JPH07253792A - Muffling device - Google Patents

Abstract

PURPOSE:To surely and stably muffle even a noise arriving while propagating through a space having a large reverberation and a duct in the frequency range of the noise to be muffled. CONSTITUTION:This device is provided with a sensor microphone 8 outputting a noise signal, an additional sound source speaker 17 radiating an additional sound, an error microphone 13 outputting a residual noise signal and a signal processing part 12 performing an adaptive control based on the noise signal and the residual noise signal and outputting an additional sound signal Then, the signal processing part 12 has a correcting filter 20 and a feedback correcting filter 23 for stabilizing the adaptive control, a white noise generating part generating a white noise and an LPF relatively attenuating the frequency component of the white noise data corresponding to a frequency component among frequency components disturbing a muffling operation and performs an identifying procession correcting filter coefficients of the correcting filter 20 and the feedback correcting filter 23 by using the output data of the LPF.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a muffler, and more particularly to a muffler suitable for being installed in a duct or a space having a large reverberation.

[0002]

2. Description of the Related Art Conventionally, as this type of silencer, there has been known a silencer that actively reduces noise in the passenger compartment of an automobile or the passenger compartment of an aircraft, as described in Japanese Patent Laid-Open No. 16889/1993. Has been. This silencer includes a noise detection means such as a crank angle sensor for detecting a noise generation state of an engine, and a loudspeaker (additional sound source speaker) for generating an additional sound having the same amplitude as the noise of the noise generation source (source sound). ) And a microphone that detects residual noise in the passenger compartment (error microphone)
And a transfer function between the additional sound source speaker and the error microphone (to be exact, the voltage-sound-pressure conversion characteristic of the additional sound source speaker and the sound of the error microphone) based on the output signal of the error microphone and the output signal of the noise detecting means. A controller (signal processing unit) for generating an additional sound signal for driving an additional sound source speaker by using a control algorithm including a pressure-voltage conversion characteristic)
And an ignition switch for notifying when the engine is started, and a test signal generator for generating a test signal when the engine is started.

The characteristic of this silencer is that the temperature of the passenger compartment changes.
In order to update the above transfer function, which changes due to the influence of deterioration of the additional sound source speaker and the error microphone over time, to the corrected transfer function, the test signal generator supplies the test signal to the controller when the engine is started. is there.

[0004]

By the way, in the case of silencing noise propagating in a predetermined space or duct by using the silencing device of the above-mentioned conventional structure, there are the following inconveniences. That is, some of the noises have frequency bands that are easily reverberated due to the structure of the space and the duct. When an additional sound source having the same amplitude as the noise in the frequency band and having the opposite phase is radiated from the additional sound source speaker, the additional sound reverberates in the space or duct and interferes with the reverberant sound. Noise may be unstable and uncontrollable, and the sound pressure level may rise rather than before the silencing operation is started.

The above problem will be described in more detail with reference to FIGS. 6 and 7. FIG. 6 shows the ventilation duct 1.
It is a block diagram of the example of composition of the conventional silencer which silences the noise which propagates inside. The ventilation duct 1 shown in FIG. 6 has a substantially T-shaped cross section, and a noise detecting means 2 (a sensor microphone in this example) is attached to the inside of the beam portion 1a at a predetermined distance from the opening portion 1b. Has been. Sensor microphone 2
Detects the level of noise invading through the opening 1b of the ventilation duct 1 and converts it into a noise signal. This noise signal is input to the signal processing unit 3. The configuration of the signal processing unit 3 is similar to that of the signal processing unit described in Japanese Patent Laid-Open No. 16889/1993.

On the other hand, the opening 1 of the beam 1a of the ventilation duct 1
An error microphone 4 that detects the level of residual noise attenuated by this silencer and converts it into a residual noise signal is provided near the inside of c. The residual noise signal is input to the signal processing unit 3. In addition, the additional sound source speaker 5 is provided on the upstream side of the error microphone 4 and on the bottom portion 1e of the base portion 1d of the ventilation duct 1, and emits an additional sound that should cancel noise in the opening 1c of the ventilation duct 1. Further, in the vicinity of the outside of the opening 1c of the ventilation duct 1, a measurement microphone 6 for detecting noise radiated from the opening 1c is provided. In addition, the sensor microphone 2,
The error microphone 4, the additional sound source speaker 5, and the measurement microphone 6 are provided at positions (unit: mm) having the dimensions shown in FIG. The diameter of the beam portion 1a and the base portion 1d of the ventilation duct 1 is 100 mm.

