JP3364325B2 - Active noise control 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 active noise control system using an adaptive digital filter constructed by using a microprocessor, a digital signal processor and the like.

[0002]

2. Description of the Related Art FIG. 7 is a block diagram of a duct silencing system using an active noise control device according to the prior art (Japanese Patent Laid-Open No. 5-67948). There is a noise source 1 in the duct 2, a microphone 3 as a noise signal producing means, a speaker 4 as a secondary sound wave outputting means, and a microphone 5 as an error signal producing means are arranged in this order in the duct opening direction. Each of the microphone 3 and the microphone 5 includes an amplifier, a filter, and an A / D converter (not shown), and the noise signal u (n) and the first error signal e ₁ (n) converted into digital signals, respectively. Is output.

The noise signal u (n) is subjected to a convolution operation with the filter coefficient sequence a _i (n) by the equation (1) in the first transversal filter 6 which is a signal processing means, and the operation result is the control signal. It is output to the speaker 4 as y (n).

[Equation 1] Here, N represents the number of taps of the first transversal filter 6.

The speaker 4 includes a D / A converter, a filter, and an amplifier, which are not shown, and receives the control signal y (n) to output 2 signals.
Generates the next sound wave. The primary sound wave emitted from the noise source 1 interferes with the secondary sound wave emitted from the speaker 4, and the microphone 5 detects the interference result.

The adaptive means is roughly divided into two blocks. One of them is as follows. Noise signal u (n)
Is filtered by a filter 7 having a transfer function H (z) to create a _first reference signal r ₁ (n). The transfer function H (z) is a transfer function in the process in which the control signal y (n) is detected by the microphone 5, and also includes the characteristics of the amplifier, filter, and A / D converter included in the microphone 5.

The first coefficient updating unit 8 uses the first reference signal r ₁ (n) and the first error signal e ₁ (n) to calculate the filter coefficient sequence a _i (i = 0 by equation (2). , 1,2, ..., N-1) are updated. a _i (n + 1) = a _i (n) −μr ₁ (ni) e ₁ (n) (2) where μ is a small positive value. With this update formula,
The root mean square value of the first error signal e ₁ is gradually minimized.

Another block of the adapting means is for suppressing an increase in the gain of the first transversal filter 6 in the frequency band where the sound is not muted, and is as follows. The control signal y (n) is filtered by the filter 9 that passes only the frequency band in which the gain increase of the first transversal filter 6 is to be suppressed, and the second reference signal r ₂ (n) is created. The second reference signal r ₂ (n) is input to the second transversal filter 10.

The second transversal filter 10 is
It has the same filter coefficient sequence a _i as the first transversal filter 6. In other words, the first and second transversal filters 6 and 10 share the filter coefficient sequence a _i . In the second transversal filter 10, the calculation of the equation (3) is performed, and the second error signal e ₂ (n) is created.

[Equation 2]

The second coefficient updating section 11 uses the second reference signal r ₂ (n) and the second error signal e ₂ (n), and formula (4)
Thus, the filter coefficient sequence a _i (i = 0, 1, 2, ..., N-1) is updated. a _i (n + 1) = a _i (n) −νr ₂ (ni) e ₂ (n) (4) where ν is a small positive value. With this update formula,
The root mean square value of the second error signal e ₂ is gradually minimized.

By repeating such a series of processing,
The expression (4) suppresses the influence on the frequency components other than the sound deadening target, and the expression (2) optimizes the gain in the frequency band of the sound deadening target to perform the sound deadening.

Although the above filter operation is processed by programming of a microprocessor or a digital signal processor, it is necessary to execute the update of the filter coefficient a _{i in} the equations (2) and (4) in one sampling period. Too much. For this reason, it is difficult to shorten the sampling interval, and there is an inconvenience that the time required to obtain a desired silencing effect becomes long. Therefore, it has been proposed to reduce the amount of calculation by alternately updating the filter coefficients of equations (2) and (4) for each sampling.

[0012]

In the prior art, the equation
Although updating the two systems of the filter coefficient a _i shown in (2) and equation (4) is useful for improving the stability of the operation of the active noise control system, the calculation necessary for updating the filter coefficient is performed. There was a problem that the amount increased. Further, in the conventional technology, the formula (2)
And a method of reducing the calculation amount per sampling by alternately updating the filter coefficient of the equation (4) for each sampling, in this case, the first purpose is Since the update of the filter coefficient that minimizes the root mean square value of the error signal e ₁ is performed only once in the two samplings, the minimization speed of the root mean square value of the first error signal is Become slow. In other words, there is a problem that it takes about twice as long to obtain the same sound deadening effect.

