JPH0926793A - Active silencing device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an active silencing device which can silence a three- dimensional space with a single sound source without specifying the frequency and direction. SOLUTION: A monopole sound source is composed of a speaker 6 and an external noise detecting microphone 2 which is opposed to the speaker. This source is controlled by a control system in which the system is approximated by a first-order time lag system in a minute frequency band by limiting the frequency range.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an active muffling device that actively cancels external noise to muffle the space.

[0002]

2. Description of the Related Art In recent years, an active silencer has been put into practical use, which adds a secondary sound to noise and cancels it by overlapping them to positively eliminate the noise. The simplest configuration is
This is shown in FIG. For example, the noise propagating from the noise source 1 in the air-conditioning duct is detected by the noise detecting microphone 2, the amplitude and phase of the input signal are adjusted by the delay circuit 3 and the inverting circuit 4, and then amplified by the amplifier 5. Then, the secondary sound is added from the rear speaker 6. The control method is open control that does not have a feedback loop, so that signals do not diverge and high-speed control is possible, but when the acoustic characteristics from the microphone 2 to the speaker 6 change, it has the drawback that it cannot follow it. .

Therefore, as shown in FIG. 16, an error microphone 7 is installed behind the speaker 6 to detect a residual sound. A feedback circuit 8 that adjusts the gain and phase of the control circuit so that the residual sound is always minimized is provided to respond to changes in the acoustic characteristics from the microphone 2 to the speaker 6. In this case, since the error microphone 7 detects the additional sound from the speaker 6 in addition to the noise, there is a drawback that it diverges and causes howling, and a physical howling countermeasure is required. Further, a processing speed for performing feedback at high speed is required.

The above control can be realized by an analog circuit, but recently, in many cases, an adaptive control algorithm for optimally changing a transfer function by a digital signal processing circuit is used. Compared to analog circuits, digital circuits
The calculation accuracy of the circuit element is improved, and there is no secular change.
Since filter coefficients can be replaced, development time can be shortened, cost can be reduced, LSI can be easily implemented, and if DSP (Digital Signal Processor) is used, signal processing compatible with various systems is possible only by program development. There are advantages such as

By the way, when it is desired to control a three-dimensional wide space different from a one-dimensional space such as a duct line, a method of using a plurality of systems shown in FIG. 15 or 16 can be considered. A signal other than the residual sound, that is, the sound of the speaker of another device is also input to the sensor 7, so that signal interference between the devices may occur. Therefore, a multi-channel system in which a plurality of sensors and speakers are collectively controlled to mute a large space is conceivable, but it is not practical because the system becomes complicated. A monopole sound source that refers only to the sound around the error sensor is effective as a method for solving this problem.

The present invention relates to active noise reduction using a monopole sound source (hereinafter referred to as a monopole method). The monopole method uses a monopole sound source composed of a speaker provided close to a microphone for detecting external noise. The closed control active muffling shown in FIG. 16 described above requires a minimum of two microphones, that is, the monitor microphone 2 that detects external noise and the error microphone 7 that detects residual noise. The microphone 7 is unnecessary and the structure is simple. In addition, the monopole method enables silencing that does not depend on the external acoustic path.

The prior art relating to the silencer of the monopole method is (1) Japanese Unexamined Patent Publication No. 4-166898 (Applicant: Nishiwaki Giken, Matsushita Electric Works) and (2) Japanese Unexamined Patent Publication No. 5-210.
No. 391 (Applicants: Nishiwaki Giken, Matsushita Electric Works) and the like are known. In these conventional techniques, a monopole sound source is used to mitigate a traveling wave in a pipe such as an air conditioning duct, that is, a one-dimensional sound wave, and an optimal sound source arrangement method is shown. I haven't touched on the method. Further, according to the above publication, an analog filter is used, but since adaptive control cannot be performed, it is presumed that sufficient performance cannot be exhibited in a three-dimensional sound field in which the transmission system may fluctuate.