Next, FIG. 7 shows a result of detecting the noise radiated from the opening 1c of the ventilation duct 1 by the measuring microphone 6 before and after the start of the silencing operation by the silencing device of FIG. In FIG. 7, the horizontal axis is the noise frequency [Hz], the vertical axis is the noise sound pressure relative level [dB], the broken line shows the measurement result before the silencing operation is started, and the solid line is the silencing operation start. The subsequent measurement results are shown. As can be seen from this figure, it is 1 after the muffling operation is started except near 600 Hz.
Although the noise is reduced by 0 to 30 dB, in the vicinity of 600 Hz, the sound pressure is increased by 10 dB or more after the silencing operation is started. This means that in the ventilation duct 1 having the structure shown in FIG. 6, noise near 600 Hz is likely to be reflected. As described above, when the conventional silencer described above is used to muffle noise in a ventilation duct or space having a structure in which sound of a predetermined frequency easily echoes, the noise of that frequency rather increases. There was an inconvenience that it would end up.

The present invention has been made in view of the above-mentioned circumstances, and even if noise enters through a space or a duct having a large reverberation within a frequency range of noise to be silenced,
It is an object of the present invention to provide a muffling device that can reliably muffle sound.

[0009]

In order to solve the above-mentioned problems, the invention according to claim 1 is based on a noise detecting means for detecting an incoming noise and outputting a noise signal, and an additional sound signal inputted. An additional sound emitting means for emitting an additional sound to cancel the noise at or near a predetermined listening position of the noise, and a residual noise signal provided at or near the listening position to detect a residual noise signal. Based on the residual noise detection means for outputting, the noise signal and the residual noise signal,
Signal processing means for performing adaptive control to constantly minimize the residual noise at or near a predetermined listening position of the noise, generating an additional sound signal for emitting the additional sound, and supplying the additional sound signal to the additional sound emitting means. The signal processing means includes a correction filter for stabilizing the adaptive control, and white noise data generation means for generating white noise data limited to the frequency band of the noise. In the muffling device that performs identification processing of the filter coefficient of the correction filter using the white noise data before starting, the signal processing means corresponds to a frequency component that interferes with the silencing operation among frequency components of the noise, A band limiting unit that relatively attenuates a frequency component of the white noise data is provided, and the output data of the band limiting unit is used to output the correction filter of the correction filter. It is characterized by performing the identification processing for correcting the filter coefficients.

According to a second aspect of the invention, there is provided the silencer according to the first aspect, wherein the signal processing means has a frequency characteristic of a noise signal output from the residual noise detecting means immediately before the start of the muffling operation. And a frequency characteristic calculating means for calculating the residual noise frequency characteristic which is the frequency characteristic of the residual noise signal output from the residual noise detecting means after a predetermined time has elapsed after the silencing operation is started. Among the frequency components of the residual noise, it is determined whether or not there is a frequency component in which the difference between the residual noise frequency characteristic and the corresponding noise frequency characteristic immediately before noise suppression exceeds a preset threshold value. Has a determination means for instructing the band limiting means to relatively attenuate the frequency component of the white noise data corresponding to the frequency component. It is.

[0011]

According to the first aspect of the invention, the signal processing means identifies the filter coefficient of the correction filter using the white noise data before starting the muffling operation. Next, the noise detection means detects the incoming noise and outputs a noise signal. Further, the residual noise detecting means detects the interference noise due to the interference of the incoming noise and the additional sound emitted from the additional sound emitting means at or near the listening position, and outputs the residual noise signal. As a result, the control means performs adaptive control based on the noise signal and the residual noise signal, generates an additional sound signal for emitting an additional sound having a phase opposite to that of the noise waveform component, and supplies the additional sound signal to the additional sound emitting means. The additional sound emitting means emits an additional sound at or near the listening position based on the input additional sound signal. When the noise canceling operation is performed, if there is a noise frequency component that interferes with the noise canceling operation, the signal processing means limits the frequency component of the white noise data, which corresponds to the noise frequency component, to relatively attenuate the frequency component. Identification processing is performed to correct the filter coefficient of the correction filter using the output data of the means. Therefore, according to the configuration of claim 1, even when noise in a space or duct having a large reverberation is silenced, it is possible to perform an accurate and stable silencing operation in the frequency range of the noise to be silenced.

According to the second aspect of the present invention, the frequency characteristic calculation means calculates the noise frequency characteristic immediately before the muffling operation immediately before the muffling operation is started, and calculates the residual noise frequency characteristic after a lapse of a predetermined time after the muffling operation is started. calculate. Further, the determining means determines whether or not, among the frequency components of the residual noise, there is a frequency component in which the difference between the residual noise frequency characteristic and the corresponding noise frequency characteristic immediately before noise suppression exceeds a preset threshold value. In some cases, the band limiting means is instructed to relatively attenuate the frequency component of the white noise data corresponding to the frequency component.

