EP0530523B1

EP0530523B1 - Active silencer with improved method of selecting coefficient sequence

Info

Publication number: EP0530523B1
Application number: EP92113366A
Authority: EP
Inventors: Masaki Eguchi; Hiroyuki Iida
Original assignee: Sharp Corp
Current assignee: Sharp Corp
Priority date: 1991-08-06
Filing date: 1992-08-05
Publication date: 1998-03-18
Anticipated expiration: 2012-08-05
Also published as: JP2886709B2; DE69224780D1; JPH0540486A; DE69224780T2; EP0530523A2; EP0530523A3

Description

BACKGROUND OF THE INVENTION Field of the Invention

The present invention relates to silencing apparatus, and particularly to an active silencer for eliminating noise of a rotating machine such as a blower, an engine and the like.

Description of the Background Art

An active silencer is one for suppressing noise of a rotating machine such as a blower, an engine and the like. The active silencer suppresses noise by emitting sound wave of the same amplitude with phase shift of 180° from the noise from silencing sounding means to cause sound wave interference. Fig. 4 shows its basic structure.

In Fig. 4, for example, a noise signal sequence u (n) detected by a noise detection microphone 3 inside a duct is subjected to convolution operation of the following expression (1) with a predetermined coefficient sequence h (i) in an adaptive digital filter 4.

The operation result y (n) is outputted to a speaker 6 as a silencing signal.

In the expression (1), N designates the number of taps of the digital filter. The coefficient sequence h (i) of the adaptive digital filter 4 is adjusted so that the silencing error becomes minimum by a coefficient control portion (not shown) according to a silencing error signal e (n) at a reference point detected by a silencing error detecting microphone 5.

In the system structure shown in Fig. 4, the sound emitted from the speaker 6 may be fed back to the noise detecting microphone 3 to cause howling. Also, with noise coming from outside into the silencing error detecting microphone 5, mis-adjustment of the coefficient sequence h (i) may take place also to cause howling. In the conventional art, a method is used in which acoustic feedback, effects of external noise and the like are reduced by enhancing the directivity with a microphone array.

Also, for the purpose of suppressing the acoustic feedback, as shown in Fig. 5, an example has been reported in which a noise detecting microphone 3 and a silencing error detecting microphone 5 are provided at equal distances with respect to a speaker, and a difference of respective detection signals is obtained by a subtracter 7 and used as an input signal u (n) of an adaptive digital filter to cancel acoustic feedback.

In an active silencer with such a structure as shown in Fig. 4, even if the directivity of microphone is somewhat enhanced, or if the structure of Fig. 5 is used, the possibility of occurrence of howling can not be removed. Furthermore, when reducing the distance between the noise detecting microphone 3 and the speaker 6 to reduce the size of system, the possibility of occurrence of howling further increases because a cause of acoustic feedback increases. In this way, a measure to be taken for howling is an important problem in configuring an active silencer. Furthermore, when performing coefficient updating with the LMS algorithm which is widely used as an adaption algorithm of coefficient of the adaptive digital filter 4, the coefficient sequence h (i) gradually diverge if an input signal corresponds to a narrow band, resulting in a possibility of occurrence of overflow. The overflow of coefficient sequence h (i) causes degradation of silencing accuracy and occurrence of howling.

Prior art document GB-A-2 154 830 discloses an attenuation system of sound waves. In this system, when sound waves are intentionally attenuated by using destructive interference, temporal changes sometimes cause less than optimum performance, and the transfer function of a signal processing device driving a sound generator from a sound detector is modified at intervals as a result of sequential measurements of the transfer function between the sound detector and a further sound detector downstream from the generator. For this purpose a data processor calculates the required transfer function and causes a data processor to vary the coefficients of a digital filter comprising the processing device.

Further, prior art document EP-A-0 071 947 discloses an apparatus for reducing vibrations of a stationary induction apparatus. In this apparatus, a vibration is detected and a vibration applying force capable of suppressing the detected vibrations is applied to the stationary induction apparatus by at least one vibration applying device, in which the phase and amplitude of the vibration applying force of the vibration applying device are successively and repeatedly adjusted so as to decrease the sum of squares of the respective amplitudes of vibration detected by vibration sensors. A calculation for obtaining the sum of squares of the detected amplitudes of vibration and the control of the phase and amplitude of the vibration applying force based upon the calculated sum of squares may be carried out in accordance with a programm stored in a microcomputer.

Finally, prior art document US-A-5 022 082 discloses a method for reducing a convergence time of an active acoustic attenuation system upon start-up or upon a given sensed parameter change. The weights of an adaptive filter model coefficient weight vector are started at or changed to values which are closer to the converged value than an initial or present nonconverged value is to the converged value. The filter model converges in a shorter time as the weights change and are updated from their starting or changed value to the converged value.

SUMMARY OF THE INVENTION

It is an object of the present invention to enhance reliability of silencing in an active silencer; it is another object of the present invention to prevent howling in an active silencer.

To solve this object the present invention provides an active silencer as specified in claim 1.

Preferred embodiments of the invention are especially described in the subclaims.