As another known example, MFB (Motiona1 Feedbac
References (ACTIV)
E CONTROL OF PERIODIC NOISE USING MOTIONAL FEEDBAC
K, Proc.of Inter-Noise 94, PP.1203-1206, Usagawa etc.)
There is. This is a method of feeding back the impedance change of the speaker instead of the signal of the microphone close to the speaker, and belongs to the monopole method in principle. The control method is adaptive control using a digital signal processing circuit. In the above literature, for active muffling of random sound without periodicity, the Filtered-X LMS algorithm is conventionally used, but a sensor that refers to external noise is required, and the cost due to the demand for speedup of the control circuit is high. increase,
They point to disadvantages such as slow convergence speed. Therefore, paying attention to periodic sound, Synchronized LMS, Notch Filter, Delayed-
We have presented control methods such as X Harminics Synthesizer, but require a sensor to detect the periodic component.

[0009]

As described above, adaptive control by a digital signal processing circuit is generally used in active noise reduction, and a feedback type LMS (Least Mean Square) is used.
e: Least squares algorithm is widely used.
However, this method requires a sensor corresponding to an error sensor which detects a reference signal in the related art. Also, adaptive control of feedback systems that do not satisfy causality is difficult. On the other hand, spatial muffling using a plurality of devices is difficult to practically use because it requires a complicated processing device to control the interference between the microphone and the speaker sound source. Therefore, a monopole sound source that does not require a reference signal is conceivable, but this method means that a signal already input is canceled by using a signal with a time delay, which does not satisfy the causality. Therefore, there is a problem that the LMS algorithm does not operate properly.

The present invention has been made in view of the above-mentioned circumstances, and solves the above-mentioned drawbacks of the prior art, and it is possible to mute a three-dimensional space with a single sound source without specifying the frequency and direction. An object is to provide a silencer.

[0011]

In order to achieve the above object, the invention of claim 1 limits a frequency range of a monopole sound source composed of a speaker and a microphone for external noise detection facing the speaker. Therefore, in the minute frequency band, the system configured is controlled by a control system which is approximated to be a first-order lag system.

Further, the invention of claim 2 is characterized in that a digital signal processing circuit is used as a control circuit, a filter which operates by an adaptive algorithm is provided, and the control system is provided in the filter.

The invention of claim 3 is characterized in that a wide range of spatial muffling is possible by using one or a plurality of monopole sound sources.

Further, the invention of claim 4 is characterized in that it is possible to muffle the sound in a wide frequency band by using a plurality of filters operating by an adaptive algorithm in the signal processing circuit.

A fifth aspect of the present invention is characterized in that the active silencing device according to the first to fourth aspects is provided on a vertical plate.

The invention according to claim 6 is characterized in that the active silencer according to any one of claims 1 to 4 is provided at the entrance of the room space to provide sound insulation inside and outside the room space.

The invention according to claim 7 is characterized in that the active processing device according to any one of claims 1 to 4 is provided on the inner wall surface of the duct.

[0018]

According to the invention of claim 1, when the filter in the monopole method limits the frequency range, attention is paid to the fact that the system constructed in the minute frequency band is a simulation of a time delay system, and the noise signal is When it can be considered that the delay time is steady and the delay time is sufficiently short, the active damping can be stably achieved by adaptively controlling the first-order delay system for each frequency.

According to the second aspect of the present invention, the digital signal processing circuit is used in the control circuit and the filter is operated by the adaptive algorithm. Therefore, if the frequency range is limited, the system configured in the minute frequency band is 1 It is possible to easily realize a control system that is approximate to a second-delay system.
By using a digital signal processing circuit, signal processing corresponding to various systems can be performed only by developing a program, so that the development period of the entire system can be shortened and the cost can be reduced.

According to the invention of claim 3, by using one or more monopole sound sources, it is possible to perform control without specifying the propagation direction of the sound wave. Therefore, it is possible to mute the three-dimensional space. Also, by using multiple monopole sound sources,
It is possible to mute without receiving interference between sound sources,
As a result, it is possible to mute a wider space.

According to the invention of claim 4, by using one or a plurality of monopole sound sources, it is possible to perform control without specifying a frequency. Therefore, it is possible to mute sound in an arbitrary frequency range. Further, by using a plurality of monopole sound sources, it is possible to muffle sound without receiving interference between the sound sources, and thereby mute sound in a wider frequency range.

