JP2524046B2 - Electronic silencing method and device - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an electronic noise reduction method and apparatus, and more particularly to an additional sound having the same sound pressure and a phase opposite to that of periodic noise or pseudo periodic noise generated from a noise source. The present invention relates to an electronic silencing method and apparatus for silencing sound by the sound wave interference.

[0002]

2. Description of the Related Art Conventionally, in this type of electronic muffler, an additional sound having a phase opposite to that of noise and having the same sound pressure is generated from a speaker in a predetermined region to be muffled, but a drive signal for driving this speaker is An adaptive digital filter is created based on an input signal from a sensor microphone or the like for detecting noise and an error sensor for detecting an interference sound of noise and additional sound in a predetermined region to be muted.

FIG. 10 shows the basic configuration of this type of electronic silencer. The adaptive digital filter 1 uses the input signal x _(n) indicating noise to drive the speaker drive signal y _(n).
Is output. In the figure, d _(n) is the input signal x
_(n) is the desired response in the error sensor, e _(n)
Is an error signal detected by the error sensor.
C is a transfer function from the speaker to the error sensor.

By the way, the adaptive digital filter 1 can be realized by an FIR filter having a filter coefficient of a predetermined tap length and an adaptive algorithm for controlling the filter coefficient. The adaptive algorithm is the input signal x.
_{From the information of (n)} and the error signal e _(n) , the filter coefficient of the adaptive digital filter is adjusted so that the energy of the error signal e _(n) becomes the minimum under some evaluation criteria.

Next, a method for setting the above filter coefficient to an optimum value will be described. The output y _(n) of the adaptive digital filter 1 is given by the convolution of the input x _(n) and the filter coefficient w _(i) of the predetermined tap length I, so that

[0006]

[Equation 14]

And the error signal e
_(n) is the following equation,

[0008]

(Equation 15)

Can be expressed as In the equation (15), c _(j) is the filter coefficient (C ^) of the tap length J indicating the transfer function C from the speaker to the error sensor, and r _{(n )} is the input x _(n). Is a reference signal obtained by filtering with the above filter coefficient (C ^),

[0010]

[Equation 16]

[0011] The following vector representation for simplicity,

[0012]

[Equation 17]

When the above equation (15) is obtained,

[0014]

(Equation 18)

Can be expressed as Here, when the mean square error (MSE) E [e _(n) ² ] is calculated, the following formula 18 is obtained.

[0016]

[Formula 19]

And MSE becomes a quadratic function of the filter coefficient. The second derivative is first, and if the derivative is set to 0, the solution having the minimum value J _min is obtained. Now, the FX algorithm (Filtered, which is the steepest descent algorithm)
-x LSM (least-mean-square) algorithm)
Using the instantaneous squared error e _(n) ² itself as the estimator of E J, the estimator of the derivative (gradient ∇) of J (∇
^ _N ) is given by

[0018]

(Equation 20)

Then, using the above (∇ ^ _n ),
The filter coefficient of the adaptive digital filter is recursively updated by the following equation.

[0020]

[Equation 21]

Here, μ is a positive scalar and is a step size parameter for controlling the magnitude of the correction amount in each repetition. The equation [21] above is the gradient vector (∇
^ _N ), which means that the filter coefficient is sequentially updated in the opposite direction (in the direction of the steepest descent of the error surface), and if this is continued, the MSE finally reaches the minimum value J _min , and the filter coefficient becomes the optimum value. I will have it.

The above-mentioned FX algorithm is a case where there is one error signal e _(n), but when there are a plurality of error sensors etc., a MEFX algorithm (Multipule Error Filteredjk-x Algorithm) which is an extension of the above FX algorithm is used.
As an electronic muffler to which m) is applied, there is one disclosed in Japanese Patent Publication No. 1-501344. On the other hand, an electronic silencer for limiting the noise to be silenced to periodic noise and silencing the periodic noise has been proposed (US Patent No. 415).
3815).

As shown in FIGS. 11 and 12, this electronic muffler stores some discrete data corresponding to the muffling waveform in the memory 12 in the waveform generator 11. The adaptive unit 13 selects the optimum sound deadening data from these data, and uses the signal from the error sensor 14 that detects the error after the sound wave of the noise source 15 and the sound wave for sound deadening interfere with each other to reduce the sound deadening sound in the memory 12. Correct the data. Further, the selected silencing data is appropriately read by a synchronization signal from the sensor 16 that detects the cycle of the periodic noise, and is output to the speaker 17.