The active silencer includes especially first noise detecting means for detecting a noise signal of an object of noise silencing, operating means for performing operation on the basis of the detected noise signal and a predetermined coefficient sequence to produce a silencing signal, silencing means for generating sound wave on the basis of the produced silencing signal to reduce noise at a predetermined position with respect to the object of noise silencing, second noise detecting means for detecting a noise signal at the predetermined position which is changed because of activation of the silencing means, and control means for controlling the operating means so that a predetermined coefficient sequence takes a most appropriate value on the basis of average power of a detection output of the second noise detecting means or another value representing a noise level.

In an active silencer made as described above, the control means controls the operating means so that a predetermined coefficient sequence takes the most appropriate value on the basis of average power of a detection output of the second noise detecting means or another value representing a noise level, so that the reliability is improved and howling can be prevented.

The foregoing and other objects, features, aspects and advantages of the present invention will become more apparent from the following detailed description of the present invention when taken in conjunction with the accompanying drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

Fig. 1 is a block diagram showing a structure of an active silencer according to one embodiment of the present invention.

Fig. 2 is a diagram for illustrating control condition of a coefficient sequence in the active silencer of Fig. 1.

Fig. 3 is a flow chart illustrating one example of contents of control corresponding to Fig. 2.

Fig. 4 is a block diagram showing a basic structure of a conventional active silencer.

Fig. 5 is a system block diagram showing a structure of one example having an object of preventing howling based on the active silencer of Fig. 4.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

Fig. 1 is a block diagram of an active silencer according to one embodiment of the present invention.

In the figure, there is a noise source 1 inside a container 2 such a duct having an opening 20 on its one side, and a noise detection microphone 3 is provided opposing thereto. A silencing error detection microphone 5 is provided at a reference point, and a silencing speaker 6 is provided at an intermediate position thereof as shown in the figure.

A noise signal u (n) detected by the noise detection microphone 3 and converted into a digital signal in an A/D converter 9 through an amplifier 11 is transmitted to a digital signal processor (DSP) through a data bus 16. A convolution operation designated by the above-noted expression (1) is applied to the noise signal u (n) there with a predetermined coefficient sequence h (i), and the operation result y (n) is outputted to a D/A converter 8 as a silencing signal through the data bus 16. Connected to the DSP 14 are a memory 21 for convolution operation coefficient storing coefficient sequence h (i), a digital filter 22 of FIR (Finite Impulse Response) type, for example, and a memory 23 for storing a noise signal u (n).

The silence signal converted into an analog signal by the D/A converter 8 is outputted into the container 2 from the speaker 6 through the amplifier 12. The result of silencing is detected by the silencing error detecting microphone 5 and captured as a silencing error signal e (n) into the DSP 14 through an amplifier 13, an A/D converter 10 and the data bus 16.

A coefficient sequence h (i) stored in the operating coefficient memory 21 is updated so that the silencing error signal e (n) becomes the minimum.

As clearly seen from the expression (1), calculation of silencing signal y (n) requires noise signals from the present time to N units time before, and the past data are stored in the memory 23 for storing noise signal u (n) through the data bus 16 together with a coefficient sequence h (i) which is the most appropriate for silencing.

In order to keep the coefficient sequence h (i) the most appropriate all the time, it is necessary to update the same. Fig. 2 is a diagram showing outline of a control system of coefficient sequence h (i) in the present invention along the time axis. As to the coefficient sequences h (i), in the initial state, a coefficient sequence in a stable operation or the most appropriate coefficient sequence which is measured in advance can be set. In the figure, T represents a period in which coefficient sequence h (i) is updated, and t1-t10 indicate times of respective operations. The coefficient sequence h (i) is updated during the period T from t1 to t2 and the average power w is calculated on the basis of the silencing error signal during the period of t2 to t3, and further the coefficient sequence h (i) at that time is stored in the memory 15. Subsequently, the coefficient sequence h (i) is further updated during the period T from t3 to t4. Next, the average power w is calculated again on the basis of the silencing error signal from t4 to t5 period. If the calculated power is smaller than the previous average power, the coefficient sequence h (i) at that time is stored in the memory 15. This is for retaining the coefficient sequence most appropriate in this period. If the calculated power is larger than the previous average power, the coefficient sequence h (i) stored in the memory 15 at the previous time is copied into the convolution operation coefficient memory 21 connected to the DSP 14. Also, the memory for storing noise signal u (n) is cleared as needed.

Subsequently, until the time T passes again, the coefficient sequence h (i) is updated and similar processings are continuously performed thereafter. Furthermore, check is always made for an abnormal value which exceeds a prescribed value of an input/output signal and overflow in operation processing during the coefficient updating period. At t8 that such an abnormality has been detected, whether the time T has passed or not, the coefficient sequence h (i) saved in the memory 15 at the previous time is called during the time of t8 to t9 and copied into the convolution operation coefficient memory 21, and the memory 23 for storing noise signal u (n) is cleared to zero.

Fig. 3 is one example representing the above-described silencing process flow in a flow chart.

When the operation starts, a noise signal u (n) and an error signal e (n) are inputted into the DSP 14 (step S101).

Next, a coefficient sequence h (i) is updated (step S102).