According to the fifth aspect of the present invention, since the active muffling device is provided on the vertical plate, it is possible to muffle the space facing the vertical plate. Therefore, for example, even when the whole office is noisy due to the noise of the office automation equipment, etc. by raising the vertical plate in a part of the office,
A quiet space can be formed in a part of the office.

According to the invention of claim 6, since the active muffling device is provided at the entrance of the room space, the entrance of the room space is opened, and even if people can freely enter and leave the room space, It can provide sound insulation inside and outside. As a result, the room space can be made a quiet space that is shielded from external noise. Further, even when a musical instrument is played indoors, the sound does not leak outside.

According to the invention of claim 7, by disposing a plurality of silencers in series with respect to the sound propagation path in the duct, the silencer of the present invention prevents mutual interference between monopole sound sources. Since there is no sound source, a muffling effect proportional to the number of installed monopole sound sources can be expected. Therefore, it is possible to muffle the noise in the duct within an arbitrary attenuation amount and frequency range, and prevent the noise of the air conditioner from being propagated to, for example, an office room that should be quiet.

[0025]

An embodiment of the present invention will be described below with reference to the accompanying drawings.

A schematic block diagram of the monopole active control system is shown in FIG. The signal received by the microphone 2 is input to the digital signal processing board 11. And
A space to be converted into a digital signal by the A / D conversion circuit 12, signal-processed by the adaptive filter 13, inversely converted to an analog signal by the D / A conversion circuit 14, amplified by the amplifier 15, and silenced from the speaker sound source 6. Emitted as sound waves. The emitted sound wave propagates through the space and the microphone 2
Is detected by Therefore, a control loop which is an input signal of the adaptive filter 13 is formed through the acoustic path including the microphone 2 and the secondary sound source speaker 6.

Here, the transfer characteristic of the secondary sound source speaker 6 can be approximately regarded as a simple time delay at low frequencies. Therefore, the monopole active control means that the signal already input is erased by using the signal delayed in time, and the causality is not satisfied. However, if the noise signal is steady and the delay time is sufficiently short, adaptive control is possible. That is, the sensor output with respect to noise is a sine wave as shown in FIG. 2A, the impulse response of the adaptive filter 13 is a delta function that only reverses the sign, and the impulse response of the acoustic path is delayed by the time τ. When fτ << 1 is satisfied, the output of the adaptive filter is (b) of FIG. 2, and the signal synthesized by the sensor output is the sum of (a) and (b) of FIG. Noise will be attenuated as shown in FIG.

FIG. 3 is a block diagram of a control system for a monopole sound source. For the speaker 16 of the noise source 17, the spectrum of the sound wave at the position of the microphone 2 is X
(F), the conversion gain from the sound pressure of the microphone to the voltage is H _m
(f), the transfer function of the adaptive filter is H _f (f), and the conversion gain from the voltage of the secondary sound source speaker to the sound pressure is H _s (f). At this time, if the microphone output voltage is Y (f), the following equation is obtained. Y (f) = Y (f) H _f (f) H _s (f) H _m (f) + X (f) H _m (f) (1) Therefore, the microphone output Y (f) is Y (f) = X (f) H _m (f) / (1-H _f (f) H _s (f) H _m (f)) (2). From the above equation, the stability condition is 1> | H _f (f) H _s (f) H _m (f) | (3). Also, the sound pressure P _s (f) at the position of the error sensor is P _s (f) = X (f) / (1-H _f (f) H _s (f) H _m (f)) (4) Becomes Practically, the transfer function H _m (f) of the microphone can be regarded as flat, and the transfer function H _s (f) of the speaker can be regarded as a high-pass filter with a time delay. _The condition required for _f (f) is that the time delay of H _f (f) is sufficiently small so that the effect on causality due to the inclusion of H _f (f) does not worsen. The value of the expression -H _f (f) H _s (f) H _m (f) is as small as possible within the range that satisfies That is,
If the low cutoff frequency of the adaptive filter is f1 and the high cutoff frequency is fh, the phase rotation Δθ in this frequency range is τ because the time delay of the acoustic path including the characteristics of the adaptive filter is τ, so Δθ = 2π Since (f _h −f ₁ ) τ (5), if this value is sufficiently smaller than 2π, the conditional expression (3) for stability is satisfied, and stable noise control is possible.