[0013]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing an electrical configuration of a silencer (which is applied to a ventilation duct 7 of a house) which is an embodiment of the present invention. In this embodiment, the frequency of noise to be silenced is 1 kHz or less. In FIG. 1, the sensor microphone 8 is attached inside the ventilation duct 7 in the vicinity of the opening 7 a and detects the level of noise propagating in the ventilation duct 8 and converts it into a noise signal.
This noise signal is input to an amplifier (hereinafter referred to as AMP) 9. The AMP 9 amplifies the noise signal. LPF1
0 has a cutoff frequency of 1 kHz, and passes the low frequency component of the output signal of the AMP9. The analog / digital converter (hereinafter referred to as A / D converter) 11 converts the low frequency component of the noise signal output from the LPF 10 at a sampling frequency of 2.5 kHz into digital noise data, and then inputs the digital noise data to the signal processing unit 12. To do.

On the other hand, in the vicinity of the opening 7b inside the ventilation duct 7, there is provided an error microphone 13 for detecting the level of residual noise attenuated by this silencer and converting it into a residual noise signal. The residual noise signal is AM
Input to P14. The AMP 14 amplifies the residual noise signal. The LPF 15 has a cutoff frequency of 1 kHz, and passes the low frequency component of the output signal of the AMP 14. The A / D converter 15 has a sampling frequency of 2.
The low-frequency component of the residual noise signal output from the LPF 15 at 5 kHz is converted into digital residual noise data ε (i) and then input to the signal processing unit 12. Further, the additional sound source speaker 17 is provided in the ventilation duct 7 on the upstream side of the error microphone 13.
It is embedded in the inner surface of the and emits an additional sound to cancel the noise at the opening 7b of the ventilation duct 7.

The signal processing unit 12 includes an adaptive filter (hereinafter,
An ADF) 18, a controlled filter (hereinafter referred to as CNF) 19, a correction filter 20 and the like are roughly configured.
7 to the error microphone 13, the transfer function G11 between the sensor microphone 8 and the error microphone 13 has an inverse characteristic F (correctly, in advance, of the transfer function G11) as an identification and adaptive control operation. The inverse characteristic F) from which the effect has been removed is successively identified, and digital additional sound data for canceling noise is generated and output. This effect is added by returning the additional sound from the additional sound source speaker 17 (returning to the original).

The noise data output from the A / D converter 11 is input to the non-inverting input terminal of the differential amplifier 21 of the signal processing unit 12. The output data of the differential amplifier 21 is
It is input to the CNF 19 and the correction filter 20. On the other hand, the residual noise data ε (i) output from the A / D converter 16 is input to the ADF 18 and the frequency characteristic calculation unit 22.

The ADF 18 outputs the noise data XC output from the correction filter 20 and corrected by the feedback correction filter 23 and the correction filter 20.
Based on the value of (i), the residual noise data ε (i) is used to make full use of an adaptive control algorithm such as Least Mean Squre (LMS), and the position of the error microphone 13 is constantly measured. So that the residual noise at
Adaptation processing (adaptive operation) of a required transfer function (filter coefficient) is performed. A filter coefficient j corresponding to the inverse characteristic F of the transfer function G2 between the sensor microphone 8 and the error microphone 13 is set in the CNF 19. This filter coefficient j
Are updated by the ADF 18 every sampling period.

The frequency characteristic calculator 22 calculates the frequency characteristic of noise at the detection point of the error microphone 13 immediately before the start of the silencing operation and the frequency characteristic of residual noise at the detection point of the error microphone 13 at the silencing operation, and the storage device 24 Are stored as noise frequency characteristic data immediately before muffling and residual noise frequency characteristic data. In this embodiment, the frequency characteristics of noise and residual noise of 1 kHz or less are obtained by subjecting 512 points of sampling data to Fourier transform, and equally dividing the result into 16 blocks on the frequency axis.
The average of each divided block is calculated, and this is averaged 50 times in time.

The D / A converter 25 converts white noise data output from a white noise generator (not shown) at a sampling frequency of 2.5 kHz, output data of the CNF 19 and the like into analog signals. Further, the LPF 26 has a cutoff frequency of 1 kHz, is configured by an anti-aliasing filter, and passes the low frequency component of the output signal of the D / A converter 25. AMP27 is LP
After amplifying the output signal of F26, the additional sound source speaker 17
To enter.

The correction filter 20 is composed of an FIR digital filter and contributes to the stabilization of the adaptive control algorithm, and the additional sound source speaker 17 is provided.
1 to the error microphone 13, the path shown in FIG. 1 (point J → D / A converter 25 → LPF 26)
→ AMP27 → additional sound source speaker 17 → error microphone 1
3 → AMP14 → LPF15 → A / D converter 16 →
The filter coefficient h1 corresponding to the transfer function G11 of 1 kHz or less at the point K) is set. This correction filter 2
An adaptive control algorithm including 0 is called a Filtered-XLMS algorithm and was proposed by Widrow et al. In 1985.