Next, presence or absence of overflow is examined (step S103).

If there is no overflow, a silencing signal y (n) is calculated by the convolution operation (step S104).

The calculated silencing signal y (n) is checked as to whether it is inside a range of prescribed values or not (step S105).

If the silencing signal y (n) is within the range of prescribed values, it is outputted to the speaker 6 (step S106).

A determination is made as to whether the time T of updating has passed or not (step S107). If it has passed, it is at t2 in Fig. 2 because it is immediately after operation start.

Next, the average power w in the T period immediately before is calculated on the basis of the error signal e (n) (step S108).

A check is made as to whether the average power w has increased as compared to the previous time or not (step S109).

If the average power w has not increased as compared to the previous time, or if it is immediately after the operation start, the coefficient sequence h (i) is retained in the memory 15 (step S110).

When there is an overflow in step S103, when the silencing output y (n) is out of the range of prescribed values in step S105, or when the average power w is increased than the previous time in step S109, the most appropriate coefficient sequence h (i) retained in the memory 15 is called and the memory for storing noise signal u (n) is cleared (step S111).

When the time counter which counts time T relating to the series of updating and calculation of the average power w is cleared (step S112), and when the time counter does not attain the time T in the process of step S107, the flow returns to step S101.

According to the present invention, in an active silencer in which, a silencing signal is produced by applying convolution operation to a signal sequence detected by sound wave detecting means and a predetermined coefficient sequence and outputted to silencing sounding means to eliminate noise at a reference point by sound wave interference, and the coefficient sequence is properly changed on the basis of a silencing error signal detected at the reference point, the most appropriate coefficient sequence h (i) is retained such as a coefficient sequence in a stable silencing operation or an obtained coefficient sequence which is measured in advance. When an abnormality of system such as occurrence of howling is detected by detection of an abnormal value which exceeds prescription of input/output, overflow in operation processing, and the like, or when the noise level at the reference point increases, by calling a previously retained most appropriate coefficient sequence, replacing the same by the present coefficient sequence, and further clearing to 0, a buffer storing a noise signal sequence u (n) to be subjected to convolution operation with a coefficient sequence h (i), the feedback loop of system is cut in an instant, so that howling can be stopped in an instant. Also, since a coefficient sequence h (i) is exchanged for that in a stable state as described above, shift to silencing condition from abnormal condition can be implemented in an instant. Furthermore, in the present invention, modification of hardware structure of a conventional active silencer is not required, resulting in a costal advantage. Furthermore, howling can be effectively suppressed only with an improvement of controlling software and stable control is always possible, so that it has excellent and practical applicability to a conventional active silencing system.

Although the present invention has been described and illustrated in detail, it is clearly understood that the same is by way of illustration and example only and is not to be taken by way of limitation, the scope of the present invention being limited only by the terms of the appended claims.

Claims

An active silencer for reducing noise at a predetermined position by emitting a sound wave with an opposite phase to the noise for noise reduction of an object to be silenced, comprising:

first detecting means (3) for detecting a noise signal of the object to be silenced;

second detecting means (5) for detecting a noise signal at said predetermined position for every predetermined period;

sound wave generating means (6, 14) for producing said sound wave by multiplying the noise signal detected by said first detecting means (3) by a coefficient updated for every predetermined period;

comparing means (14) for comparing a value representing a noise level calculated on the basis of said detected noise signal detected by said first detecting means and a value representing a noise level of a previous time; and

coefficient storing means (15) responsive to an output by said comparing means (14) indicating that said calculated representing value is smaller than the previous representing value for storing the coefficient used at that point,

characterized in that
said coefficient storing means (15) copies the previous representing value into a convolution operation coefficient memory (21) connected to said comparing means (14), if said calculated representing value is larger than the previous representing value for retaining a coefficient sequence most appropriate.
The active silencer according to claim 1, characterized in that said sound wave generating means comprises updating means (14) for changing a coefficient sequence used as a predetermined coefficient sequence for every predetermined period,

said convolution operation coefficient memory (21) stores said predetermined coefficient sequence and, when the coefficient sequence is changed by said updating means (14), storing said changed coefficient sequence in place of the predetermined coefficient sequence already stored, and

noise signal storing means (23) store a noise signal detected by said first detecting means (3).
The active silencer according to claim 2, characterized in that said updating means replaces the predetermined coefficient sequence stored in said convolution operation coefficient memory (21) with a most appropriate coefficient sequence stored in said convolution operation coefficient memory (21) under a predetermined condition.
The active silencer according to claim 3, characterized in that said predetermined condition is met when at least one of conditions is met, said conditions including, when the value representing the noise level calculated on the basis of the noise signal detected by said second detecting means (5) is equal to or larger than the value representing the noise level at the previous time, when said predetermined coefficient sequence overflows, when said produced silencing signal is out of a predetermined range, when a noise signal detected by said first detecting means (3) exceeds a prescribed value and when a noise signal detected by said second detecting means (5) exceeds a prescribed value.
The active silencer according to claim 3, characterized in that said updating means (14) cancels a content stored by said noise signal storing means (23) under said predetermined condition.