In the case of using the means of the low-frequency content of the noise obtained by the monopole active control scheme that operates theoretically completely stable _{H f (f) H s (} f) H m (f) is As close as possible to -1,
Obtained when its absolute value does not exceed 1, and its value is 6d
It turns out that it is B. Further, it can be seen that a low-pass filter with f ₁ = 0 is suitable as the adaptive filter satisfying the above-mentioned two conditions.

On the other hand, when the characteristics of the acoustic path are known, further noise attenuation is possible. The phase rotation Δθ of the adaptive filter is set so as to satisfy equation (5), and the center frequency fc of the adaptive filter is set to H _f (fc) H _s (fc) H _m (fc)
If the phase is set to be π, an attenuation amount of 6 dB or more can be obtained. In this case, a band pass filter is suitable as the adaptive filter 13.

For example, the adaptive control using a low-pass filter as the adaptive filter 13 will be described in order to operate stably even when the characteristics of the acoustic path are unknown.
As described in the previous section, there are two conditions for adaptive control: that the time delay τ is as small as possible and that the gain satisfies Expression (3). Therefore, the present inventor has decided to define a low-pass filter in the frequency domain and inversely transform it to obtain the filter coefficient. That is, in the (n + 1) th update, the microphone output is subjected to an appropriate time window, M samples are taken out, Fourier transform is performed, and the power spectrum Pn is calculated. Obtain new coefficients H _{n + 1} (f) from H _n (f), P _n (f) and P _n-1 (f). Inverse Fourier transform. The first half of the real part is cut out with appropriate weight and used as the impulse response. The filter coefficient was designed by the following four processes.

Here, the updating formula of the real part of the filter coefficient is as follows. H _{n + 1} (f) ＝ H _n (f) + μ (P _n-1 (f) -P _n (f)) ^1/2 if P _n (f) <P _n-1 (f) H _{n + 1} (f) ＝ H _n (f) -μ (P _n (f) -P _n-1 (f)) ^1/2 if ((P _n-1 (f) <P _n (f) & H _n ( f)-(P _n (f) -P _n-1 (f)) 1/2 ≧ 0 H _{n + 1} (f) = 0 otherwise where μ is a step size parameter.
The imaginary part of the filter coefficient is always 0.

Here, _assuming that the sampling period is T _s , the tap length of the filter is L, and the Hanning window of length M is used as the time window, the frequency resolution is 2 / MT _s . On the other hand, the A / D converter required to realize the adaptive filter,
Assuming that the total time obtained by adding the time delay due to the calculation time required for the D / A converter and the filtering and the time delay from the speaker to the microphone is T _d , the effective frequency band of this monopole active control system is fT. _d << 1
Is the low frequency range. Therefore, the frequency point that can be used in DFT is 2pT _d / MT _s except for DC (0th frequency point).
It is up to the frequency point p where << 1. In addition, L> M / 2p is also required for the adaptive filter to operate properly.

FIG. 4 shows the number of updates 0 (ANC OFF) and 300.
The power spectrum of the error sensor output when 0 (ANC ON) is shown. As can be seen from the figure, when ANC is ON, cancellation is performed from DC to 200 Hz. In this figure, the amount of cancellation varies,
Separately, as a result of examination with an FFT analyzer, 160H from DC
Up to z, a constant cancellation amount of about 5 dB was obtained, and at frequencies higher than that, both were almost the same.

As data for the next experimental example, the distance between the speaker 6 and the microphone 3 was 3 cm in the case shown in FIG. 1 in which a speaker and a microphone were used. Further, the noise source speaker 16 was installed at a position 90 cm from the microphone 2 and the level was 76 dB (A) at the position of the microphone 2. The delay time from the drive voltage output of the speaker 6 to the output of the microphone 2 including the delay times of the A / D converter 12 and the D / A converter 14 was 0.78 mS. Therefore, since the sampling cycle is 0.04 mS, the number of delay points (T _d / T _s ) is about 20.
At this time, in order to satisfy the cancellation condition sufficiently, P
Must be 1. That is, the actual frequency range is 0 <
f ≦ 50 Hz.