The feedback correction filter 23 has an F
The additional sound radiated from the additional sound source speaker 17 is for preventing the additional sound radiated from the additional sound source speaker 17 from being converted into an electric signal by the sensor microphone 8 to cause acoustic feedback. Up to 8, to be precise, the route shown in FIG. 1 (point J → D / A converter 25 → LPF26 → AMP27 →
Additional sound source speaker 17 → sensor microphone 8 → AMP 9 → L
PF10 → A / D converter 11 → Point L) 1k
A filter coefficient h2 corresponding to a transfer function G12 below Hz is set.

The additional sound data output from the CNF 19 is input to the inverting input terminal of the differential amplifier 21 through the feedback correction filter 23 and subtracted from the noise data in the differential amplifier 21, whereby feedback correction is performed. Done. The algorithm including this feedback correction was proposed by Warnaka et al. In 1984.

In the above structure, first, before the muffling operation is performed, the filter coefficients h1 and h2 of the correction filter 20 and the feedback correction filter 23 are initially set.
Need to ask. This initial setting operation will be described with reference to FIG. 2, parts corresponding to the respective parts in FIG. 1 are designated by the same reference numerals, and the description thereof will be omitted. FIG. 2 shows a white noise generator 28 not shown in FIG.
, LPF 29, and subtractors 30a and 30b are shown. In FIG. 2, the ADF 18, 18 and the CNF are shown.
Although two subscripts a and b are attached to 19 and 19, respectively, these are merely shown for convenience of explanation of the initial setting operation.

The LPF 29 is a digital filter that outputs the output data of the white noise generator 28 with its band limited to 1 kHz or less. In this embodiment, as described above, the frequency characteristics of noise and residual noise of 1 kHz are equally divided into 16 blocks on the frequency axis. Therefore, the LPF 29 correspondingly corresponds to this block. 16 narrow-band band-pass filters (hereinafter, referred to as “B”) that pass only the output data of the white noise generator 28 in the frequency band.
It is called PF). And LPF29
When each of the BPFs is turned on by the determination unit 31 described later, the output data of the white noise generation unit 28 in the corresponding frequency band is passed. Therefore, if all the BPFs are turned on, the output data of the white noise generator 28 will be 1 kHz.
The output signal is output from the LPF 29 after being band-limited to z or less.

First, the white noise generator 28 produces M series data. Next, the produced M series data is 16
In LPF 29 in which all BPFs are turned on, 1
White noise data X (i) band-limited to below kHz
(I is time), and ADF 18a, 18b and CNF
It is input to the D / A converter 25 as well as being input to 19a and 19b. As a result, the D / A converter 25
The white noise data X (i) input to the D / A converter 25 is converted into an analog signal, and then the LPF is input to the LPF.
The signal is input to the AMP 27 via 26, amplified, and input to the additional sound source speaker 17. Therefore, 1k corresponding to the white noise data X (i) from the additional sound source speaker 17
Since white noise (sound waves) whose band is limited to Hz or less is emitted and detected by the error microphone 13, the level of the detected white noise is converted into a noise signal and output in the error microphone 13.

Next, the noise signal corresponding to the white noise data X (i) output from the error microphone 8 is AMP14.
After being amplified in (1), it is converted into digital noise data in the A / D converter 16 via the LPF 15, and is input to the non-inverting input terminal of the subtractor 30a (not shown in FIG. 1). At the time of initial setting, the error microphone 1
Since the output value from 3 is not residual noise, it is simply referred to as noise data. The same applies below.

On the other hand, the white noise data X (i) input to the ADF 18a and the CNF 19a is subjected to convolution operation with the correction filter coefficient h1 (initial value is random or 0) obtained by the ADF 18a, and then The obtained calculation result Yc (i) is output from the CNF 19a and input to the inverting input terminal of the subtractor 30a. In the subtractor 30a, the result Yc (i) of the convolution operation processed by the CNF 19a is subtracted from the noise data output from the A / D converter 16, and the obtained subtraction result is output as error data εc (i). , Is input to the ADF 18a.

As a result, in the ADF 18a, the LMS algorithm processing of about 5 seconds shown in the equation (1) is executed based on the error data εc (i) and the white noise data X (i), and the additional sound source speaker 17 Error microphone 1
The transfer function G11 up to 3 is obtained, and the coefficients of the ADF 18a are all updated to the filter coefficient h1 corresponding to the obtained transfer function G11. Then, this filter coefficient h1
Is set in the correction filter 20 as the filter coefficient h1 during the actual muffling operation. h1 (i + 1) (n) = h1 (i) (n) -2μX (i−n) εc (i) ... (1) In the equation (1), i is time, n is the tap number of the ADF 18a, μ is a convergence coefficient.

On the other hand, the white noise (sound wave) radiated from the additional sound source speaker 17 and corresponding to the white noise data X (i) is also detected by the sensor microphone 8. Therefore, the white noise detected by the sensor microphone 8 is detected. The noise level is converted into a noise signal and output. Next, the noise signal corresponding to the white noise data X (i) output from the sensor microphone 8 is amplified by the AMP 9, and then the LPF 10
After that, it is converted into digital noise data in the A / D converter 11, and is input to the non-inverting input terminal of the subtractor 30b not shown in FIG.