[0024]

By the way, the conventional FX
The algorithm requires calculations such as the convolution operation [Equation 14] expression in the adaptive digital filter and the convolution operation [Equation 16] expression for creating the reference signal,
Especially when the number of noise sources, error sensors, speakers, etc. increases, enormous calculations have to be executed within a minute sampling period. Therefore, there are restrictions such as cost and DSP processor capability. At present, it is practically used for periodic noise or pseudo periodic noise.

On the other hand, the electronic muffling apparatus described in US Pat. No. 4,153,815 does not require the above-mentioned convolution calculation, but does not perform adaptive control based on the LMS algorithm and cannot expect a sufficient muffling effect. . The present invention has been made in view of the above circumstances, and it is possible to significantly reduce the amount of calculation for calculating the filter output of an adaptive digital filter, and to improve periodic noise or pseudo periodic noise. An object of the present invention is to provide an electronic silencing method and device capable of silencing.

[0026]

SUMMARY OF THE INVENTION In order to achieve the above object, the present invention uses a speaker to generate an additional sound having an opposite phase and the same sound pressure with respect to periodic noise or pseudo periodic noise generated from a noise source. An electronic muffling method for generating noise to muffle sound waves in a predetermined region by the sound wave interference, wherein a drive signal for driving the speaker has an adaptive digital filter having a first filter coefficient that is sequentially updated every predetermined sampling period. And a transfer function from the speaker to the error sensor, wherein the first filter coefficient is an error signal from an error sensor arranged in the predetermined region for detecting an interference sound wave. A filter obtained by filtering the input signal with a digital filter having a second filter coefficient And Reference signal, an electronic silencing method is updated based on the current first filter coefficients,
The period of the periodic noise or the pseudo periodic noise is detected for each period, and the input signal of the adaptive digital filter and the digital filter is an impulse generated in synchronization with the period detection, and the period is detected. Each time, the tap length of the adaptive digital filter and the tap length of the reference signal are made to match based on the detected cycle.

[0027]

According to the present invention, impulses synchronized with the fundamental period of periodic noise or pseudo-periodic noise are generated, whereby an adaptive digital filter for producing a speaker driving signal and a reference signal are provided. It eliminates the need for convolutional operations in the digital filter used to create it. Further, every time the basic period is detected, the tap length of the adaptive digital filter and the tap length of the reference signal are matched on the basis of the detected period so that they can be theoretically handled as an FX algorithm. .

[0028]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A preferred embodiment of an electronic silencing method and apparatus according to the present invention will be described in detail below with reference to the accompanying drawings. FIG. 1 is a block diagram showing the basic configuration of an electronic silencer according to the present invention. In the figure, the noise source 20 generates periodic noise or pseudo periodic noise, and corresponds to, for example, an engine, a motor, a fan, a press machine or the like.

The synchronization detector 22 generates a synchronization pulse synchronized with the cycle of the periodic noise generated from the noise source 20. For example, a pickup which detects the rotation of the engine or the like and directly generates the synchronization pulse. Although a sensor or the like can be considered, the present invention is not limited to this, and may include a sensor microphone that detects periodic noise and generate a synchronization pulse in synchronization with the detection period by detecting the period of the periodic noise from the detection signal. Good.

The controller 24 executes a new algorithm (Synchronized Filtered-x algorithm; hereinafter referred to as SFX algorithm) according to the present invention, which will be described later, and outputs a drive signal y _(n) for driving the speaker 26. The error sensor 28 is arranged in a predetermined region to be silenced, detects an interference sound wave (silence error) between the periodic noise from the noise source 20 and the additional sound generated from the speaker 28, and indicates the silencing error. The error signal e _(n) is output to the controller 24.

Next, the SFX algorithm will be described. The SFX algorithm also uses the conventional FX algorithm. The input x _(n) in the FX algorithm is

[0032]

[Equation 22]

If an impulse is used as shown in, the adaptive type
Filter output y of digital filter _(n)Is

[0034]

(Equation 23)

[0035] Where N is 1 of the periodic input signal
It is the number of points per cycle, and I is the tap length of the adaptive digital filter. Then, if the tap length of the adaptive digital filter is determined so that I = N is always satisfied, the following equation is obtained by

[0036]

[Equation 24]

It becomes as simple as Similarly, [Equation 16]
The reference signal r _(n) in the conventional FX algorithm shown in the equation is

[0038]

(Equation 25)

It becomes simple as shown in. SF like this
X algorithm is a filter in FX algorithm
Output y _(n)And reference signal r_(n)When creating
An algorithm that can eliminate the need for convolution
It is. Next, the SFX algorithm will be described with reference to FIGS.
4 will be used for further details.

