JPH0635486A - Noise cancellation system - Google Patents

Abstract

PURPOSE:To constitute a incoherent systems as the secondary propagation system from an adaptive filter to each observation point and to improve the follow-up performance of noise cancellation. CONSTITUTION:An adaptive signal process part 21c determines the coefficient of the adaptive filter 21a by performing an adaptive signal process so as to cancel noises at respective observation points, and the adaptive filter 21a filters a reference signal inputted through a 2nd digital filter 21e according to the determined coefficient value to generate plural noise canceling signals and inputs the noise canceling signals to a speaker through a 1st digital filter 21d. The characteristics of the 1st digital filter 21d are so determined that the canceling sound propagation systems from the adaptive filter output terminal to the respective observation points do not interfere with one other, and the characteristics of the 2nd digital filter 21e are so determined that the total characteristics of the 2nd digital filter 21e, the adaptive filter 21a whose coefficient is set to an initial value, the 1st digital filter 21d, and the canceling sound propagation systems 23 are nearly equal to the opposite-sign characteristics of the propagation characteristics of a noise propagation system 22.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a noise canceling system, and more particularly to a noise canceling system capable of canceling noises at a plurality of positions (observation points) in an automobile and listening to a comfortable audio sound.

[0002]

2. Description of the Related Art As a noise countermeasure, a method using a sound absorbing material (passive control) has been conventionally known. However, in the method using the sound absorbing material, it is troublesome to form a quiet area where noise is small, and there is a problem that the bass cannot be effectively eliminated. In particular, in order to prevent noise in the passenger compartment of an automobile, there is a problem that the weight of the automobile increases and the noise cannot be effectively eliminated. For this reason, a method (active control) of radiating a noise canceling sound having a phase opposite to that of the noise from the speaker to reduce the noise has been in the spotlight and is being put to practical use in a part of indoor space such as a factory or an office. Also, a method has been proposed in which noise is reduced by active control even in the passenger compartment of an automobile.

FIG. 13 is a block diagram of a conventional device for realizing noise cancellation, and 11 is an engine which is a noise source, and 1 is a noise source.
Reference numeral 2 is a rotation speed sensor that detects the engine rotation speed R, and 13 is a reference signal generation unit that generates a sine wave signal of a constant amplitude having a frequency according to the engine rotation speed R as a reference signal S _N. Reference numeral 14 denotes a noise canceling controller, which receives the reference signal S _N generated from the reference signal generating unit 13 and also receives the noise Sn and the canceling sound at the noise canceling position (observation point, for example, near the driver's ear) in the vehicle interior. The synthesized sound signal of Sc is input as the error signal Er,
Adaptive signal processing is performed so that the error signal is minimized, and a noise cancel signal Nc is output. The noise cancellation controller 14 includes an adaptive signal processing unit 14a, an adaptive filter 14b having a digital filter configuration, and D / which converts an adaptive filter output into an analog noise cancellation signal Nc.
An A converter 14c, a filter 14d created based on a transfer function of a cancellation sound propagation system (secondary sound propagation system) from a speaker to a noise cancellation point, and an A / D converter 14e that converts an analog error signal into a digital signal.
And have. Reference numeral 15 is a power amplifier for amplifying the noise canceling signal Nc, 16 is a canceling speaker for radiating the noise canceling sound Sc, 17 is arranged at a noise canceling point, and detects a synthesized sound of the noise Sn and the canceling sound Sc,
It is an error microphone that outputs a synthetic sound signal as an error signal Er.

The adaptive signal processing unit 14a receives the error signal Er at the noise canceling point and the reference signal S _N ′ corresponding to the noise inputted through the filter 14d, and cancels the noise at the noise canceling point using these signals. Adaptive filter 14b
Determine the coefficient of. For example, the adaptive signal processing unit 14a determines the coefficient of the adaptive filter 14b so that the error signal Er input from the error microphone 17 is minimized according to a well-known LMS (Least Mean Square) adaptive algorithm. The adaptive filter 14b digitally filters the reference signal S _N according to the coefficient determined by the adaptive signal processor 14a, and outputs the noise cancel signal N _C from the DA converter 14c. It should be noted that the reference signal S _N must be a signal having a high correlation with the noise S _n desired to be erased, and a sound having no correlation with the reference signal is not erased. When the engine 11 rotates, its rotational speed R is detected by the rotational speed sensor 12, and the reference signal generator 13 generates a reference signal S _N (see FIG. 14 (a)) having a frequency corresponding to the engine rotational speed R, Input to the noise cancellation controller 14. At this time, a periodic engine sound (periodic noise) generated from the engine 11 propagates through the air having a noise propagation system (primary sound propagation system) having a predetermined transfer function and reaches a noise cancellation point. Therefore, the noise (engine sound) Sn at the noise canceling point is slightly weakened in level and delayed, as shown in FIG.
As shown in 4 (b).

First, the noise canceling controller 14 has a noise canceling signal Nc whose phase is opposite to that of the reference signal S _N , for example.
Then, the cancel speaker 16 outputs the cancel sound Sc shown in FIG. 14 (c). However, since the noise Sn is out of phase with the noise Sn, the noise is not canceled by the cancel sound Sc, and the error signal Er is generated. The noise canceling controller 14 performs adaptive signal processing so that the error signal Er becomes minimum, and the adaptive filter 14
The coefficient of b is determined, and finally, in the ideal case, FIG.
As shown in (1), the phase of the cancel sound Sc is made opposite to the phase of the noise Sn, and the levels are matched to cancel the noise.