On the other hand, the white noise data X (i) input to the ADF 18b and the CNF 19b is the filter coefficient h2 for feedback correction obtained by the ADF 18b.
After the convolution operation is performed (the initial value is random or 0), the obtained operation result Yh (i) is output from the CNF 19b and input to the inverting input terminal of the subtractor 30b.
In the subtractor 30b, the result Yh (i) of the convolution operation processed by the CNF 19b is subtracted from the noise data output from the A / D converter 11, and the obtained subtraction result is output as error data εh (i). , ADF18
Input to b.

As a result, in the ADF 18b, the LMS algorithm processing of about 5 seconds shown in the equation (2) is executed based on the error data εh (i) and the white noise data X (i), and the additional sound source speaker 17 Sensor microphone 8
To the transfer function G12, and the coefficient of the ADF 18b is the filter coefficient h corresponding to the obtained transfer function G12.
All updated to 2. Then, this filter coefficient h2 is
It is set in the feedback correction filter 23 as a filter coefficient h2 for feedback correction during actual muffling operation. h2 (i + 1) (n) = h2 (i) (n) -2μX (i−n) εh (i) ... (2) In equation (2), i is time, n is the tap number of the ADF 18b, μ is a convergence coefficient. The processing described above is performed during one sampling cycle. By repeating this processing for about 10 to 20 seconds, the filter coefficient h1 of the correction filter 20 and the feedback correction filter coefficient h2 of the feedback correction filter 23 are obtained.

When the above-described initial setting is completed, the noise frequency characteristic calculation operation immediately before muffling at the detection point of the error microphone 13 immediately before the start of the muffling operation is executed. Figure 1
In, the error microphone 13 detects the noise level in the vicinity thereof, converts it into a noise signal, and outputs it. The noise signal output from the error microphone 13 is amplified by the AMP 14, and then passes through the LPF 15 and the A / D converter 16
Is converted into digital noise data and input to the frequency characteristic calculation unit 22. The frequency characteristic calculation unit 22 sets A
/ D converter 16 samples the noise signal 51
After Fourier transforming the sampling data of two points, the result is divided equally into 16 blocks on the frequency axis,
The average of each divided block is calculated. By averaging this for 50 times in time, the frequency characteristic calculation unit 22 calculates the frequency characteristic of noise at the detection point of the error microphone 13 immediately before the start of the silencing operation, and the storage device 24 stores each of the 16 blocks. Each time, it is stored as noise frequency characteristic data immediately before muffling.

Next, the muffling operation is executed. In FIG. 1, the level of noise that has entered the ventilation duct 7 through the opening 7a of the ventilation duct 7 is detected by the sensor microphone 8, converted into a noise signal, amplified by the AMP 9, and passed through the LPF 10 to A / It is converted into digital noise data in the D converter 11. And A / D
The noise data output from the converter 11 is input to the non-inverting input terminal of the differential amplifier 21 of the signal processing unit 12. On the other hand, the additional sound data subjected to the convolution operation with the filter coefficient j (initial value is 0) in the CNF 19 is subjected to the convolution operation with the filter coefficient h2 in the feedback correction filter 23, and then added to the differential amplifier 21. It is input to the inverting input terminal. As a result, the differential amplifier 2
1, the output data of the feedback correction filter 23 is subtracted from the noise data to perform the feedback correction.

Next, the output data of the differential amplifier 21 is C
It is input to the NF 19 and the correction filter 20. As a result, in the CNF 19, the convolution operation is performed with the filter coefficient j to generate additional sound data, which is then converted into an analog signal in the D / A converter 25, and the LPF
The signal is input to the AMP 27 via 26, amplified, and input to the additional sound source speaker 17. However, since the initial value of the filter coefficient j is 0, immediately after the muffling operation is started, no sound is emitted from the additional sound source speaker 17. Also,
In the correction filter 20, convolution calculation is performed on the output data of the differential amplifier 21 with the filter coefficient h1 set at the time of initial setting and the coefficient correction operation described later, and the calculation result XC (i) is stored in the ADF 18. Is entered.

On the other hand, in the error microphone 13, after the level of residual noise is detected and converted into a residual noise signal,
Amplified in AMP14 and passed through LPF10 to D / A
In the converter 11, digital residual noise data ε
Converted to (i). As a result, the ADF 18 uses the residual noise data ε (i) based on the value of the noise data XC (i) output from the correction filter 20 and corrected by the feedback correction filter 23 and the correction filter 20. Then, the adaptation processing (adaptive operation) by the LMS algorithm shown in the equation (3) is performed, so that the residual noise at the position of the error microphone 13 is constantly minimized between the sensor microphone 8 and the error microphone 13. The inverse characteristic F of the transfer function G2 is obtained. And ADF
The 18 coefficients are all updated to the filter coefficient j corresponding to the inverse characteristic F of the obtained transfer function G2. This gives C
All the filter coefficients of NF19 are replaced with this filter coefficient j. j (i + 1) (n) = j (i) (n) -2μXC (i−n) ε (i) ... (3) In the equation (3), i is time, n is the tap number of the ADF 18, μ is a convergence coefficient.