FIG. 2 (A) shows an example of periodic noise to be silenced, and FIG. 2 (B) shows a synchronization pulse generated in synchronization with the periodic noise. When the sync pulse is input now, the current cycle T of the periodic noise can be detected from the time difference from the previously input sync pulse. Then, this cycle T is set to a predetermined sampling cycle τ (for example,
The value divided by the sampling frequency 2 kHz is the number of points N (= T / τ) described above (see FIG. 2C).

On the other hand, as shown in FIG. 3A, the filter coefficient (C
Let J be the tap length of ^). Then, in the case of J> N, as shown in FIG. 2B, the filter coefficient (C ^) is added one after another in a cycle of the number of points N to synthesize a coefficient (C
Create ~). Incidentally, this coefficient (C to) will be referred to as a pseudo period sequence.

If J <N, as shown in FIG. 2C, the number of points is added by adding 0 to the filter coefficient (C ^) of the tap length J by the number of points (N-J). Create N pseudo-periodic sequences (C-). A reference signal (r_ _(n) ) is created as shown in FIG. 4 from the pseudo period sequence (C to) created in this way. That is, as shown in the figure, a pseudo period sequence (C to) for two periods is stored, and the read range for one period (the number of points N) is shifted at every predetermined sampling period to shift the reference signal (r).
_ _(N) ) is created.

FIG. 5 is a block diagram showing an embodiment of the controller 24 shown in FIG. This controller 24
Is a register 30, a cycle detection circuit 32, a reference signal generation unit 34, a clock generator 36, a multiplier 38, 4
0, an adder 42, and a delay circuit 44. The register 30 corresponds to an adaptive digital filter, so that the sync pulse is applied from the sync detector 22 (FIG. 1) and the sampling clock having a predetermined sampling period is applied from the clock generator 36 ( 2B and 2C). Then, the register 30 uses k as a reference when the sync pulse is input (k = 0).
When the th sampling clock is input, the kth filter coefficient w _k is output.

Each time the cycle detecting circuit 32 inputs a sync pulse, it detects the current cycle T of the periodic noise from the time difference from the previously input sync pulse, and divides this cycle T by the sampling cycle τ. A signal indicating the number N (= T / τ) is output to the pseudo period sequence generation circuit 34A of the reference signal generation unit 34. The reference signal creation unit 34 is composed of a pseudo periodic sequence creation circuit 34A and registers 34B and 34C, and is provided with a synchronizing pulse and a sampling clock, and the period detection circuit 32 adds the number N of points.

The register 34B stores in advance the filter coefficient (C ^) of the tap length J indicating the transfer function from the speaker 26 to the error sensor 28, and the pseudo period sequence creating circuit 34A inputs the synchronizing pulse every time. , The number of points N added from the cycle detection circuit 32, and the register 34
3 based on the filter coefficients (C ^) stored in B
Execute the process shown in to create a pseudo period sequence (C ~),
The pseudo period sequences (C to) are stored in the two storage units of the register 34C.

When the k-th sampling clock is input to the register 34C from the pseudo-cycle sequence (C to) for the above two cycles, as shown in FIG. 4, when the synchronization pulse is input (k = 0), k-th reference signal (r_
_(n) ) is read. The filter coefficient (C ^) stored in the register 34B is obtained as follows by a known method.

FIG. 6 shows the speaker 26 to the error sensor 28.
It is a block diagram which shows the structure at the time of calculating the filter coefficient (C ^) which shows the transfer function C up to. As shown in the figure, the noise signal generator 50 outputs the noise signal x _(n) to the speaker 26 via the D / A converter 52 and the amplifier 54, and also has an adaptive type having a filter coefficient of the tap length J. It is output to the digital filter 56 and the control unit 58. still,
The noise signal generator 50 generates, for example, an M-sequence pseudo-random signal as a noise signal.

The noise signal detected by the error sensor 28 is applied to the positive input of the adder 64 via the amplifiers 60, 62. On the other hand, a filter output y _{(n) obtained} by filtering the input x _{(n )} from the adaptive digital filter 56 is added to the negative input of the adder 64.
Adds these input signals and outputs it as an error signal e _(n) to the control unit 58.

The control unit 58 controls the input signal x_(n)And error belief
Issue e_(n)Error signal e based on _(n)So that it becomes 0
To the adaptive digital filter by executing the LMS algorithm.
The filter coefficient of the filter 56 is updated. In this way
The filter coefficient of the adaptive digital filter 56 is sped up.
F which shows the transfer function C from the power supply 26 to the error sensor 28.
The filter coefficient (C ^) is identified, and the filter coefficient (C ^) is identified.
Ask for. Note that this filter coefficient (C ^) can be
Required when the silencer is activated.