The present invention has been made in consideration of the above-mentioned problems.
The present invention provides an active noise control method and device that can realize updating of filter coefficients having the same effect as updating of filter coefficients of (2) and equation (4) with a smaller amount of calculation.

[0014]

According to the present invention, in an active noise control device that emits a secondary sound wave to mitigate by sound wave interference with respect to a primary sound wave emitted from a noise source, there is a correlation with the primary sound wave. Noise signal generating means for outputting the noise signal u, signal processing means for filtering the noise signal u with a first transversal filter having a predetermined filter coefficient sequence a _i , and outputting a control signal y, and the control signal A secondary sound wave output means for inputting y to generate a secondary sound wave and an error signal generating means for outputting a _first error signal e ₁ correlated with the interference result of the primary sound wave and the secondary sound wave are constituted. in adaptation means, the first and the reference signal r ₁ obtained by filtering the noise signal u by the transfer function H of the process of being detected (z) in the control signal y is the error signal creating means, first having a predetermined frequency component 2 Sum signal r ⁺ and a difference signal r of the reference signal r ₂ ^- a, and the first and the error signal e _1, the second reference signal r ₂ of the first transversal filter same filter coefficient sequence and a _i A sum signal e ⁺ and a difference signal e ⁻ with the second error signal e ₂ filtered by the second transversal filter are created, and using these four signals r ⁺ , r ⁻ , e ⁺ , e ⁻ , The update amount of the filter coefficient sequence a _i is calculated.

At this time, the i-th filter coefficient a _i (n) at the time n is expressed by the following two equations a _i (n + 1) = a depending on whether n−i is an even number or an odd number. ^{_{i (n) -μr + (ni}} ) e + (n) (5) a i (n + 1) = a i (n) -μr - (ni) e - a (n) (6) using alternately Update. Where μ is a number with a small positive value.

Equations (5) and (6) can be used alternately depending on whether n is an even number or an odd number of n instead of ni.

As a method for creating the _second reference signal r ₂ , a frequency component of the control signal y is band-limited by a filter, and a frequency component of the input signal u is band-limited by a filter. There is a method for creating white noise by band limiting with a filter, and a method for creating white noise by limiting the frequency component of the repetitive signal of the filter coefficient sequence a _i with a filter. Further, the level of the second reference signal r ₂ is set to an appropriate level according to the degree of influence of the update of the filter coefficient of the equation (4), and the band limiting filter plays the role of this level setting. It also includes.

Further, level detecting means for detecting the amplitude level of the second error signal e ₂ is provided, and when the level is below a predetermined reference, the i-th filter coefficient a _i at time n of the adaptive means. (n) is a formula shown in the prior art
It can also be updated by (2).

The update of the filter coefficient a _i is performed by the conventional method.
Even when (2) and the equation (4) are alternately used for each sampling, when the amplitude level of the _second error signal e ₂ is below a predetermined reference, the equation (4) is not applied and the equation (2) is not applied. It is good to use only.

[0020]

According to the present invention, the reference signal r ₁ at time n (n), r ₂ sum signal of (n) r ⁺ (n) and the difference signal r ^- (n), and the error signal e ₁ (n) , E ₂ (n) sum signal e ⁺ (n) and difference signal e
^- (n), respectively _{^{r + (n) = r 1}} (n) + r 2 (n) (7) r - (n) = r 1 (n) -r 2 (n) (8) e + (n ) = E ₁ (n) + e ₂ (n) (9) e ⁻ (n) = e ₁ (n) −e ₂ (n) (10).

In the prior art, among the 2N times of filter coefficient updating operations by the equations (2) and (4), r ₁ related to the sound deadening object is involved only in the N times of the equation (2). Both r ⁺ and r ⁻ include r ₁ , and when the filter coefficient a _i is updated using these, r ₁ related to the muffling target is always involved. Therefore, the update of the filter coefficient a _i always reflects the _first reference signal r ₁ related to the muffling target, and the calculation efficiency for muffling is doubled.