In order to simplify the explanation, the above is an example in which one noise source, one canceling sound source (speaker), and one noise canceling point (observation point) are provided. However, in reality, there are a plurality of noise sources and a plurality of points (observation points) where noise is desired to be canceled. Therefore, one speaker cannot cancel noise at each observation point, and a plurality of speakers exist. FIG. 15 is a configuration diagram of a conventional noise canceling device when there are K noise sources, M speakers, and L observation points. Reference numeral 21 is a noise canceling controller that operates to cancel noise at each observation point, and 22 is a primary sound virtual propagation system (noise propagation system) that represents a system in which noise propagates from each noise source (not shown) to each observation point. ), 23 includes the characteristics of the speaker (not shown),
Secondary sound propagation system (canceling sound propagation system) that expresses a system in which cancellation sound propagates from each speaker to each observation point, 2
Reference numeral 4 denotes a signal synthesizing unit expressing the function of the microphone at each observation point, the adding units 24 _{1 to} 24 ₁ ′ correspond to the microphones at the first observation point, and the adding units 24 _{2 to} 24 ₂ ′ at the second observation point. It corresponds to a microphone, and the adding units 24 _{L to} 24 _L ′ are the L-th
It corresponds to the microphone at the observation point. _{d d1n} ~d dLn is an external noise is not a cancellation target in each observation point.

[0007] In the noise cancellation controller 21, 21a are input to the reference signal x _a1n ~x _AKN corresponding to noise generated from the noise source (output from the reference signal generating unit (not shown)), the noise cancellation signals y _a1n ~ A multi-input-multi-output adaptive filter (hereinafter simply referred to as an adaptive filter) for inputting _yaMn into each speaker, 21b is a secondary sound propagation system 2
3 is a filter for creating a filtered X signal created by using each element (propagation element) of the transfer function matrix of 3 and a reference signal x _{a1 n
xaKn} is input, 21c is the synthesized sound signals (error signals) e _{1n to} e _{Ln at} each observation point and the filtered X output from the filtered X signal creation filter 21b.
It is input a signal r _111n ~r _LMKn, an adaptive signal processing section for determining the coefficients of the adaptive filter 21a performs adaptive signal processing to cancel the noise at each observation point by using these signals.

FIG. 16 is an explanatory diagram of the primary sound virtual propagation system 22. As shown in FIG. 16 (a), each of K noise sources NG _{1 to} N is generated.
The noise generated from G _K propagates through the primary sound propagation system 22 having a predetermined frequency / phase characteristic and is provided at each observation point by a microphone.
Reach (MIC _{1 to} MIC _L ). Therefore, assuming that the transfer characteristic of the propagation system in which the noise from the i-th noise source NGi reaches the j-th microphone MIC _j is H _ji , the primary sound virtual propagation system 22 is as shown in FIG. 16 (b). Expressed, its transfer function matrix (H) is as follows.

[0009]

[Equation 1]

Each element H _ij of the transfer function matrix (H) is modeled by the FIR type digital filter shown in FIG. That is, the delay element DL that sequentially delays the input signal by one sampling time and the coefficient h ₀ ,
It is modeled by a digital filter including a multiplication unit ML that multiplies h ₁ , h ₂ , ... And an addition unit AD that adds the outputs of the multiplication units. FIG. 18 is an explanatory diagram of the secondary sound propagation system 23. As shown in FIG. 18 (a), the noise canceling sounds generated from the speakers SP _{1 to} SP _M are secondary sound propagation having a predetermined frequency / phase characteristic. It propagates through the system 23 and reaches the microphones MIC _{1 to} MIC _L provided at each observation point. Therefore, _assuming that the transfer characteristic of the secondary sound propagating system in which the cancel sound based on the i-th noise cancel signal y _ain reaches the j-th microphone MIC _j is C _ji , the secondary sound propagating system 23 is shown in FIG. It is expressed as shown in b), and its transfer function matrix (C) is as follows.

[0011]

[Equation 2]

Each element of the transfer function matrix (C) has the FIR shown in FIG. 17, as in the case of the primary sound virtual propagation system 22.
Modeled by type digital filters. That is,
Delay element D for sequentially delaying the input signal by one sampling time
It is modeled by a digital filter including L, a multiplication unit ML that multiplies each delay element output by the coefficients c ₀ , c ₁ , c ₂ , ..., And an addition unit AD that adds each multiplication unit output. Figure 1
9 is a block diagram of the filtered X signal producing filter 21b produced by using each element C _ij of the transfer function matrix (C) of the secondary sound propagation system 23. Updating adaptive signal processing unit 21c and the reference signal _{x a1n} ~x aKn, the coefficients of the adaptive filter by performing adaptive signal processing based on the synthesized speech signal e _1n to e _Ln between noise and canceling sound at each observation point and the adaptive filter 21a is input to a speaker is inputted to the reference signal x _a1n ~x _AKN generate noise cancellation signals y _a1n ~y _aMn, to cancel the noise of each observation point. By the way, the noise cancellation signal y output from the adaptive filter 21a
_{a1n to} yaMn _do not reach the observation point as they are, but reach them with the frequency / phase characteristics of the secondary sound propagation system 23 _added . Therefore, the adaptive signal processing unit 21c causes the reference signal x
_Filtered X using a signal obtained by adding the characteristics of the secondary sound propagation system 23 to the reference signal without directly using _a1n to x _aKn
The LMS (MEFX LMS) algorithm is adopted to perform more advanced noise cancellation control. That, filtered-X In LMS algorithm, reference is input to the adaptive filter 21a signal x _a1n ~x _AKN filter 2
The coefficients of the adaptive filter 21a are updated using the signals _r11In to _rLMKn (filtered X signals) filtered by 1b and the synthesized sound signal at each observation point.