Now, referring to FIG. 5, the error signal e _(n) input from the error sensor 28 (FIG. 1 ₎ and the reference signal (r_ _(n) ) output from the reference signal generator 34 are sampled at each sampling clock. Based on and
The multipliers 38 and 40 and the adder 42 are

[0051]

(Equation 26)

(Where μ = step size parameter)
The filter coefficient (w_ _{(n + 1)} ) is calculated based on
Incidentally, 27 is a setting device for setting the step size parameter, and outputs a signal indicating 2μ to the multiplier 38. The filter coefficient (w_ _{(n + 1)} ) calculated in this way is
It is output to the register 30 via the delay circuit 44, the filter coefficient (w_ (n _{+ 1))} by the filter coefficient register 30 (w_ _(n)) is updated.

Next, the processing procedure in the controller 24 will be described with reference to the flowcharts shown in FIGS. When the sync pulse synchronized with the cycle of the periodic noise is input from the sync detector 22 (FIG. 1), the process of interrupt A shown in FIG. 7 is executed. As shown in the figure, when the processing of the interrupt A is first started, other interrupts are prohibited (step 100). Then, the current period T of the periodic noise is detected from the time difference from the previously input synchronization pulse, and the number N of points obtained by dividing the period T by the sampling period τ
(= T / τ) is calculated (step 102).

Next, at step 104, the calculated point number N and the error sensor 2 from the speaker 26 are detected.
The tap length J of the filter coefficient (C ^) indicating the transfer function up to 8 is compared, and when N <J, the filter coefficient (C
From ^),

[0055]

[Equation 27]

Based on, the pseudo period sequence (C
~) Is created (step 106), and if N> J,
From the second filter coefficient (C ^),

[0057]

[Equation 28]

On the basis of
~) Are created (step 108). Next, the calculated pseudo period sequence (C to) is stored in each of the two storage units of the register 34C (step 110), and subsequently, a counter for counting the number of sampling clocks from the input of the synchronization pulse (FIG. After setting the count value k (not shown) to 0, interruption is enabled (steps 112 and 11).
4).

On the other hand, when a sampling clock having a predetermined sampling period is input from the clock generator 36 (FIG. 5), the processing of interrupt B shown in FIG. 8 is executed. As shown in the figure, when the processing of the interrupt B is first started, other interrupts are prohibited (step 200). Then, the kth filter coefficient w _k is output from the register 30 based on the count value k of the counter (step 202).

Next, the error signal e _(n) input from the error sensor 28 (FIG. 1 ₎ and the reference signal generating unit 34
K-th reference signal (r_ _(n) ) output from
From the following equation,

[0061]

[Equation 29]

The filter coefficient (w_ _{(n + 1)} ) is calculated based on the above, and the register 3 is calculated with this filter coefficient (w_ _{(n + 1 )} ).
The filter coefficient (w_ _(n) ) of 0 is updated (step 2)
04). Then, after incrementing the count value k of the counter by 1, the interrupt is enabled (step 20).
6, 208). In the above embodiment, the case where the number of noise sources, the number of error sensors, and the number of speakers are each one has been described, but the SFX algorithm can be applied to a so-called MEFX algorithm used when there are a plurality of noise sources.

Next, a case where the SFX algorithm according to the present invention is applied to an error scanning (hereinafter referred to as ES) algorithm will be described. The ES algorithm is an algorithm disclosed in Japanese Patent Laid-Open No. 3-274897, and when a plurality of error sensors are provided,
In the process of updating the filter coefficient for each sampling, pay attention to the instantaneous error output of an error sensor.Since all the information related to this error output is known from the system configuration, this error output and the input indicating noise are input. Based on this, a filter coefficient of the adaptive digital filter is calculated according to a predetermined algorithm (for example, FX algorithm), and the filter coefficient is updated with the calculated filter coefficient. At the time of the next sampling, attention is paid to another error sensor, and the same algorithm as described above is executed. That is, this is an algorithm for updating the filter coefficient while scanning the error sensors one after another.

FIG. 9 is a block diagram showing the configuration of the main part when the SFX algorithm according to the present invention is applied to the ES algorithm, showing the case where the number of noise sources is 1, the number of error sensors is 2, and the number of speakers is 2. There is. As shown in the figure, the electronic muffler mainly includes a controller 7
1, 72, speakers 81 and 82, and an error sensor 91,
It is composed of 92.

The error sensors 91 and 92 are arranged in predetermined regions to be muted, respectively, and detect a sound in which noise from a noise source interferes with additional sound from the speakers 81 and 82,
The error signals e _{1 (n)} and e _{2 (n)} indicating the interference sound are amplified by the amplifier 8
3 and 84 and A / D converters 85 and 86 to output to the controllers 71 and 72. Controllers 71 and 7
Reference numeral 2 denotes registers 71A and 72A, reference signal generation units 71B, 71C and 72B, 72C, ES, respectively.
It is composed of ES algorithm execution units 71D and 72D for executing the algorithm.