When the equations (5) and (6) are alternately used to update the filter coefficient a _i , the update amount δ ⁺ _i (n) of the filter coefficient at the i-th tap in the equation (5) and the equation (6 ), The update amount δ ^- _i (n) of the filter coefficient at the i-th tap is expressed by equation (11) and equation (12)
It is represented by. δ ⁺ _i (n) = δ _i (n) -μ (r ₁ (ni) e ₂ (n) + r ₂ (ni) e ₁ (n)) (11) δ ^- _i (n) = δ _i (n ) + Μ (r ₁ (ni) e ₂ (n) + r ₂ (ni) e ₁ (n)) (12) where δ _i (n) is the same as the expression (2) and the expression (4) in the prior art. It is the update amount when used, and μ = ν is assumed.

Update amounts of these filter coefficients δ ⁺ _i , δ ^- _i
Compared with the update amount of the prior art, Equation (11) and Equation (1
Although it differs by the second term on the right-hand side of 2), equations (5) and (6)
Since the updating is performed alternately, these quantities cancel each other out in the adaptation process, and a convergence value of the filter coefficient that is almost equivalent to that in the prior art is obtained. Furthermore, equation (5)
Since the update amount for minimizing the first error signal is included in each of the update equations of and (6), even if equation (5) and equation (6) are executed alternately The speed of obtaining the muffling effect is hardly deteriorated, and the number of times of updating the filter coefficient required to obtain the same muffling effect is reduced to half that of the conventional technique.

The same effect can be obtained by alternately using the equations (5) and (6) based on the even-odd number of n-i or the even-numbered odd number of n. However, when they are alternately used based on ni, the storage capacity for holding the sum signal r ⁺ and the difference signal r ⁻ of the reference signal can be halved.

The present invention is intended to suppress a gain increase in a specific frequency band of the first transversal filter, and the second reference signal r ₂ has only the frequency component to be controlled. It is required to be a signal. If the control signal y, which is the output of the first transversal filter, is filtered by the filter that passes only the frequency component to be suppressed, when the frequency component appears in the control signal y, it is attempted to suppress it. The action works, and the gain increase is suppressed.

When the noise signal u is used instead of the control signal y, when a frequency component to be suppressed appears in the noise signal u, an action of suppressing the frequency component works. Control signal y
When white noise is used instead of, the effect of always suppressing the gain increase of the frequency component to be suppressed works. Further, when the repetitive signal of the filter coefficient sequence a _i is used instead of the control signal y, if the gain of the frequency component to be suppressed increases in the first transversal filter, the effect of suppressing it will be obtained. Works. With either method, the gain increase in the frequency band that is not targeted for muffling is appropriately suppressed.

A level detecting means for detecting the amplitude level of the second error signal e ₂ is provided, and when the detected level falls below a predetermined reference, the filter coefficient a _i is updated by the formula (2) shown in the prior art. (11)
Also, the influence of the second term on the right side in Expression (12) is reduced.
Immediately after the start of the control, the signal in the frequency band in which the gain increase should be suppressed is not included in the control signal y so much that the contribution to the _second error signal e ₂ is small. Therefore, by using this method, a desired silencing effect can be achieved quickly. After that, when the gain of the frequency component to be suppressed increases, the filter coefficient a _i is updated by the equations (5) and (6), and the appropriate filter coefficient a _i is always set.

Equation (2) is used when the amplitude level of the second error signal e ₂ is below a predetermined standard, and otherwise, equation (2) and equation (4) are alternated for each sampling as in the conventional case. Even with the configuration in which the filter coefficient a _i is updated by using the equation (2), the time required to achieve the desired silencing effect is shortened since the equation (2) is mainly applied immediately after the control is started.

[0029]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 is a block diagram when an active noise control system according to a first embodiment of the present invention is applied to muffling a duct. There is a noise source 1 in the duct 2, and a microphone 3 as a noise signal generating means for receiving a noise from the noise source 1 and converting the noise into a signal in order toward the duct opening direction, and a sound wave for canceling the noise. A speaker 4 is provided as a secondary sound wave output means for generating, and a microphone 5 is provided as an error signal output means for receiving a noise and a secondary sound wave and outputting an error between them as a signal.

Each of the microphone 3 and the microphone 5 includes an amplifier, a filter and an A / D converter (not shown) and outputs a digitized signal. The microphone 3 outputs a noise signal u (n), and the microphone 5 outputs a first error signal e ₁ (n). Here, n in parentheses represents the nth sampling, and the same applies to the symbols used below.