Therefore, in the CNF 19, the newly input noise data is subjected to the convolution operation with the newly set filter coefficient j to generate new additional sound data, and then the D / A converter 25. Is converted into an analog signal, input to the AMP 27 via the LPF 26, amplified, and input to the additional sound source speaker 17. As a result, a new additional sound (silent sound wave) corresponding to the new additional sound data is radiated from the additional sound source speaker 17, and interferes with the noise of the opposite phase at the detection point of the error microphone 13 to weaken the noise. The level of the sound pressure of the residual noise after the interference is detected by the error microphone 13,
After being converted into a residual noise signal, it is amplified in the AMP 14, is converted into digital residual noise data ε (i) in the A / D converter 16 via the LPF 15, and is input again to the ADF 18 of the signal processing unit 12. The adaptive active control is performed by the ADF 18 in one sampling period (A
(1 noise data output from the / D converter 11)
Repeated every time, and finally, at the detection point of the error microphone 13, the noise and the additional sound have the same amplitude and opposite phase, sufficiently cancel each other, and are emitted from the opening 7b of the ventilation duct 7 to the outside. Noise is actively silenced.

Further, after 30 seconds from the start of the muffling operation, the residual noise frequency characteristic calculation operation at the detection point of the error microphone 13 is executed in parallel with the muffling operation. The frequency characteristic calculation unit 22 subjects the 512 points of sampling data obtained by sampling the residual noise signal by the A / D converter 16 to Fourier transform, and then divides the result equally into 16 blocks on the frequency axis. Calculate the average at. By averaging this over 50 times, the frequency characteristic calculation unit 22 calculates the frequency characteristic of the residual noise at the detection point of the error microphone 13 during the silencing operation, and the storage device 24 stores each of the 16 blocks. It is stored as a residual noise frequency characteristic for each.

Next, the judging operation is executed. The determination unit 31 (see FIG. 3), which is not shown in FIGS. 1 and 2, reads the residual noise frequency characteristic and the noise frequency characteristic immediately before silencing for each of the 16 blocks stored in the storage device 24, and for each block. The difference between the residual noise frequency characteristic and the noise frequency characteristic immediately before noise reduction is calculated, and each calculation result is 10 dB.
The block numbers exceeding the number are stored in a predetermined storage area of the storage device 24. The block numbers are assigned from 1 to 16 in order from the low frequency side to the high frequency side of the noise frequency. Then, the calculation result of the difference calculation is 1
Only when there is a block exceeding 0 dB, that is, when there is a frequency band which easily reverberates as described in the section "Problems to be solved by the invention", the filter coefficient h1 of the correction filter 20 and the feedback correction filter 2 shown below.
A filter coefficient correction operation for correcting the filter coefficient h2 of 3 is executed.

Next, the filter coefficient correction operation will be described with reference to FIG.
Will be described with reference to. In FIG. 3, parts corresponding to the parts in FIG. 2 are designated by the same reference numerals, and description thereof will be omitted.
First, in FIG. 3, an LP not shown in FIGS.
Among the 16 BPFs 291-2916 of F29, the BPF corresponding to the block number stored in the predetermined area of the storage device 24 is turned off. Here, in the ventilation duct 7 of FIG. 1, since the sound in the frequency band near 600 Hz is likely to reverberate, in the above determination operation, the residual noise frequency characteristic of the block corresponding to the frequency band near 600 Hz and the noise frequency characteristic immediately before muffling are It is assumed that the calculation result of the difference calculation of 1 exceeds 10 dB. Therefore, in the present case, LPF2
Of the 16 BPFs 291 to 2916 of 9, 600 Hz
The BPF whose pass band is the nearby frequency band is turned off.

Next, referring to FIG. 3, the white noise generator 28
Creates M-series data. Next, the produced M-sequence data has a frequency of 1 kHz or less in the LPF 29 in which only the predetermined BPF is turned off, and the BPF in which the MPF is turned off.
The frequency band of the noise data X is also band-limited and then added in the adder 32 not shown in FIGS.
(I) (see FIG. 4), which is input to the ADFs 18a and 18b and the CNFs 19a and 19b, and also to the D / A converter 25. As a result, the noise data X (i) input to the D / A converter 25 is converted into an analog signal in the D / A converter 25, and then LP
It is input to the AMP 27 via F26, amplified, and input to the additional sound source speaker 17. Therefore, 1 kHz corresponding to the noise data X (i) from the additional sound source speaker 17
Noise (sound waves) whose frequency band is less than z and whose frequency band is near 600 Hz is radiated, and the error microphone 1
The noise level detected by the error microphone 13 is converted into a noise signal and output. Next, the noise signal corresponding to the noise data X (i) output from the error microphone 8 is amplified by the AMP 14, and then passed through the LPF 15 to the A / D converter 1
6 is converted into digital noise data, and the subtractor 3
It is input to the non-inverting input terminal of 0a.