The registers 71A and 72A and the reference signal generators 71B, 71C and 72B, 72C are
Since it has the same functions as the register 30 and the reference signal generation unit 34 of the controller 24 shown in FIG. 5, detailed description thereof will be omitted here. However, the reference signal generators 71B, 71C, 72B, and 72C are provided with filter coefficients (C ^ ₁₁ ), (C ^ ₂₁ ), (C ^ ₂₁ ), (C ^ ₂₁ ), (C ^ ₂₁ ), which indicate transfer functions between the speakers 81 and 82 and the error sensors 91 and 92, respectively.
A register (not shown) for storing ^ ₁₂ ) and (C ^ ₂₂ )
have.

Reference signal producing sections 71B, 71C,
Each of 72B and 72C inputs a pseudo pulse sequence (C to ₁₁ ) each time a synchronization pulse synchronized with the period of the periodic noise is input.
(C to ₂₁ ) (C to ₁₂ ) and (C to ₂₂ ) are created, and thereafter, every time the internal sampling clock is input, a pseudo period sequence (C to ₁₁ ) (C to ₂₁ ) (C to ₁₂ ) and reference signal by shifting the _{(C~ 22) (r_ 11)} ,
(R_ _21), create a (r_ ₁₁₎ and (r_ _22), and outputs the ES algorithm execution unit 71D, the 72D.

The ES algorithm execution unit 71D uses the MEF
The register 71A by the adaptive algorithm (ES algorithm) that equivalently approximates the X algorithm in the adaptive process.
Those of calculating the filter coefficients (w_ _11), to update the filter coefficients of the register 71A at the calculated filter coefficients _{(w_ 11) (w_ 11)} , said reference signal (r_ _11), predetermined and (r_ ₂₁₎ The ES algorithm shown in the following equation is executed based on the error signals e ₁ (n) and e ₂ (n) sampled at the sampling period of.

[0069]

[Equation 30]

[0070]

[Equation 31]

That is, at a certain time (n) at the time of sampling, the filter coefficient (w_ ₁₁ ) is given by the formula (30).
_n , reference signal (r_ ₁₁ ) _n and error signal e
Calculate the filter coefficient (w_ ₁₁ ) _{n + 1} based on _{1 (n)} ,
At the next sampling time (n + 1), the filter coefficient (w_ ₁₁ ) _{n + 1} , the reference signal (r_ ₂₁ ) _{n + 1} and the error signal e _{2 (n + 1} ) are obtained as shown in the equation (31). ₎ , The filter coefficient (w_ ₁₁ ) _{n + 2} is calculated.

As described above, the ES algorithm focuses on the error signal of one error sensor for each sampling period, updates the corresponding filter coefficient by the FX algorithm by the reference signal relating to this error signal, and then performs the next sampling. At times, focusing on another error sensor, the same processing as above is executed. In the case of the MEFX algorithm that updates the filter coefficient by using a plurality of error signals e _{1 (n)} and e _{2 (n)} at the same time,

[0073]

[Equation 32]

And the calculation amount in one sampling period
Compared to the above [Equation 30] and [Equation 31]
Increase with the number of sensors. Well, ES algorithm
The execution unit 71D includes the calculation units 73, 74, 75 and the selection unit 7.
6, and the calculation unit 73 makes a reference at a certain time (n).
Signal (r_₁₁)_nAnd error signal e_{1 (n)}Based on [number
[30] The second term on the right side of the equation is calculated, and this is selected by the selection unit 7
It outputs to the calculating part 75 via 6. The calculation unit 75 is a fill
Coefficient (w_₁₁)_nHas a storage unit for storing
Filter coefficient (w_₁₁)_nAnd the selection unit 7
Add the output from 6 and add this to the new filter coefficient
(W_₁₁) _{n + 1}As the filter coefficient (w_
₁₁)_{n + 1}At the next time (n + 1) the filter of register 71A
Transferred as a coefficient, and the filter coefficient of the register 71A is updated.
Do new.

Further, the arithmetic unit 73 has a reference signal (r_ ₂₁ ) _{n + 1} and an error signal e _{2 (n + 1) at the} next time _{(n + 1).}
The second term on the right side of [Equation 31] is calculated based on
This is output to the arithmetic unit 75 via the selection unit 76, and the arithmetic unit 75 performs the same processing as described above to update the filter coefficient of the register 71A. The other ES algorithm execution unit 72D also performs the same process as the ES algorithm execution unit 71D to update the filter coefficient of the register 72A.