The signal processing means comprises a first transversal filter 6 having N taps. In the first transversal filter 6, the noise signal u (n) is subjected to a convolution operation with the filter coefficient sequence a _i according to the equation (1), and the operation result is output to the speaker 4 as the control signal y (n). The speaker 4 includes a D / A converter, a filter and an amplifier (not shown), and receives the control signal y (n) to generate a secondary sound wave. The primary sound wave emitted from the noise source 1 interferes with the secondary sound wave emitted from the speaker 4 and is attenuated. The microphone 5 detects this interference result.

In the adaptive means, the noise signal u (n) is filtered by the filter 7 having the transfer function H (z),
A first reference signal r ₁ (n) is created. The transfer function H (z) is a transfer function in the process in which the control signal y (n) is detected by the microphone 5, and also includes the characteristics of the amplifier, filter, and A / D converter included in the microphone 5. Such a filter 7 has a finite impulse response (F
(IR) digital filter or infinite impulse response (IIR) digital filter.

The control signal y (n) is filtered by the filter 9 that passes only the frequency band in which the gain increase of the first transversal filter 6 is to be suppressed, and the second reference signal r ₂ (n) is created. To be done. This filter 9 is also an FIR digital filter or I
It is composed of an IR digital filter. The level of the second reference signal r ₂ (n) is appropriately set by the filter 9.

The second reference signal r ₂ (n) is input to the second transversal filter 10. The second transversal filter 10 shares the same filter coefficient sequence a _i as the first transversal filter 6. In the second transversal filter 10, the calculation of the equation (3) is performed, and the second error signal e ₂ (n) is created.

The first reference signal r ₁ (n) and the second reference signal r ₂ (n) are added by the adder 13,
The sum signal r ⁺ (n) is output. On the other hand, the subtracter 14 takes the difference between the first and second reference signals r ₁ (n) and r ₂ (n) and outputs the difference signal r ⁻ (n). First error signal e
Similarly, ₁ (n) and the second error signal e ₂ (n) are added by the adder 15 and the sum signal e ⁺ (n) is added to the subtracter 16
Then, the difference is taken and the difference signal e ⁻ (n) is output. These sum signal and difference signal are defined by equations (7) to (10).

In the coefficient updating unit 12, these four signals r
^{+ (N), r - (} n), e + (n), e - enter the (n), the filter coefficient sequence using equation (5) and _{(6) a i (i =} 0,1 , 2, ..., N-1)
Will be updated. At this time, the filter coefficient sequence a _i is updated by alternately using the equations (5) and (6) depending on whether n is an even number or an odd number. In other words, equation (5) for each sampling
And equation (6) are applied alternately.

Equation (5) may be used when n is an even number, and equation (6) may be used when n is an odd number, and conversely, equation (5) may be used when n is an odd number.
Equation (6) may be used when (5) is used and even numbers are used. In either case, the update of the filter coefficient sequence a _i in the n-th sampling is performed according to one of the equations (5) and (6), and in the next sampling is performed according to the other equation.

Filter coefficient sequence a according to equation (5) or equation (6)
_i The update of the past N number of the sum signal, that is ^{r + (n-N + 1} ) from r ⁺ to (n), or past the N difference signal namely r ^- (n-
N + 1) to r ⁻ (n) are required. In one update of the filter coefficient sequence, only one set of these, for example, N sum signals is used, but in the next update of the filter coefficient sequence, the other set is used. Sum signal r
⁺ And the difference signal r ^- the N content must be constantly stored and held past respectively.

In the present embodiment, the coefficient updating unit 12 is provided with a storage means (not shown), and the sum signal r ⁺ and the difference signal r ⁻ are stored and held for the past N pieces, respectively, and the oldest signal of each piece. Is rewritten with the latest sum signal r ⁺ and difference signal r ⁻ .

In order to suppress the gain increase in the specific frequency band of the first transversal filter 6, the second reference signal r ₂ is required to be a signal having only the frequency component to be controlled. To be done. In the above configuration, it is the output control signal y of the first transversal filter 6, since the filtering by filter 9 passes only the frequency band is a suppression target, the second to the reference signal r ₂ is , And only the frequency components to be suppressed in the control signal y are included. Therefore, the first transversal filter 6 is appropriately suppressed in gain increase.