In FIG. 19, C_ijIs the secondary sound propagation system 23
Each element C of the transfer function matrix (C) at_ij(Figure
18) is an FIR type digital filter that realizes
It Refer to each filter 21b for creating filtered X signal
Signal x_a1n~ X_aKnAll propagation elements on (all
Filter through the filter corresponding to the propagation element of
Code X signal r_111n~ R_LMKnIs designed to output
It That is, the reference signal x _a1nIs the first speaker
From all the observation points to C₁₁~ C_L1To activate
Lardard X signal r_111n~ R_L11nAnd the reference signal x
_a1nPropagation element from the 2nd speaker to all observation points
C₁₂~ C_L2To act on the filtered X signal r _121n~ R
_L21nTo output the reference signal x_a1nTo the Mth spin
Propagation element C from the marker to all stations_1M~ C_LMTo act
Filtered X signal r_1M1n~ R_LM1nIs output and
Like the reference signal x_a2n~ X_aKnAll propagation elements
Let Incidentally, R₁₁= (R_111n, R_211n, ・ ・ ・ R_L11n) R_{twenty one}= (R_121n, R_221n, ・ ・ ・ R_L21n) R_M1= (R_1M1n, R_2M1n, ・ ・ ・ R_LM1n) R_MK= (R_1MKn, R_2MKn, ・ ・ ・ R_LMKn).

FIG. 20 shows a multi-input / multi-output adaptive filter 2.
It is a block diagram of 1a, and has the same structure as the primary sound virtual propagation system 22 and the secondary sound propagation system 23. A _11n- A _MKn is F
For example, delay elements DL1 and D configured by an IR type digital filter for sequentially delaying an input signal by one sampling time
It is realized by L2, multiplication units ML1 to ML3 that multiply the outputs of the delay elements by coefficients a ₀ , a ₁ , and a ₂ , and addition units AD1 to AD2 that add the outputs of the multiplication units. The number of delay stages is not limited to two. Noise cancellation signal y _a1n to be input to the first speaker is obtained by adding the reference signal x _a1n ~x _AKN to input to the digital filter A _11n ~A _1Kn, each reference signal x _a1n ~x _AKN Digital filter A _21n
~ A noise cancellation signal y _a2n to be input to the second speaker is obtained by inputting to A _2Kn and adding,
... Each reference signal x _{a1n to} x _aKn is _converted to digital filter A
_By inputting and adding _M1n to _AMKn , the noise cancellation signal _yaMn input to the Mth speaker is obtained.

Each FIR digital filter A _{11n to} A _MKn in the adaptive filter 21a has three coefficients (two-stage delay).
In the above configuration, the adaptive signal processing unit 21c performs adaptive signal processing for each of the three coefficients of the FIR digital filters A _{11n to} A _MKn to determine the coefficient value. Ie one FIR
The coefficients a ₀ , a ₁ and a ₂ of the type digital filter A _ijn are calculated by the following equation to determine the coefficients a ₀ , a ₁ and a ₂ .

[0016]

[Equation 3]

In equation (1), (n) is the value at the current sampling time, (n-1) is the value one sampling time before, (n-2) is the value two sampling times before, and (n + 1). Means the value from the current time to the next sampling time. Further, μ is a constant of 1 or less, R _ij is an output signal of the filter 21b, and e _n is a synthetic sound signal of noise and cancellation sound at each observation point, which are expressed as follows.

[0018]

[Equation 4]

[0019] according According to the noise cancellation device, adaptive signal processing unit 21c and the filtered-X signal r _111n ~r _LMKn which is the output of the filter 21b, the synthesized speech signals e _1n ~ the noise and canceling sound at each observation point Based on e _Ln , adaptive signal processing is executed to determine the coefficients of the respective FIR digital filters A _{11n to} A _{MKn forming} the adaptive filter 21a, and the adaptive filter 21a receives the reference signals x _{a1n to} x _aKn as input. input to the speaker SP ₁ to SP _M (FIG. 18) to generate a noise cancellation signal y _a1n ~y _aMn, acts as the speaker cancels the noise of each observation point by generating a canceling sound.

In FIG. 21, the number of noise sources K = 2 and the number of speakers M =
2 is a configuration diagram of a specific conventional noise canceling device when the number of observation points (the number of microphones) L = 2, and 21a is configured by four FIR type digital filters A ₁₁ , A ₂₁ , A ₁₂ and A _22. adaptive filter, each propagation elements C ₁₁ and 21b the secondary propagation system transfer function matrix _{(C), C 21, C} 12, C
Filters for creating a filtered X signal in which ₂₂ is a digital filter, 21c-1 to 21c-4 are adaptive signal processing units (MEFs).
X LMS algorithm), SP ₁ and SP ₂ are speakers,
MC ₁ and MC ₂ are microphones installed at the observation points.

[0021]

According to the conventional noise canceling apparatus, as described above, the noise canceling signal y _a1n
~ Y _aMn is generated and input to each speaker SP _{1 to} SP _M ,
Each speaker produces a cancel sound and acts to cancel the noise at each observation point. However, in the conventional noise canceling device, the secondary sound propagation system usually has a cancel sound transmission path extending from all actuators (speakers) to all sensors (microphones) (see FIG. 18). Such a system is called an interfering system, but the interfering system is not a control target and causes the following problems. That is, depending on the characteristics of the secondary sound propagation system, some actuator outputs (cancellation sounds) may have opposite actions at some observation points. In such a case, noise (vibration) at each observation point may occur. ) Alternately and repeatedly becomes larger and smaller.

For example, even when a predetermined speaker outputs a cancel sound so as to sufficiently cancel the noise at the corresponding observation point, the cancel sound from another speaker arrives due to the interference system. Therefore, the noise canceling device performs control so as to further cancel noise at the observation point, and inputs a new noise canceling signal to the speaker. However, this new noise canceling signal acts in a direction that does not cancel the noise, and the noise is increased. As a result, the noise canceling device acts so as to cancel the noise by generating a noise canceling signal in the direction of reducing the noise next. After that, such operation is repeated, and the noise (vibration) at the observation point is repeatedly increased or decreased, and the noise cannot be canceled at all times. Such a phenomenon is particularly prominent when rapid adaptive control is performed. originally,
Since the adaptive control system tries to always perform optimal control by following the characteristic fluctuation of the external system by performing adaptation, the phenomenon as described above becomes a particularly large defect in the adaptive control system.

Further, the noise canceling device performs adaptive signal processing to sequentially determine the coefficients of the adaptive filter, and finally converges to a predetermined coefficient value (optimal value) to cancel noise.
By the way, if the initial value of the coefficient of the adaptive filter deviates significantly from the value to be converged (optimal value), the time required for the adaptive filter to reach the optimum value becomes long, and the noise canceling followability deteriorates. Occurs. Since the conventional noise canceling device does not consider such a point, it takes a long time to cancel the noise, and when the noise state changes, the change cannot be followed and the noise cannot be canceled sufficiently. There was a problem. From the above, the first object of the present invention is to provide a noise canceling method capable of solving the problem caused by the interfering system, that is, the phenomenon that the noise is increased or decreased at the observation point. A second object of the present invention is to provide a noise canceling system with good followability.

[0024]

[Means for Solving the Problems] The above problems include a cancel sound generation source (speaker) that outputs a cancel sound to cancel noise at a plurality of observation points, respectively.
A sensor (microphone) that detects the synthesized sound of the noise and the cancel sound at each observation point, and an adaptive signal that cancels the noise at each observation point using the synthesized sound signal at each observation point and the reference signal according to the noise An adaptive signal processing unit that performs processing to determine the coefficient of the adaptive filter; an adaptive filter that performs digital filter processing on the reference signal based on the coefficient to generate a plurality of noise cancellation signals and inputs the noise cancellation signal to the speaker; This is achieved by the first and second digital filters provided before and after.

[0025]

The canceling sound is output from the speaker in order to cancel the noise at each of the plurality of observation points, and the synthesized sound of the noise and the canceling sound at each observation point is detected by the microphone. The adaptive signal processing unit determines the coefficient of the adaptive filter by performing adaptive signal processing so as to cancel the noise at each observation point by using the synthetic sound signal at each observation point and the reference signal corresponding to the noise, and the adaptive filter is determined. The reference signal input through the second digital filter is subjected to digital filter processing based on the obtained coefficient value to generate a plurality of noise cancellation signals, and the noise cancellation signals are output to the speaker through the first digital filter. To enter.
The characteristics of the first digital filter are set so that the cancel sound propagation systems from the output end of the adaptive filter to each observation point do not interfere with each other. The noise canceling signal of the propagation system does not have an adverse effect and the repetition of sound at the observation point disappears, and the noise can be canceled at all times.
Further, the characteristics of the second digital filter are the total of the second digital filter, the adaptive filter having the coefficient initially set, the first digital filter, and the entire cancel sound propagation system from each cancel sound source to each observation point. Since the characteristic is set to be substantially equal to the opposite sign characteristic of the noise propagation system characteristic, in other words, the coefficient value of the second digital filter is close to the value (optimal value) at which the coefficient of the adaptive filter converges. Since it is set to, it is possible to perform noise cancellation with good followability.

In order to make the cancellation sound propagation systems non-interfering with each other, the diagonal terms of the transfer function matrix representing the overall characteristics of the first digital filter and the entire cancellation sound propagation system from each speaker to all the observation points. The gain of is much larger than the gain of the off-diagonal term. In this case, if the first digital filter characteristic is set to the inverse characteristic (inverse filter characteristic) of the cancellation sound propagation system, a non-interference system can be obtained, and the pair of the total transfer function matrix of the first digital filter and the cancellation sound propagation system can be formed. The angular term can be represented by a delay element, and the filter for generating a filtered X signal can be realized only by a delay element, which eliminates the conventional convolution operation for generating a filtered X signal and can reduce the amount of calculation.

[0027]

【Example】

(a) Overall configuration of the first embodiment of the present invention FIG. 1 is a configuration diagram of a noise canceling system according to an embodiment of the present invention, in which K noise sources exist and M speakers are used.
This is a case where there are L observation points (microphones). In the figure, 21 is a noise canceling controller that operates to cancel noise at each observation point, and 22 is a system in which noise propagates from K noise sources (not shown) to L observation points (noise propagation system). ) Representing the primary sound virtual propagation system, 2
3 is a secondary sound propagation system (cancellation sound propagation system) that represents a system in which the cancellation sound propagates from each of the M speakers to each observation point, including the characteristics of the speaker (not shown), and 24 is each observation point In the signal synthesizer that expresses the function of the microphone in
The adders 24 _{1 to} 24 ₁ ′ correspond to the microphones at the first observation point, the adders 24 _{2 to} 24 ₂ ′ correspond to the microphones at the second observation point, ... Adders 24 _{L to} 24 _L ′
Is the microphone at the Lth observation point, and d
_{d1n ~d dLn} is an external noise is not a cancellation target in each observation point.

Noise canceling controller 21
21a is a noise cancellation signal y_{a1 n}~ Y_aNnOutput
Multi-input-multi-output adaptive filter, 21b is a filter
This is a filter for creating the X signal, and the noise you want to cancel.
Reference signal x according to_a1n~ X_aKn(A reference signal (not shown)
The second digital fill, which will be described later)
21c at each observation point
Synthetic sound signal e_1n~ E_LnAnd a filtered X signal generation file
Filtered X signal r output from the filter 21b ₁₁~ R
_NKIs input, and the noise at each observation point is
Adaptive signal processing is performed to cancel the sound.
The adaptive signal processing unit for updating the filter coefficient, 21d is adaptive
The multi-input multi-output first provided in the latter stage of the filter 21a
It is a digital filter, from the adaptive filter to each observation point.
So that the respective secondary sound propagation systems at
21e is an adaptive filter 21a.
Multi-input multi-output second digital filter provided in front of
It is First digital filter 21e and coefficient
The adaptive filter 21a set to the period value and the first digital
The total characteristics of the secondary filter 21d and the secondary sound propagation system 23 are
It is almost equal to the inverse sign characteristic of the propagation characteristic of the primary sound virtual propagation system 22.
Is set to be. Primary sound virtual propagation system 22
The secondary sound propagation system 23 has the same structure as the conventional one, and
Have the structures shown in FIGS. 16 and 18, respectively.

First Digital Filter The first digital filter 21d is composed of a large number of FIR type digital filters B _{11 to} B _MN as shown in FIG. B _{11 to} B _MN are elements when the characteristics of the digital filter 21d are expressed by a transfer function matrix. F
The IR type digital filters B _{11 to} B _MN are delay elements DL1 and D1 for sequentially delaying the input signal by one sampling time.
L2 ... and the coefficients b ₀ , b ₁ , b ₂ ...
It is realized by the multiplication units ML1, ML2, ML3 ... For multiplying by and the addition units AD1, AD2. Adaptive filters 21a noise cancel signal outputted from the y _a1n ~y _ANN is inputted to the FIR type digital filter B ₁₁ .about.B _1N respectively, the noise cancellation signal y to be input to the first speaker by adding their outputs _b1n is obtained, and each noise cancellation signal y _a1n ~
y _aNn are respectively FIR type digital filters B _{21 to} B _2N
The noise cancellation signal y _b2n to be input to the second speaker is obtained by inputting to and adding to each of the noise canceling signals y _{a1n to} y _aNn to the digital filters B _{M1 to} B _MN. The noise cancellation signal y _bMn input to the M-th speaker is obtained by adding and adding.

This first digital filter 21d is constructed.
FIR type digital filter B ₁₁~ B_MNCoefficient of
b₀, B₁, B₂... is the output end of the adaptive filter 21a
From the sound source to each observation point
Determined to be Wataru. Specifically, each speaker
From the entire secondary sound propagation system 23 to the first observation point
Transfer function matrix expressing the overall characteristics of the digital filter 21d
Rix's diagonal term gain is greater than off-diagonal term gain
Each coefficient is determined so as to be large. Figure 3 shows the first
Each FIR digital filter in the digital filter 21d
Ruta B ₁₁~ B_MN6 is an explanatory diagram of a coefficient value determination method of FIG. Spi
If the number of markers M = 2 and the number of observation points L = 2, the first digital
Transfer function mat of the secondary filter 21d and the secondary sound propagation system 23
Ricks (B) and (C) are each a matrix of 2 rows and 2 columns.
Can be expressed as a quad, and the first digital filter and secondary sound
The transfer function matrix (P) showing the overall characteristics of the propagation system is also
It can be expressed by a matrix of 2 rows and 2 columns, and equation (2) holds.

[0031]

[Equation 5]

In equation (2), the transfer function matrix (C) of the secondary sound propagation system 23 is known by measurement, and
Each element of the total transfer function matrix (P) is “diagonal term P
The gains of ₁₁ and P ₂₂ are considerably larger than the gains of the off-diagonal terms P ₁₂ and P ₂₁ (the gain of the diagonal term >> the gain of the off-diagonal term). For example, the gain of off-diagonal terms is set to zero. Therefore, (2) the elements B ₁₁ .about.B ₂₂ of the digital filter 21d is Motomari from the equation, it can determine the coefficients of the FIR type digital filter for realizing these elements.

FIG. 4 is an explanatory view of the comprehensive secondary sound propagation system from the output end of the adaptive filter 21a to each observation point, and the noise cancellation signal y _a1n output from the adaptive filter 21a.
˜y _{a Nn} reach the observation point via their own secondary sound propagation systems 23-1 to 23-L (transfer functions are P _{11 to} P _LN ), and y _c1n ˜y.
_cLn . In this case, each secondary sound propagation system is non-interfering with another secondary sound propagation system, and the noise cancellation signal y _ain of the other system does not interfere with a certain secondary sound propagation system. . That is, the noise cancellation signal y _{a in} operates to cancel noise only the i observation point,
A non-interfering system is formed without affecting any other observation points.

Second Digital Filter The second digital filter 21e is composed of a large number of FIR type digital filters W _{11 to} W _NK as shown in FIG. Each of the FIR digital filters W _{11 to} W _NK has a delay element D that sequentially delays the input signal by one sampling time.
L1, DL2, ... And coefficients w ₀ , w ₁ ,
Multipliers ML1, ML2, ML3 for multiplying w ₂ ...
And addition units AD1, AD2 for adding the outputs of the respective multiplication units
・ ・ Consist of The coefficients w ₀ , w ₁ , w ₂ ... Of each FIR type digital filter are determined by a method described later so that the coefficients of the adaptive filter 21a converge to the optimum values in a short time, and thus good followability is obtained. You can cancel the noise. By _{inputting the} reference signal x _a1n to the FIR digital filters W _{11 to} W _N1 , signals yw _11n to yw _N1n are obtained from the respective filters, and the reference signal x _a1n is obtained.
_{By inputting a2n} to the FIR type digital filters W _{12 to} W _N2 , signals yw _12n to yw _N2n are obtained from the respective signals,
..., the reference signal x _aNn is the FIR digital filter W
Signal yw _1Kn ~yw _NKn is obtained by inputting the _1K to _W-NK.

Adaptive Filter The adaptive filter 21a is composed of a large number of FIR type digital filters A _{11 to} A _NK and adders AD1 to AD _N as shown in FIG. 6, and each FIR type digital filter has a second digital filter 21e. Signals output from yw _{11n to} yw _NKn
Is input, and the FIR digital filter A _ij (j = 1 to 1
The i-th noise cancellation signal y _ain is obtained by adding the output of K) to the adder AD _i .

[0036]Filter for creating filtered X signal The filtered X signal producing filter 21b is shown in FIG.
Having a structure, the first digital filter 21d and the secondary sound transmission
Diagonal term of total transfer function matrix (P) of carrier system 23
FIR type that realizes the transfer function (the off-diagonal term is 0)
Digital filter P₁₁, P_{twenty two}・ ・ ・ ・ ・ ・ P_NNSet of K
It is composed of Is it the second digital filter 21e?
Signal yw output from_11n~ Yw_N1nIs the first set of FIR type
Digital filter P₁₁, P_{twenty two}・ ・ ・ ・ ・ ・ P_NNEntered in
Filtered X signal r₁₁~ R_N1And the signal yw_12n~
yw_N2nIs the second set of FIR digital filters P₁₁, P
_{twenty two}・ ・ ・ ・ ・ ・ P_NNTo the filtered X signal r₁₂~
r_N2Is obtained, and the signal yw is obtained._1Kn~ Yw_NKnIs the Kth group
FIR type digital filter P ₁₁, P_{twenty two}・ ・ ・ ・ ・ ・ P_NNTo
Input and filtered X signal r_1K~ R_NKIs obtained.

When the number of noise sources associated with the adaptive signal processing unit and other parts is 2, the number of speakers is 2, and the number of observation points is 2, the noise canceling controller 21 has the configuration shown in FIG. 8 (first digital filter). Is omitted). 21 in the figure
Reference numeral a is an adaptive filter, reference numeral 21b is a filtered X signal generating filter 21c-11, 21c-21, 21c-12, 21c-22 is an adaptive signal processing unit (LMS), and reference numeral 21e is a second digital filter. Each adaptive signal processing unit 21c-11, 21c-21, 21c-12, 21c-22
Determines the coefficients of the FIR digital filters A ₁₁ , A ₂₁ , A ₁₂ , and A _{22 in} the adaptive filter 21a. For example, the adaptive signal processing unit 21c-ij (i, j = 1, 2) uses the FIR
The three coefficients a ₀ , a ₁ , and a ₂ of the type digital filter A _ij are determined by performing the calculation of the following equation.

[0038]

[Equation 6]

In the equation (1) ', (n) is the value at the current sampling time, (n-1) is the value one sampling time before, and (n-2) is 2
The value before the sampling time, (n + 1) means the value from the current time to the next sampling time. Further, μ is a constant of 1 or less, and r _ij is a filtered X signal creation required filter 21b.
, E _in is an error signal (synthetic sound signal) at each i observation point.

[0040]Method for determining characteristics of second digital filter For example, all the elements A of the adaptive filter 21a_ijTransfer function of
Set to 1 to force the adaptive filter to do no adaptation.
Even if it is, the coefficient of the second digital filter 21e
If is set to a value close to the optimum value, some noise
You can cancel. Therefore, in advance, the second data
Calculate and set the coefficient value of the digital filter 21e
Then, the coefficient of the adaptive filter 21a is changed from 1 by adaptive control.
If updated, the coefficient of the adaptive filter 21a will be optimal in a short time.
To cancel noise with good followability by converging on the value.
Will be able to. FIG. 9 shows the second digital filter 2
Each FIR digital filter W in 1e ₁₁~ W_NKof
It is explanatory drawing of a coefficient value determination method. Of the adaptive filter 21a
Setting the transfer functions to all 1s, in other words, the adaptive filter
Each FIR type digital filter A constituting the filter 21a_ij
Coefficient a₀~ A₂Initialize so that the transfer function is 1.
Then, the noise cancellation system becomes as shown in Fig. 9.
It However, the number of noise sources, the number of speakers, noise cancellation
It is assumed that the number of points (observation points) is 2. Scarecrow
Consider the case where noise is canceled in the system
Then, the following formula is established.

[0041]

[Equation 7]

From the equations (3) and (3) ', the following equations (H) + (C) (B) (A) (W) = 0 (4) (H) + (P) (A) ( W) = 0 (4) '(where (P) = (C) (B)) holds. In the equation (4), (H) and (C) are known by measurement, (B) is already known because it has already been obtained, and (A) is known because all of its transfer functions are set to 1. . Therefore, the second from equation (4)
The respective elements W _{11 to} W _NK of the digital filter 21e are obtained, and the respective coefficients of the FIR type digital filter realizing the elements can be determined. For example, if the number of noise sources K = 2, the number of speakers M = 2, and the number of observation points L = 2, the transfer function matrix (B) of the first digital filter 21d, the transfer function matrix (W) of the second digital filter 21e, The transfer function matrix (H) of the primary sound virtual propagation system 22, the transfer function matrix (C) of the secondary sound propagation system 23, and the transfer function matrix (A) of the transfer function matrix of the adaptive filter 21a are respectively expressed as a matrix as follows. it can.

[0043]

[Equation 8]

By substituting these into the equation (4), the respective elements W _{11 to} W ₂₂ of the second digital filter 21e are obtained. Thus, if each transfer function element of the second digital filter 21e is obtained and set, the coefficient of the second digital filter becomes a value (optimum value) at which the coefficient of the adaptive filter 21a should finally converge. The values are close to each other and good followability can be achieved. FIG. 10 is a characteristic explanatory diagram of the second digital filter 21e. From the equation (4) ′, the total characteristics of the second digital filter 21e, the adaptive filter 21a having the initial value set, and the non-interference system (see FIG. 4) are (P) (A) (W) = − (H) The above-mentioned comprehensive characteristic is equivalent to the reverse sign characteristic of the primary sound virtual propagation system 22.

Overall Operation A cancellation sound is output from the speaker in order to cancel noise at each of a plurality of observation points. The signal synthesizer 24 detects the synthesized sound signals e _{1n to} e _Ln of the noise and the cancel sound at each observation point and inputs them to the adaptive signal processor 21c. Second digital filter 21e based on the coefficient value (fixed) which is set in advance, subjected to FIR digital filter processing to the reference signal x _a1n ~x _AKN being input a plurality of signals yw _11n ~yw _NKn Occurred and filtered X
The signal is input to the signal generation filter 21b and the adaptive filter 21a. The filtered X signal generation filter 21b applies a filtering process to the input signals yw _{11n to} yw NKn _according to the element P _ij of the non-interference system matrix (P) to generate filtered X signals r _{11 to} r _NK and adapts them. The signal is input to the signal processing unit 21c. Adaptive signal processing unit 21c and the output signal r ₁₁ ~r _NK of filtered-X signal generation filter 21b, the synthesized speech signal e _1n to e of the noise and the cancel sound at each observation point
Based on _Ln , the adaptive algorithm process is executed according to the equation (1) ′ to determine and update the coefficients of each FIR type digital filter A _{11 to} A _NK (see FIG. 6) forming the adaptive filter 21a.

The adaptive filter 21a uses the second digital filter 2 based on the coefficient value determined by the adaptive signal processing.
First digital filter 21 in the signal yw _11n ~yw _NKn being input from 1e subjected to FIR digital filter processing to generate a plurality of noise cancellation signal y _a1n ~y _ANN
Enter in d. The first digital filter 21d _performs a FIR digital filtering process on the input noise canceling signals y _{a1n to} y _aNn based on a preset coefficient value (fixed) to _generate a plurality of noise canceling signals y _{b1n to} y _{b1n to} y.
_{Generate bMn} and input to each of M speakers. As a result, each speaker outputs a cancel sound and acts to cancel the noise at each observation point, and the above operation is repeated thereafter.

In this case, the first digital filter 21d
The characteristics of are set so that the canceling sound propagation systems from the output end of the adaptive filter to each observation point do not interfere with each other. Therefore, for one canceling sound propagating system, noise canceling of another canceling sound propagating system is performed. The signals do not interfere with each other, and the noise level at the observation point does not repeat. In addition, the second digital filter 2
The characteristic of 1e is that the total characteristic of the second digital filter 21e, the adaptive filter 21a whose coefficient is initially set, the first digital filter 21d, and the secondary sound propagation system 23 is the propagation characteristic of the primary sound virtual propagation system 22. In other words, the coefficient value of the second digital filter 21e is set to a value close to the value (optimal value) at which the coefficient of the adaptive filter 21a converges. Therefore, it is possible to perform noise canceling control with good followability.

(B) Second Embodiment of the Present Invention In the above, in the total transfer function matrix (P) of the first digital filter 21d and the secondary sound propagation system 23,
The characteristics of the digital filter 21d were determined so as to satisfy the condition that "the gain of the diagonal term is considerably larger than that of the non-diagonal term (gain of the diagonal term >> gain of the non-diagonal term)". As a special case that satisfies this condition, the characteristic of the first digital filter 21d may be set to the inverse characteristic (inverse filter characteristic) of the secondary sound propagation system 23. By doing so, it is possible to form a non-interfering system, and the diagonal term of the total transfer function matrix of the first digital filter 21d and the secondary sound propagating system 23 can be expressed by the delay element. The X signal generation filter 21b can be realized by only a delay element, the conventional convolution calculation for generating the filtered X signal becomes unnecessary, and the calculation amount can be reduced. Note that (B) (C) = (P)

[0049]

[Equation 9]

When is satisfied, (B) is said to have the inverse characteristic (inverse filter characteristic) of (C). FIG. 11 is an explanatory diagram of the comprehensive secondary sound propagation system from the output end of the adaptive filter to each observation point when the inverse filter characteristic is used. Noise canceling signal y _a1n output from the adaptive filter 21a
y _ANN is the y _C1N ~y _CLN respectively reached the observation point via its own secondary sound propagation system 23-1 to 23-L. The transfer function of each secondary sound propagation system 23-1 to 23-L is the product of the constant p _ij and the delay term of Δ _ij sampling time. In this case, the respective secondary sound propagation systems do not interfere with each other from the other secondary sound propagation systems, and the noise cancellation signal y _ain of the other system does not interfere with the certain secondary sound propagation system. An interference system is formed. Further, the diagonal term of the total transfer function matrix (P) becomes a delay element, so that each filter of the filtered X signal generating filter 21b can be realized by only the delay element.

As described above, the first digital filter 2
When the transfer function matrix (B) of 1d has the inverse characteristic of the transfer function matrix (C) of the secondary sound propagation system 23, (P) (A) (W) =-(H) is obtained from the equation (4) ′. . Therefore, the second digital filter 21e, the adaptive filter 21a having the initial value set, and the non-interference system (see FIG. 11).
(See FIG. 12) is equivalent to the inverse sign characteristic -H of the propagation characteristic of the primary sound virtual propagation system 22 as shown in FIG. Although the present invention has been described above with reference to the embodiments, the present invention can be variously modified according to the gist of the present invention described in the claims, and the present invention does not exclude these.

[0052]

As described above, according to the present invention, the first digital filter is provided after the adaptive filter, and the cancel sound propagation systems from the adaptive filter to each observation point are made non-interfering with each other. The noise canceling signals of other canceling sound propagating systems do not interfere with the canceling sound propagating system, and the noises at the observation point do not repeat, and noise can be canceled at all times. Also, a second digital filter is provided in front of the adaptive filter, and the second digital filter, the adaptive filter having an initial value set, and the first digital filter are provided.
Since the total characteristics of the digital filter and the secondary sound propagation system are set to be substantially equal to the inverse sign characteristics of the propagation characteristics of the primary sound propagation system, the coefficient value of the second digital filter is set to the coefficient of the adaptive filter. The value can be set to a value close to the convergent value (optimal value), and noise canceling with good followability can be performed.

Further, according to the present invention, since the first digital filter characteristic is set to the inverse filter characteristic of the secondary sound propagation system to form the non-interference system, the first digital filter and the secondary sound propagation are formed. The diagonal term of the total transfer function matrix of the system is only the delay element. Therefore, the filter for generating the filtered X signal in the noise canceling controller can be realized only by the delay element, and the tatami for the conventional filtered X signal generation can be realized. The complicated calculation becomes unnecessary and the calculation amount can be reduced.

[Brief description of drawings]

FIG. 1 is a configuration diagram of an embodiment of the present invention.

FIG. 2 is a configuration diagram of a first digital filter.

FIG. 3 is an explanatory diagram of a characteristic determining method of a first digital filter.

FIG. 4 is an explanatory diagram of a comprehensive secondary sound propagation system from an adaptive filter to an observation point.

FIG. 5 is a configuration diagram of a second digital filter.

FIG. 6 is a configuration diagram of an adaptive filter of the present invention.

FIG. 7 is a configuration diagram of a filter for generating a filtered X signal according to the present invention.

FIG. 8 is a configuration diagram of a noise canceling controller of the present invention when there are two noise sources, speakers, and observation points.

FIG. 9 is an explanatory diagram of a characteristic determining method of a second digital filter.

FIG. 10 is a characteristic explanatory diagram of a second digital filter.

FIG. 11 is an explanatory diagram of a comprehensive secondary sound propagation system when the inverse filter characteristic is used.

FIG. 12 is a characteristic explanatory diagram of a second digital filter.

FIG. 13 is a configuration diagram of a conventional noise canceling device.

FIG. 14 is a waveform diagram for explaining a noise canceling operation.

FIG. 15 is a configuration diagram of a conventional noise canceling device when there are a plurality of noise sources, speakers, and observation points.

FIG. 16 is an explanatory diagram of a primary sound virtual propagation system.

FIG. 17 is a configuration diagram of a digital filter that realizes each element of a transfer function matrix.

FIG. 18 is an explanatory diagram of a secondary sound propagation system.

FIG. 19 is a configuration diagram of a filter for generating a filtered X signal.

FIG. 20 is a configuration diagram of an adaptive filter.

FIG. 21 is a configuration diagram of a conventional noise canceling device when there are two noise sources, speakers, and observation points.

[Explanation of symbols]

 21 .. Noise canceling controller 21a .. Adaptive filter 21b .. Filtered X signal creation filter 21c .. Adaptive signal processing unit 21d .. First digital filter 21e .. Second digital filter 22 .. Primary sound virtual Propagation system 23 .. Secondary sound propagation system 24 .. Signal synthesizer

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Shuji Sawada 1-4-1 Chuo, Wako, Saitama Stock Company Honda R & D Co., Ltd.

Claims (2)

[Claims] 1. A cancel sound is output to cancel noise at each of a plurality of observation points, a combined sound of the noise and the cancel sound at each observation point is detected, and a combined sound signal and a noise source at each observation point are detected. The reference signals corresponding to the noise generated from are input, the coefficient of the adaptive filter is determined by performing adaptive signal processing so as to cancel the noise at each of the observation points using these signals, and the noise signal is output to the adaptive filter. In the noise canceling method in which a plurality of noise canceling signals obtained by inputting are input to the canceling sound generation source, the canceling sound from the adaptive filter to each observation point is provided by providing the first and second digital filters before and after the adaptive filter. The characteristics of the first digital filter provided after the adaptive filter are set so that the propagation systems do not interfere with each other. Set,
In addition, the overall characteristics of the second digital filter, the adaptive filter whose coefficient is set to the initial value, the first digital filter, and the entire cancel sound propagation system from each cancel sound source to each observation point are A noise cancellation method characterized in that the characteristic of the second digital filter is set so as to be substantially equal to the inverse sign characteristic of the propagation characteristic of the entire noise propagation system from each noise source to each observation point. 2. The noise cancellation according to claim 1, wherein the first digital filter characteristic is set to an inverse filter characteristic of an entire cancellation sound propagation system from each cancellation sound generation source to each observation point. method.