In FIG. 5, the number of updates 0 (ANC OFF) and 300
The power spectrum of the microphone output when 0 (ANC ON) is shown. The difference between the two was 0.3 dB, which was almost no difference. This is because the tap length of the adaptive filter is 128.
This is because it is as short as (5.12 ms, cycle of about 200 Hz), and the effect can be expected to be improved by using a high-speed DSP and grasping the transfer characteristics of the microphone and the speaker in advance.

As an experimental example of the active silencer in the above-mentioned adaptive algorithm, first, the simulation result by the computer will be shown. First, P ₀ = 0, μ = 0.01, L = 128, M =
The state of convergence was investigated for the case of 256, T _d / T _s = 1,2,5. The results are shown in FIGS. 6, 7 and 8. In each figure, (a), (b) and (c) are frequency points p =
It shows the relationship between the power of the microphone output and the number of updates in the cases of 1, 2, and 3.

Further, FIGS. 9A and 9B show the impulse response and transfer function of the filter when T _d / T _s = 1 and the number of updates is 10000, respectively. Further, FIG. 10 shows the power spectrum (100 times average) of the microphone output when the number of updates is 0 and 10000. From this result, it is found that the proposed adaptive algorithm is effective.

In the simulation, the result of an experiment using a low-pass filter instead of the speaker and the microphone is shown. A block diagram of the experiment is shown in FIG. Sampling cycle is 0.04mS, DSP board input (A
The delay time from the filter input before the / D converter) to the output (filter output after the D / A converter) is 0.254 mS.
(Approximately T _d / T _s = 6 points), the number of adaptive filter taps is 128, the number of FFT points is M = 512, and the step size parameter μ is 0.0001. Further, since the adapted frequency range is 1 ≦ p ≦ 3 in the frequency number of FFT, the actual frequency is 50 Hz ≦ f ≦ 150 Hz.
At this time, even if p takes the maximum value of 3, 2pT _d / M
Since T _s ≈0.07 <1.0, the canceling conditional expression (3) is sufficiently satisfied.

If the devices are installed in parallel in the space, it is possible to obtain a wide space in which the sound is reduced by the volume reduction of one device (for example, 5 dB). For example, as shown in FIG. 12, a partition used in an office or the like is provided with this device, and a large number of the devices can be arranged to realize an active sound insulating wall, and noise through the partition can be effectively silenced. This makes it possible to create a quieter space with a partition having the same height as the conventional one, or to create a space with the same quietness as the conventional one with a lower partition. This device can be attached not only to a partition but also to an outdoor soundproof wall, which can block traffic noise and the like.

Further, as shown in FIG. 13, when the external sound field and the internal sound field communicate with each other only at the entrance and exit, such as in a concert hall and an anechoic room, by providing this device at the entrance and exit, It is possible to reduce the intrusion of noise or the leakage of sound to the outside. This is provided with the monopole sound source 20 of this embodiment, which is composed of a microphone and a speaker, at three locations of a closed room space 21 and an entrance 22. As a result, external noise is transmitted from the entrance 22 to the room space 21.
Even when a concert or the like is not performed inside the room space 21 or inside the room space 21, the sound leaking out from the entrance is greatly reduced even though the entrance is open. If the monopole sound sources 20 are arranged in parallel in double and triple at the entrance / exit, it is possible to make a quieter indoor space.

Further, as shown in FIG. 14, for noise propagating in a duct such as an air conditioner to which many conventional active noise suppressors are applied, the device 20 is arranged in series with the acoustic path. By doing so, at the outlet of the duct, a noise reduction effect of (volume reduction of one monopole sound source device) × (number of devices) can be expected. As described above, the present apparatus can be arbitrarily arranged in parallel or in series, and has an advantage that mutual interference does not occur.