In updating the filter coefficient a _i according to the present embodiment, the update amount of each coefficient is δ ⁺ _{i in} equation (11) or equation (12).
Δ ^- _i . That is, the update amount of each coefficient a _i in one sampling is δ ⁺ ₀ (n), δ ⁺
₁ (n) ・ ・ ・ δ ⁺ _N-2 (n), δ ⁺ _N-1 (n), and according to equation (6) δ ^- ₀ (n), δ ^- ₁ (n) ・ ・ ・ δ _{^{- n-2 (n),}} δ - the _n-1 (n).
If μ = ν in the prior art equations (2) and (4) and the update amount is δ _i when both equations are used at the same time, the update amount δ ⁺ _i or δ ⁻ _{i in} the present embodiment is given by the equation ( As is clear from 11) and equation (12), μ (r ₁ (ni) e ₂ (n) + r ₂ (ni) e ₁ (n)) rather than δ _i
Will only increase or decrease.

However, in this embodiment, the equation (5) and the equation
Since the update of (6) is performed alternately every sampling, these quantities cancel each other out in the adaptation process. Therefore, it is possible to obtain the convergence value of the filter coefficient that is almost equivalent to that in the case of the conventional technique.

Furthermore, since the update amounts for minimizing the first error signal are included in the update equations of equations (5) and (6), equations (5) and ( Even if 6) is executed alternately,
The speed of obtaining the muffling effect is hardly deteriorated, and the number of times of updating the filter coefficient required to obtain the same muffling effect is reduced to 1/2 of that of the conventional technique. Therefore, the desired silencing effect can be achieved in about half the time of the prior art.

The series of filter operations and filter coefficient update operations described above are processed by programming of a microprocessor or a digital signal processor, or by filter operation dedicated hardware.

In the second embodiment of the present invention, the filter coefficient sequence a _i (i = 0, i = 0, in accordance with the equations (5) and (6) as in the first embodiment.
1,2, ..., N-1) is updated, but the way of using equations (5) and (6) is different. In this embodiment, the filter coefficient sequence a _i is updated by alternately using the equations (5) and (6) depending on whether n−i is an even number or an odd number. Equation (5) may be used when ni is an even number, and equation (6) may be used when it is an odd number. Conversely, equation (5) may be used when n-i is an odd number, and equation (5) may be used when n-i is an odd number. Equation (6) may be used for

In sampling at a certain time, the filter coefficients of even-numbered taps are all expressed by equation (5) or equation (6).
And the filter coefficients of the odd-numbered taps are all updated by the other equation. In the next sampling, even and odd numbers of n are reversed, so that n−i
The filter coefficient of the even-numbered taps and the filter coefficient of the odd-numbered taps are
It is updated by the other formula in (6) that was not used previously.

[0047] That is, the update amount of each coefficient a _i in one sampling, for _{^{example, δ + 0 (n),}} δ - 1 (n) ··· δ +
_N-2 (n), δ ^- _N-1 (n), and in the next sampling δ ^- ₀ (n), δ ⁺ ₁ (n) ・ ・ ・ δ ^- _N-2 (n), δ ⁺ _N It becomes _-1 (n).
Similar to the first embodiment, the update amounts δ ⁺ _i and δ ^- _i of this embodiment are the same.
Also, the update amount δ _{i of the} conventional technique is μ (r ₁ (ni) e ₂ (n) + r ₂ (n
-i) e ₁ (n)) only differ for each coefficient a _i, for each sampling [delta] ⁺ _i and [delta] ^- Since _i is replaced, these quantities are offset from one another in the adaptation process. Therefore, it is possible to obtain the convergence value of the filter coefficient that is almost equivalent to that in the case of the conventional technique.

Furthermore, since the update amounts for minimizing the first error signal are included in the update equations of the equations (5) and (6), the update by both equations is performed alternately. Even if it is executed, the speed at which the muffling effect is obtained hardly deteriorates. Therefore,
The number of times of updating the filter coefficient required to obtain the same muffling effect is reduced to half that of the conventional technique, and the desired muffling effect can be achieved in about half the time of the conventional technique.

[0049] In this embodiment, the updating of the filter coefficient sequence a _i, for ^{example, r + (n-N +} 1), r - (n-N + 2) ··· r + (n-1),
Like r ⁻ (n), N signals in which the sum signal r ⁺ of the reference signals and the difference signal r ⁻ are alternately arranged are required. But,
It is not necessary to store and hold N sum signals r ⁺ and difference signals r ⁻ respectively as in the first embodiment, and the signals r ⁺ and r ⁻ are alternately input to the coefficient updating unit 12 for each sampling. Good. Therefore, the required capacity of the storage unit in the coefficient updating unit 12 is halved compared to the first embodiment.