On the other hand, the noise data X (i) input to the ADF 18a and the CNF 19a is subjected to the convolution operation with the filter coefficient h1 obtained by the ADF 18a,
The obtained calculation result Yc (i) is output from the CNF 19a and input to the inverting input terminal of the subtractor 30a. In the subtractor 30a, the result Yc (i) of the convolution operation processed by the CNF 19a is subtracted from the noise data output from the A / D converter 16, and the obtained subtraction result is output as error data εc (i). , Is input to the ADF 18a. As a result, in the ADF 18a, based on the error data εc (i) and the noise data X (i),
The LMS algorithm processing of about 5 seconds shown in the above equation (1) is executed to obtain the corrected transfer function G'11, and the coefficient of the ADF 18a is the filter coefficient h corresponding to the obtained transfer function G'11. All updated to '1. And
This filter coefficient h'1 is set as the corrected filter coefficient h'1 in the correction filter 20.

On the other hand, since the noise (sound wave) corresponding to the noise data X (i) radiated from the additional sound source speaker 17 is also detected by the sensor microphone 8, the level of the noise detected in the sensor microphone 8 is detected. Is converted into a noise signal and output. Next, the noise signal corresponding to the noise data X (i) output from the sensor microphone 8 is AMP9.
After being amplified in, it is converted into digital noise data in the A / D converter 11 via the LPF 10 and input to the non-inverting input terminal of the subtractor 30b.

On the other hand, the noise data X (i) input to the ADF 18b and the CNF 19b are subjected to the convolution operation with the filter coefficient h2 obtained by the ADF 18b,
The obtained calculation result Yh (i) is output from the CNF 19b and input to the inverting input terminal of the subtractor 30b. In the subtractor 30b, the result Yh (i) of the convolution operation processed by the CNF 19b is subtracted from the noise data output from the A / D converter 11, and the obtained subtraction result is output as error data εh (i). , Is input to the ADF 18b. Thereby, in the ADF 18b, based on the error data εh (i) and the noise data X (i),
The LMS algorithm processing of about 5 seconds shown in the above equation (2) is executed to obtain the modified transfer function G'12, and the coefficient of the ADF 18b is the filter coefficient h corresponding to the obtained transfer function G'12. All updated to '2. And
This filter coefficient h'2 is set as the modified filter coefficient h'2 in the feedback correction filter 23. The processing described above is performed during one sampling cycle. By repeating this process for about 10 to 20 seconds, the filter coefficient h1 of the correction filter 20 and the filter coefficient h2 of the feedback correction filter 23 are corrected. After that, the noise frequency characteristic calculation operation immediately before noise reduction, the noise reduction operation, the residual noise frequency characteristic calculation operation, and the determination operation are sequentially executed.

The flow of various operations described above can be summarized as follows. First, when the silencer is installed in the ventilation duct 7 for the first time, an initial setting operation is performed after the power is turned on, and then a noise frequency characteristic calculation operation immediately before silence, a silence operation and a residual noise frequency characteristic calculation operation, The determination operation is sequentially executed. When it is determined in the determination operation that the noise has been sufficiently silenced by the noise reduction operation, the noise reduction operation, the residual noise frequency characteristic calculation operation, and the determination operation are sequentially executed again. On the other hand, when it is determined in the determination operation that the sound pressure level of the noisy frequency component is increasing due to the silencing operation, the filter coefficient correction operation performs the filter coefficient h1 of the correction filter 20 and the feedback correction filter. After the filter coefficient h2 of 23 is corrected, the noise frequency characteristic calculation operation immediately before noise reduction, the noise reduction operation, the residual noise frequency characteristic calculation operation, and the determination operation are sequentially executed again.

Here, FIG. 5 shows an example of the result of an actual muffling experiment performed by the muffling device of the above-described embodiment. The experiment result of FIG. 5 is performed under the same conditions as the experiment in the silencer of FIG. 6 described above. In FIG.
Similar to FIG. 7, the horizontal axis represents the noise frequency [Hz], the vertical axis represents the noise sound pressure relative level [dB], the broken line shows the measurement result before the silencing operation is started, and the solid line is after the silencing operation is started. The measurement result of is shown.