[0043]

As described above, spatial muffling is possible in principle even in the conventional muffling device, but it is necessary to control a large number of microphones and speakers at the same time, which complicates the device. In the active silencer of the present invention, it is possible to silence the three-dimensional space by a single sound source,
By arranging a plurality of independent silencing devices of the present invention, the silencing effect can be further enhanced without interfering with each other, and therefore the applicability is high. Therefore, conventionally, the application range is limited to an air conditioning duct or the like, but by using the muffling device of the present invention, it is possible to muffle a sound, and mute a three-dimensional space such as a muffling doorway or a muffling room.

[Brief description of drawings]

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

FIG. 2 is a diagram showing the principle of silencing.

FIG. 3 is a block diagram of a control system of a monopole sound source according to an embodiment of the present invention.

FIG. 4 is a diagram showing a sound deadening effect of a sound deadening device according to an embodiment of the present invention, showing a comparison between ANC ON and ANC OFF.

FIG. 5 is a diagram showing a sound deadening effect of a sound deadening device according to an embodiment of the present invention, showing a comparison between ANC ON and ANC OFF.

FIG. 6 is a diagram showing an effect of noise amount reduction in computer simulation, where T _d / T _s = 1 and (a),
(B) and (c) respectively show the relationship between the frequency, the power of the microphone output when p = 1, 2, and 3 and the number of updates.

FIG. 7 is a diagram showing an effect of noise amount reduction in computer simulation, where T _d / T _s = 2, (a),
(B) and (c) respectively show the relationship between the frequency, the power of the microphone output when p = 1, 2, and 3 and the number of updates.

FIG. 8 is a diagram showing the effect of noise amount reduction in computer simulation, where T _d / T _s = 5, (a),
(B) and (c) respectively show the relationship between the frequency, the power of the microphone output when p = 1, 2, and 3 and the number of updates.

FIG. 9 shows (a) impulse response and (b) transfer function of the adaptive filter when T _d / T _s = 1 and the number of updates is 10000.

FIG. 10: T _d / T _s = 1, update count 0 (ANC OFF) and 1
A diagram showing the microphone output when 0000 (ANC ON).

FIG. 11 is a block diagram of an experiment using a filter.

FIG. 12 is an explanatory view of a pendant equipped with a monopole sound source according to an embodiment of the present invention.

FIG. 13 is an explanatory diagram of a silencer equipped with a monopole sound source according to an embodiment of the present invention at the entrance of a room space.

FIG. 14 is an explanatory diagram of a duct including a monopole sound source according to an embodiment of the present invention.

FIG. 15 is an explanatory diagram showing an example of a conventional open-control type silencer.

FIG. 16 is an explanatory diagram showing an example of a conventional closed control type silencer.

[Explanation of symbols]

 1 Noise Source 2 Microphone 6 Speaker 11 Digital Signal Processing Board 12 A / D Converter 13 Adaptive Filter 14 D / A Converter 15 Amplifier 16 Noise Generation Speaker 17 Noise Source

Claims (7)

[Claims] 1. A monopole sound source composed of a speaker and a microphone for external noise detection facing the speaker is limited in a frequency range by limiting the frequency range.
An active silencer characterized in that it is controlled by a control system that approximates that the system constructed is a first-order lag system. 2. The active muffler according to claim 1, wherein a digital signal processing circuit is used as a control circuit, a filter that operates by an adaptive algorithm is provided, and the control system is provided at the filter. 3. The active muffler according to claim 1, wherein a wide range of spatial muffling is possible by using one or a plurality of monopole sound sources. 4. The active muffler according to claim 1, wherein the signal processing circuit includes a plurality of filters that operate according to an adaptive algorithm to enable muffling in a wide frequency band. 5. A sound deadening device according to any one of claims 1 to 4, wherein the active sound deadening device is provided on a vertical plate. 6. An active sound deadening device, characterized in that the active sound deadening device according to any one of claims 1 to 4 is provided at an entrance of a room space to provide sound insulation inside and outside the room space. 7. An active muffler duct device comprising the active treatment device according to any one of claims 1 to 4 on an inner wall surface of a duct.