As described above, in this embodiment, it is possible to obtain almost the same silencing effect and its achievement speed as in the first embodiment, and the physical storage capacity is about half, which is more efficient as an active noise control device. Good.

FIG. 2 is a block diagram showing a case where the active noise control system according to the third embodiment of the present invention is applied to muffling a duct. The difference from FIG. 1 is that the input to the filter 9 that passes only the frequency band in which the gain increase of the first transversal filter 6 is suppressed is changed to the noise signal u (n). Other configurations are as described in the first and second embodiments.

According to this configuration, when a frequency component to be suppressed appears in the noise signal u (n), the second reference signal r ₂ contains the frequency component to be suppressed, and the frequency component for the frequency is suppressed. The inhibitory action works. Therefore,
The gain increase of the first transversal filter 6 is appropriately suppressed.

FIG. 3 shows the configuration in the case where the active noise control system according to the fourth embodiment of the present invention is applied to the muffling of a duct. The difference from FIG. 1 is that a white noise generator 17 is newly added, and its output w (n) is used as an input to the filter 9. Others are the first and the second.
The configuration is the same as that of the embodiment.

Generating an M-sequence signal as white noise is easy in programming and is effective in generating noise including a frequency component to be suppressed. According to this configuration, the effect of always suppressing the gain increase of the frequency component to be suppressed works, and appropriate suppression is performed.

FIG. 4 is a block diagram showing a case where the active noise control system according to the fifth embodiment of the present invention is applied to muffling a duct. Here, the repetitive signal s (n) of the filter coefficient sequence a _i of the first transversal filter 6 is used as the input signal to the filter 9. The signal s (n) is calculated by, for example, the equation (13)
Can be defined as s (n) = a _i (n) (i = n mod N) (13) where (n mod N) represents the remainder when n is divided by N.

As described above, when the repetitive signal of the filter coefficient sequence a _i is used instead of the control signal y, if the gain of the frequency component to be suppressed increases in the first transversal filter 6, The action that tries to suppress is working. As a result, the increase in the gain of the first transversal filter 6 is appropriately suppressed. Other configurations are similar to those of the first and second embodiments.

FIG. 5 shows the configuration in which the active noise control system of the sixth embodiment of the present invention is applied to muffling a duct. In the present embodiment, in addition to the configuration of FIG. 1, a power detection unit 18 for detecting the average of the square of the second error signal e ₂ is newly provided, and the power signal P (which is the output from the power detection unit 18 The update formula of the filter coefficient used in the coefficient updating unit 12 is changed according to n).

In the coefficient updating unit 12, the power signal P (n)
Is larger than the preset reference value, the filter coefficient a _i is updated by alternately using the equations (5) and (6) as in the first or second embodiment. . Power signal P
When (n) is smaller than the reference value, the filter coefficient a _i is updated using the equation (2).

According to this configuration, when the power signal P (n) is smaller than the reference value, the filter coefficient a _i is updated by the equation (2).
Is used, the influence of the second term on the right side in Expressions (11) and (12) is reduced. At the start of control, the signal in the frequency band to be suppressed is not included in the control signal y so much and the contribution to the _second error signal e ₂ is small. Therefore, the filter coefficient a _i is updated mainly by the equation (2), and the desired silencing effect is quickly achieved. After that, if the gain of the frequency component to be suppressed increases, the contribution to the second error signal e ₂ also increases, and the updating of the filter coefficient a _i by the equations (5) and (6) functions.

In order to update the filter coefficient a _i using the equation (2), the first reference signal r
_{1 is} also input to the coefficient updating unit 12, and the past N signals are stored and held. Other configurations are similar to those of the first embodiment.

Also in the third, fourth and fifth embodiments,
In this way, the power detection unit 18 is provided, and the coefficient update unit 12
The updating formula of the filter coefficient in (4) may be changed according to the power signal P (n).

FIG. 6 is a block diagram showing the case where the active noise control system according to the seventh embodiment of the present invention is applied to the duct silencing. In the present embodiment, a power detection unit 18 for detecting the average of the square of the second error signal e ₂ is added to the system configuration of FIG. 7 showing the prior art, and other configurations have already been described with reference to FIG. 7. That's right.