Comparing the sound pressure levels near 600 Hz between FIG. 5 (this embodiment) and FIG. 7 (conventional example), in the conventional case, the sound pressure level after starting the muffling operation is 10 compared to before starting the muffling operation.
In contrast to the increase of more than dB, in this embodiment, after the muffling operation is started, it is 10d compared to before the muffling operation is started.
About B, a decrease of about 20 dB is seen as compared with the conventional device after the muffling is started, and a good muffling effect is obtained over the entire frequency band of noise to be muffled. in this way,
According to the configuration of this example, stable and reliable active silencing can be achieved over the entire frequency range of noise to be silencing.

The embodiment of the present invention has been described in detail above with reference to the drawings. However, the specific structure is not limited to this embodiment, and the design change and the like without departing from the gist of the present invention. Even this is included in this invention. For example, in the above-described embodiment, an example in which the silencer according to the present invention is applied to the ventilation duct 7 having two openings is shown.
The invention is not limited to this, and may be applied to a ventilation duct having a plurality of openings. Further, in the above-mentioned embodiment, 1
Although the example in which one sensor microphone 8 and one error microphone 13 are used has been shown, the present invention is not limited to this, and the present invention is also applied to a muffling device using a plurality of sensor microphones and a plurality of error microphones. Furthermore, in the above-described embodiment, an example in which the signal processing unit 12 is configured by hardware has been shown, but the present invention is not limited to this, and the signal processing unit 12 may be a digital
It may be configured by a signal processor (DSP) and the signal processing such as the adaptive control may be executed by a micro program, that is, software.

[0048]

As described above, according to the silencer having the configuration of the present invention, the signal processing means removes the frequency component of the white noise data corresponding to the frequency component of the noise frequency component that interferes with the silencing operation. Since the identification processing for correcting the filter coefficient of the correction filter is performed by using the band limiting means for relatively attenuating and using the output data of the band limiting means, the propagation arrives via the space or duct having a large echo. Even noise can be surely and stably silenced within the frequency range of noise to be silenced.

[Brief description of drawings]

FIG. 1 is a block diagram showing an electrical configuration of a silencer (applied to an air conditioning duct of a house) that is an embodiment of the present invention.

FIG. 2 is a block diagram for explaining a filter coefficient initial setting operation in the muffling device.

FIG. 3 is a block diagram illustrating a filter coefficient correction operation in the silencer.

FIG. 4 is a diagram showing an example of frequency characteristics of noise used in the same filter coefficient correction operation.

FIG. 5 is a diagram showing an example of an experimental result of a silencing operation in the silencing apparatus.

FIG. 6 is a block diagram showing an electrical configuration of a conventional silencer.

FIG. 7 is a diagram showing an example of an experimental result of a silencing operation in the silencing apparatus.

[Description of Reference Signs] 8 sensor microphone (noise detection means) 12 signal processing unit (signal processing means) 13 error microphone (residual noise detection means) 17 additional sound source speaker (additional sound emission means) 20 correction filter 22 frequency characteristic calculation unit (Frequency characteristic calculating means) 23 Feedback correction filter 28 White noise generating section (white noise generating means) 29 LPF (band limiting means) 31 Determination section (determination means)

Claims (2)

[Claims] 1. A noise detection means for detecting an incoming noise and outputting a noise signal, and an addition for canceling the noise at or near a predetermined listening position of the noise based on the input additional sound signal. Additional sound emitting means for emitting sound,
A residual noise detecting means which is provided at or near the listening position and detects residual noise and outputs a residual noise signal, and a predetermined listening position of the noise or its vicinity based on the noise signal and the residual noise signal And signal processing means for performing adaptive control so as to always minimize the residual noise, generating an additional sound signal for radiating the additional sound, and supplying the additional sound signal to the additional sound radiating means. A correction filter for stabilizing the adaptive control, and a white noise data generating unit for generating white noise data limited to the frequency band of the noise, and using the white noise data before starting the silencing operation. In the silencer for performing the identification process of the filter coefficient of the correction filter, the signal processing means corresponds to a frequency component that interferes with the silencing operation among frequency components of the noise. And a band limiting unit for relatively attenuating the frequency component of the white noise data, and performing an identification process for correcting the filter coefficient of the correction filter using the output data of the band limiting unit. Silencer. 2. The signal processing means calculates a noise frequency characteristic just before noise reduction, which is a frequency characteristic of a noise signal output from the residual noise detection means immediately before the noise reduction operation starts, and a predetermined time after the noise reduction operation starts. A frequency characteristic calculating means for calculating a residual noise frequency characteristic which is a frequency characteristic of the residual noise signal output from the residual noise detecting means after a lapse of time, and a silencing function corresponding to the residual noise frequency characteristic among the frequency components of the residual noise. It is determined whether or not there is a frequency component whose difference from the immediately preceding noise frequency characteristic exceeds a preset threshold value, and if there is, a frequency of the white noise data corresponding to the frequency component in the band limiting means. The muffling device according to claim 1, further comprising a determination unit that instructs to relatively attenuate the components.