[0076] registers 71A and 72A outputs a drive signal by shifting the filter coefficients respectively updated as described above (w_ ₁₁₎ and (w_ ₂₁₎ in each sampling period, these drive signals D It is output to the speakers 81 and 82 via the / A converters 83 and 84 and the amplifiers 85 and 86. In this way the speaker 8
1 and 82 are driven, and the additional sounds emitted from the speakers 81 and 82 interfere with each other so as to cancel the periodic noise in a predetermined area where the error sensors 91 and 92 are arranged.

In the case of the conventional ES algorithm,
The convolution operation for creating the reference signal was indispensable in all the reference creating units for each sampling cycle. However, when the SFX algorithm according to the present invention is applied to the ES algorithm, the SFX algorithm does not require the convolution operation in the first place. Therefore, it is possible to easily realize multi-channel without increasing the calculation amount.

Further, in the present embodiment, the sampling period is fixed to a fixed value and the tap length of the register 30 or the like is made variable according to the number of points N corresponding to the period of periodic noise. However, the present invention is not limited to this. The tap length of the register or the like may be fixed to a predetermined number, and the sampling period may be variable corresponding to the period of periodic noise. In this case, for example, the register 30
I is the tap length, and T is the period detected from the periodic noise.
_{Letting n} be the period T _n divided by the tap length I, the time τ _n is set as the sampling period. Further, _assuming that the tap length of the filter coefficient (C ^) obtained by sampling the transfer function from the speaker to the error sensor at the sampling period τ _n is J, the tap length I and the tap length J are compared, and I <J In this case, from the filter coefficient (C ^),

[0079]

[Expression 33]

Based on, the pseudo period sequence (C
~) Is created, and in the case of I> J, from the filter coefficient (C ^), the following equation,

[0081]

(Equation 34)

Based on, the pseudo period sequence (C
~), And the pseudo periodic sequence (C
It is necessary to create a reference signal based on ~).

[0083]

As described above, according to the electronic silencing method and apparatus of the present invention, an impulse synchronized with the fundamental period of periodic noise or pseudo periodic noise is generated, and each time the fundamental period is detected. , The tap length of the adaptive digital filter and the tap length of the reference signal are made to match based on the detected period and theoretically treated as an FX algorithm. The convolution operation in the digital filter and the digital filter used to create the reference signal can be omitted,
The amount of calculation for calculating the filter output of the adaptive digital filter can be significantly reduced, and a silencing effect similar to that of the FX algorithm can be obtained.

In the general application, the convergence speed in the LMS algorithm is determined by the upper limit of the step size parameter μ depending on the maximum eigenvalue of the input signal, and the eigenvalue cannot be specified. It is often difficult to increase the step size parameter μ, but since the SFX algorithm according to the present invention uses impulses as virtual inputs, there is no limit due to extremely large eigenvalues and eigenvalue uncertainties. Therefore, the step size parameter μ can be increased, and there is an advantage that the convergence is faster than that of the normal LMS algorithm.

[Brief description of drawings]

FIG. 1 is a block diagram showing a basic configuration of an electronic silencer according to the present invention.

FIG. 2 (A) is a waveform diagram showing an example of periodic noise to be silenced, and FIG. 2 (B) shows a synchronization pulse generated in synchronization with the periodic noise. (C) is a timing chart showing a sampling clock.

3A to 3C are signal waveform diagrams used to explain an SFX algorithm according to the present invention.

FIG. 4 is a diagram used for explaining a method of creating a reference signal (r_ _(n) ) from a pseudo period sequence (C to).

5 is a block diagram showing an embodiment of the controller shown in FIG.

FIG. 6 is a block diagram showing a configuration for obtaining a filter coefficient (C ^) indicating a transfer function C from a speaker to an error sensor.

FIG. 7 is a flowchart showing a processing procedure executed by the controller shown in FIG. 1 when a synchronization pulse is input.

FIG. 8 is a flowchart showing a processing procedure executed by the controller shown in FIG. 1 when a sampling clock is input.

FIG. 9 shows an ES of the SFX algorithm according to the present invention.
It is a block diagram which shows the principal part structure when applying to an algorithm.

FIG. 10 is a block diagram showing a basic configuration of a conventional electronic silencer.

FIG. 11 is an overall configuration diagram showing an example of a conventional electronic silencer for silencing periodic noise.

12 is a block diagram showing details of FIG. 11. FIG.