The updating operation of the filter coefficient in the first coefficient updating unit 8 and the second coefficient updating unit 11 is controlled according to the power signal P (n) which is the output from the power detecting unit 18. When the power signal P (n) is larger than a preset reference value, the filter coefficient a _i is updated by the first coefficient updating unit 8 using the equation (2) and the equation (4) is used. Had second
The updating of the filter coefficient a _i by the coefficient updating unit 11 of is alternately executed for each sampling. On the other hand, the power signal P
If (n) is smaller than the reference value, the second coefficient updating unit 11 does not update the filter coefficient, and only the first coefficient updating unit 8 always updates the filter coefficient a _i .

According to this structure, since the filter coefficient a _i is updated mainly by the equation (2) immediately after the start of the control operation, it is possible to prevent the deterioration of the speed of obtaining the silencing effect after the start of the control.

[0065]

According to the present invention, in an active noise control device for suppressing a primary sound wave emitted from a noise source by emitting a secondary sound wave and interfering with a sound wave, a noise signal correlated with the primary sound wave. The noise signal creating means for outputting u and the first transversal filter having the predetermined filter coefficient sequence a _i filter the noise signal u to control the signal y.
, A secondary sound wave output means for inputting the control signal y to generate a secondary sound wave, and a first error signal e ₁ having a correlation with the interference result of the primary sound wave and the secondary sound wave. A first reference signal r obtained by filtering the noise signal u with a transfer function H (z) in the process of detecting the control signal y by the error signal creating means in the adapting means. ₁ and the second reference signal r ₂ having a predetermined frequency component, the sum signal r ⁺ and the difference signal r ⁻ , the first error signal e _1, and the second reference signal r ₂ The sum signal e ⁺ and the difference signal e ⁻ of the second error signal e ₂ filtered by the second transversal filter having the same filter coefficient sequence a _i as the transversal filter are created, and these four signals r ⁺ , r ^- use the ^-, e ^+, e By updating the filter coefficient sequence a _i you are, the first reference signal r ₁ involved in silencing target becomes possible to constantly reflect the updating of the filter coefficient sequence a _i, computation efficiency is improved. Therefore, the amount of calculation can be reduced.

Further, depending on whether ni is an even number or an odd number, if the filter coefficient is updated by alternately using the equation (5) and the equation (6), the speed of obtaining the muffling effect is not deteriorated, The number of times the filter coefficient is updated can be reduced to half that of the conventional technique. Therefore, it is possible to select a cheaper hardware, which has a cost reduction effect.

Further, based on the even and odd numbers of ni, the equation (5)
By alternately using the expression (6) and the expression (6), the number of sum signals r ⁺ and difference signals r ⁻ of reference signals which need to be stored and held can be suppressed to the necessary minimum number equal to the number of filter coefficients. As a result, the physical storage capacity can be kept small, which is effective for downsizing hardware.

[0068] provided with level detection means for detecting a second amplitude level of the error signal e _2, when the second error signal e ₂ level is below a predetermined standard, the filter coefficients a _i by the formula (2) In the configuration in which the update is performed, the update formula of the formula (2) is mainly applied immediately after the control is started, so that there is an effect of preventing the deterioration of the speed of obtaining the muffling effect after the start of the control.

Even when the filter coefficient a _i is updated by alternately using the equations (2) and (4) for each sampling as in the conventional case, the amplitude level of the _second error signal e ₂ is set to a predetermined level. If you use only formula (2) when it is below the standard,
Since the formula (2) is mainly applied immediately after the control is started, the desired silencing effect can be achieved more quickly.

[Brief description of drawings]

FIG. 1 is a diagram schematically showing a configuration of an active noise control device of a first embodiment and a second embodiment of the present invention.

FIG. 2 is a diagram schematically showing a configuration of an active noise control device of a third embodiment of the present invention.

FIG. 3 is a diagram schematically showing a configuration of an active noise control device of a fourth embodiment of the present invention.

FIG. 4 is a diagram schematically showing a configuration of an active noise control system of a fifth embodiment of the present invention.

FIG. 5 is a diagram schematically showing the configuration of an active noise control system of a sixth embodiment of the present invention.

FIG. 6 is a diagram schematically showing a configuration of an active noise control system of a seventh embodiment of the present invention.

FIG. 7 is a diagram schematically showing a configuration of an active noise control device according to a conventional technique.

