thus the signal path through the switching unit 2, 4, 6 is not interrupted or opened before and after each zero crossing, whereby further instabilities of the overall system are prevented by the invention.

Claims

claims

1. A method for stabilizing an adaptive algorithm by means of which an unknown

Transfer function (H) having an input signal (x) and an actual output signal (y), the method being

that an estimated output signal (y) is generated using the input signal (x) by means of an adaptive process (3),

- That an error signal (ε) from the actual output signal (y) and the estimated output signal

(y) is generated and - that the adaptive process (3) due to the

Error signal (ε) is improved, characterized in that in response to at least one condition of at least one of the following signal paths is interrupted: - the error signal (ε) leading signal path;

a signal path leading the input signal (x);

a signal path carrying the estimated output signal (y).

2. The method according to claim 1, characterized by the condition that a predetermined signal level or a predetermined average signal power of one of the following signals is exceeded:

- input signal (x); estimated output signal (y); - error signal (ε).

3. A method for stabilizing an adaptive algorithm, by means of which an unknown transfer function (H) is estimated, the one

Input signal (x) and an actual output signal (y), the method being

(y) is generated and

in that the adaptive process (3) is improved on the basis of the error signal (ε), characterized in that at least one of the following signal paths is opened as a function of at least one condition:

a signal path leading the error signal (ε); a signal path leading the input signal (x);

a signal path carrying the estimated output signal (y).

4. The method according to claim 3, characterized by the condition that a predetermined signal level or a predetermined average input power of one of the following signals is exceeded:

- input signal (x);

estimated output signal (y); - error signal (ε).

5. The method according to any one of claims 1 to 4, characterized in that the condition for changing a signal path of the guided in this signal path signal (x, ε, y) is dependent.

6. The method according to any one of claims 1 to 5, characterized in that a plurality of signal paths are interrupted or opened simultaneously.

7. Application of the method according to one of claims 1 to 6 for active noise reduction in a room (R).

8. Apparatus for carrying out the method according to one of claims 1 to 6, characterized by

- An adaptive processor unit (3) for determining an estimated output signal (y), wherein the processor unit (3) an input signal (x) is applied,

- means (19) for determining an error signal (ε) from an actual output signal (y) and the estimated output signal (y), the error signal (ε) being supplied to the adaptive processor unit (3), and

a switching unit (2, 4, 6) in at least one of the following signal paths:

a signal path leading the error signal (ε);

a signal path leading the input signal (x); a signal path carrying the estimated output signal (y).

9. The device according to claim 8, characterized in that means for determining a level or an average power of a signal (x, ε, y) are provided, said means being operatively connected to at least one switching unit (2, 4, 6).

10. The device according to claim 9, characterized in that the means for determining a level or an average power of a signal (x, ε, y) in a signal path with the switching unit (2, 4, 6) are operatively connected in the same signal path.

11. Device according to one of claims 8 to 10, characterized in that a plurality of switching units (2, 4, 6) are activated simultaneously.

12. Device according to one of claims 8 to 11, characterized in that the switching unit (2, 4, 6) has an adjustable switching characteristic.

13. Device according to one of claims 8 to 12, characterized in that the switching units (2, 4, 6) are operatively connected to each other.

14. Use of the device according to one of claims 8 to 13 for active noise reduction in a room (R).