[Explanation of symbols]

20 ... Noise source 22 ... Synchronous detector 24, 71, 72 ... Controller 26, 81, 82 ... Speaker 28, 91, 92 ... Error sensor 30, 34B, 34C, 71A, 72A ... Register 32 ... Cycle detection circuit 34, 71B , 71C, 72B, 72C ... Reference signal creation unit 34A ... Pseudo periodic sequence creation circuit 36 ... Clock generator 38, 40 ... Multiplier 42 ... Adder 44 ... Delay circuit 71D, 72D ... ES algorithm execution unit

Claims (12)

(57) [Claims] 1. An electronic noise suppressor, which produces an additional sound having the same sound pressure in a phase opposite to that of a periodic noise or a pseudo periodic noise generated from a noise source, and muffles the sound by a sound wave interference in a predetermined region. A drive signal for driving the speaker is generated by filtering a predetermined input signal with an adaptive digital filter having a first filter coefficient that is sequentially updated every predetermined sampling period; The filter coefficient of 1 represents the input signal by a digital filter having an error signal from an error sensor arranged in the predetermined region for detecting an interference sound wave and a second filter coefficient indicating a transfer function from the speaker to the error sensor. Based on the reference signal obtained by filtering and the current first filter coefficient In the electronic silencing method updated by: detecting the period of the periodic noise or the pseudo periodic noise for each period, the adaptive digital filter and the input signal of the digital filter are generated in synchronization with the period detection. An electronic silencing method characterized in that, every time the period is detected, the tap length of the adaptive digital filter and the tap length of the reference signal are made to match each other based on the detected period. . 2. The tap length of the adaptive digital filter is I, the number of points corresponding to the detected period is N,
When the second filter coefficient is (C ^), the tap length I is made to match the point number N, and the point number N and the tap length J of the digital filter are compared. If N <J From the second filter coefficient (C ^), A pseudo-periodic sequence (C to) of the number of points N is created based on the following, and when N> J, from the second filter coefficient (C ^), the following equation, The electronic muffling method according to claim 1, wherein a pseudo period sequence (C-) having a number of points N is created based on the above, and the reference signal is produced based on the created pseudo period sequence (C-). 3. When the tap length of the adaptive digital filter is I and the detected cycle is T _n , the cycle T _n
Was the time tau _n divided by the tap length I in the sampling period, when the tap length of filter coefficients a transfer function sampled at the sampling period tau _n from the speaker to the error sensor (C ^) and J, the The tap length I and the tap length J are compared, and if I <J, from the second filter coefficient (C ^), the following equation, A pseudo period sequence (C to) having a tap length I is created based on the following equation. When I> J, from the second filter coefficient (C ^), the following equation, The electronic muffling method according to claim 1, wherein a pseudo period sequence (C-) having a tap length I is created based on the above, and the reference signal is produced based on the created pseudo period sequence (C-). 4. The generation of the reference signal based on the pseudo period sequence (C˜) is performed by using the pseudo period sequence (C˜) as (r_ _(n) ) = [r _(n) , r _(n-1)). ,…, R _{(n -I + 1)} ]
Then, the electronic muffling method according to claim 2 or 3, wherein the (r_ _(n) ) is sequentially created by shifting each of the sampling periods. 5. The error signal detected by the error sensor is e _(n) , the reference signal is (r_ _(n) ), and the current first filter coefficient is (w_ _(n ) every predetermined sampling period. ₎ ), The following equation, (However, the filter coefficient (w_ _{(n + 1)} ) is calculated based on (where, μ = step size parameter), and the first filter coefficient (w_ _(n) ) is calculated by the filter coefficient (w_ _{(n + 1 )} ).
The electronic muffling method according to claim 1, 2, 3, or 4, wherein 6. A period detecting means for detecting a period of periodic noise or pseudo periodic noise generated from a noise source, and a speaker for radiating an additional sound for canceling a sound wave propagating from the noise source in the predetermined region. An error sensor provided in the predetermined area for detecting an interference sound wave of a sound wave propagating from the noise source and an additional sound from the speaker; and a first filter coefficient that is sequentially updated every predetermined sampling period ( w_ _(n) ), which sequentially reads out the first filter coefficient (w_ _(n) ) at each sampling cycle, and uses this as a drive signal for driving the speaker, First storage means for storing a second filter coefficient (C ^) obtained by sampling the transfer function from the speaker to the error sensor at the sampling cycle, and the cycle detection Whenever the means detects a cycle, N is the number of points corresponding to the detected cycle, and J is the tap length of the second filter coefficient (C ^).
Is compared with the tap length J of the digital filter, and if N <J, then from the second filter coefficient (C ^), the following equation, A pseudo-periodic sequence (C to) with the number of points N is created based on the following equation. When N> J, the following equation is obtained from the second filter coefficient (C ^). Pseudo-periodic sequence creating means for creating a pseudo-periodic sequence (C to) of the number of points N based on the above, and the stored contents of the pseudo-periodic sequence every time the pseudo-periodic sequence (C to) is created. Second update to (C ~)
And the pseudo period sequence (C-) stored in the second storage unit for each sampling period, and the shifted pseudo period sequence (C-) is used as a reference signal (r_).
_(n) ), reference signal output means for outputting as _(n) ), an error signal e _{(n )} detected by the error sensor at each sampling cycle, and the reference signal (r_
_(n) ), the next filter coefficient (w_ _{(n + 1)} ) is calculated based on the current filter coefficient (w_ _(n) ), and the driving is performed by the calculated filter coefficient (w_ _{(n + 1)} ). An electronic silencing apparatus comprising: a computing means for updating the first filter coefficient (w_ _{(n + 1)} ) of the signal creating means. 7. The second storage means stores the pseudo-periodic sequence (C to) over a length of two cycles, and the reference signal output means stores one cycle out of the stored two cycles. 7. The reference signal (r_ _(n) ) is read while shifting the read range.
Electronic silencer. 8. The calculating means is the following equation: 8. The electronic muffler according to claim 6, wherein the filter coefficient (w_ _{(n + 1)} ) is calculated based on (where, μ = step size parameter). 9. An additional sound having the same sound pressure as a phase opposite to that of periodic noise or pseudo-periodic noise generated from a noise source in a region where sound waves can propagate in a three-dimensional direction is generated from a speaker. In the electronic silencing method for silencing sound waves in a predetermined area by the sound wave interference, the drive signal for driving the speaker is generated by an adaptive digital filter having a first filter coefficient that is sequentially updated every predetermined sampling period. The first filter coefficient is generated by filtering a predetermined input signal, and the first filter coefficient is provided between the error signals from a plurality of error sensors arranged in the predetermined area and detecting interference sound waves, and from the speaker to each error sensor. Obtained by filtering the input signal with a plurality of digital filters having a second filter coefficient representing a transfer function Reference signal and the current first
In the electronic silencing method that is updated based on the filter coefficient of, the period of the periodic noise or the pseudo periodic noise is detected for each period, and the input signal of the adaptive digital filter and the digital filter is detected by the period detection. And the tap length of the adaptive digital filter is matched with the tap length of the reference signal on the basis of the detected period every time the period is detected. Error sensor with one or more first
Error sensor, one or more second error sensors ...
In each sampling period, the filter coefficient of the adaptive digital filter is updated based on only the information relating to the first error sensor at a certain sampling time, and the second error sensor is updated at the next sampling time. The filter coefficient of the adaptive digital filter is updated only on the basis of the information concerned, and the filter coefficient of the adaptive digital filter is updated by sequentially repeating this for each of the divided error sensors. Electronic silencing method. 10. When the tap length of the adaptive digital filter is I, the number of points corresponding to the detected cycle is N, and the second filter coefficient is (C ^), the tap length is the number of points N. Match I
The number N of points is compared with the tap length J of the digital filter. If N <J, then from the second filter coefficient (C ^), the following equation, Pseudo-periodic sequence (C-) with the number of points N is created based on the following, and when N> J, from the second filter coefficient (C ^), the following equation, 10. The electronic muffling method according to claim 9, wherein a pseudo period sequence (C to) having the number of points N is created based on the above, and the reference signal is produced based on the created pseudo period sequence (C to). 11. When the tap length of the adaptive digital filter is I and the detected cycle is T _n , the cycle T
_When the time τ _n obtained by dividing _n by the tap length I is set to the sampling period, and the tap length of the filter coefficient (C ^) obtained by sampling the transfer function from the speaker to the error sensor at the sampling period τ _n is J, The tap length I and the tap length J are compared, and if I <J, from the second filter coefficient (C ^), the following equation, A pseudo period sequence (C to) having a tap length I is created based on the following equation. When I> J, from the second filter coefficient (C ^), the following equation, 10. The electronic muffling method according to claim 9, wherein a pseudo period sequence (C-) having a tap length I is created based on, and the reference signal is produced based on the created pseudo period sequence (C-). 12. The error signal of the first error sensor at a certain sampling time (n) is e _{1 (n)} and the second error at the next sampling time (n + 1) is determined every predetermined sampling period. The error signal of the sensor is e _{2 (n + 1)} , ..., The reference signal at a certain sampling time (n) is (r_1 _(n) ), and the reference signal at the next sampling time (n + 1) is (r_
_{2 (n + 1)} ), ..., the filter coefficient of the adaptive digital filter is given by the following equation: 12. The electronic muffling method according to claim 9, 10 or 11, which is sequentially updated based on (where, μ = step size parameter).