[Explanation of symbols]

1 Noise Source 2 Duct 3 Microphone 4 Speaker 5 Microphone 6 Transversal Filter 7 Filter 8 Coefficient Update Unit 9 Filter 10 Transversal Filter 11 Coefficient Update Unit 12 Coefficient Update Unit 13 Adder 14 Subtractor 15 Adder 16 Subtractor 17 White Noise Generation unit 18 Power detection unit u Noise signal y Control signal a _i Filter coefficient sequence r ₁ Reference signal r ₂ Reference signal e ₁ Error signal e ₂ Error signal r ⁺ Reference sum signal r ⁻ Reference difference signal e ⁺ Error sum signal e ⁻ Error difference signal H (z) Transfer function w White noise P Power signal s Repeated signal

─────────────────────────────────────────────────── ─── Continuation of the front page (58) Fields surveyed (Int.Cl. ⁷ , DB name) G10K 11/178 F01N 1/06 H03H 21/00

Claims (3)

(57) [Claims] 1. A primary sound wave emitted from a noise source,
In an active noise control device that emits a secondary sound wave and suppresses sound by sound wave interference, a first transformer having a noise signal creating means for outputting a noise signal u correlated with the primary sound wave, and a predetermined filter coefficient sequence a _i. A signal processing unit that filters the noise signal u with a Versal filter to output a control signal y; a secondary sound wave output unit that inputs the control signal y to generate a secondary sound wave; a primary sound wave and a secondary sound wave. Error signal generating means for outputting a _first error signal e ₁ having a correlation with the interference result, and the noise signal u by the transfer function H (z) in the process in which the control signal y is detected by the error signal generating means. The sum signal r ⁺ and the difference signal r ^{− of} the filtered first reference signal r ₁ and the second reference signal r ₂ having a predetermined frequency component, the first error signal e _1, and the second error signal e ₁ . reference Create and ^- No. r ₂ of the second sum signal e ⁺ and the difference signal e of the error signal e ₂ which is filtered by a second transversal filter having a first transversal filter same filter coefficient sequence and a _i I-th frame at time n
For the filter coefficient a _i (n), μ is a small positive number, and n−i is
Depending on whether it is an even number or an odd number, the following two equations a _i (n + 1) = a _i (n) −μr ⁺ (ni) e ⁺ (n) a _i (n + 1) = a _i (n) ^{-μr - (ni) e - (} n) active noise control apparatus characterized by having an adaptive means for updating using alternating. 2. A level detecting means for detecting the amplitude level of the second error signal e ₂ is provided, and when the level detected by the level detecting means falls below a predetermined reference,
The adapting means uses the i-th filter coefficient a at time n.
_i a (n), and the μ and a small positive number, and updates the following equation _{a i (n + 1) =} a i (n) -μr 1 (ni) e 1 (n), wherein Item 1. The active noise control device according to item 1 . 3. A primary sound wave emitted from a noise source,
In an active noise control device that emits a secondary sound wave and suppresses sound by sound wave interference, a first transformer having a noise signal creating means for outputting a noise signal u correlated with the primary sound wave, and a predetermined filter coefficient sequence a _i. A signal processing unit that filters the noise signal u with a Versal filter to output a control signal y; a secondary sound wave output unit that inputs the control signal y to generate a secondary sound wave; a primary sound wave and a secondary sound wave. Error signal generating means for outputting a _first error signal e ₁ having a correlation with the interference result, and the noise signal u by the transfer function H (z) in the process in which the control signal y is detected by the error signal generating means. the first reference signal r _1, a second reference signal r _2, the first error signal e _1, and the second reference signal r ₂ first transformer bars having a predetermined frequency component filtered Second with the same filter coefficient sequence a _i and Rufiruta
2nd filtered by the transversal filter
Error signal e ₂ is generated and the amplitude level of the _second error signal e ₂ falls below a predetermined reference, the i-th filter coefficient a _i (n) at time n is set to μ as a small positive number. , is updated by the following equation _{a i (n + 1) =} a i (n) -μr 1 (ni) e 1 (n), when the second error signal e ₂ of the amplitude level is equal to or greater than a predetermined reference, For the i-th filter coefficient a _i (n) at time n, let v be a small positive number, and the following two equations a _i (n + 1) = a _i (n) −μr ₁ (ni) e ₁ (n ) a _i (n + 1) = a _i (n) −νr ₂ (ni) e ₂ (n) is alternately used for the time n, and updating means is provided.
An active noise control device comprising: