JP7262499B2 - Active vibration noise reduction device - Google Patents

Description

TECHNICAL FIELD The present disclosure relates to an active vibration noise reduction device that reduces noise by generating and interfering with vibration noise such as vehicle interior noise accompanying engine rotation and vehicle running.

As a method for reducing noise in the vehicle interior, there is a control method using an algorithm called a direct method that does not require prior identification of the acoustic characteristic C and can follow changes in the acoustic characteristic C during control. be.

Patent Document 1 discloses a silencing system using a direct method (see FIG. 1 of Patent Document 1). This muffling system includes an adaptive filter for noise reduction (finite impulse response (FIR) filter (C)) and an adaptive filter (control FIR filter (D )), with three FIR filters acting as adaptive filters (control FIR filters (K)) representing the estimated characteristics of the transfer path (G) from the control loudspeaker to the error microphone. For adaptive updating of the control FIR filter, two virtual error signals e1 and e2 generated inside the system from the error signal e detected by the error microphone are used.

ここで、ｅ１：仮想的な誤差信号、ｅ２：仮想的な誤差信号、ｅ：誤差信号、ｒ：参照信号の時系列信号ベクトル、＊：フィルタリング計算（ＦＩＲフィルタでは畳み込み計算）、Ｃ：制御ＦＩＲフィルタ（Ｃ）のフィルタ係数、Ｋ：制御ＦＩＲフィルタ（Ｋ）のフィルタ係数、Ｄ：制御ＦＩＲフィルタ（Ｄ）のフィルタ係数、である。 The silencing system using the direct method shown in FIG. 1 of Patent Document 1 operates on the following principle.

Here, e1: virtual error signal, e2: virtual error signal, e: error signal, r: time-series signal vector of reference signal, *: filtering calculation (convolution calculation for FIR filter), C: control FIR Filter coefficient of filter (C), K: filter coefficient of control FIR filter (K), D: filter coefficient of control FIR filter (D).

２つの仮想的な誤差信号ｅ１、ｅ２が同時に最小値（０）に収束すると、仮想的な誤差信号ｅ１、ｅ２を用いて更新している制御ＦＩＲフィルタ（Ｃ）及び制御ＦＩＲフィルタ（Ｋ）も一定値に収束するため、上記式は、ｅ＝０、となる。 Thus, in the direct method, two virtual error signals e1 and e2 are calculated inside the system. The following equation is obtained by adding the two virtual error signals e1 and e2 in the above equation.

When the two virtual error signals e1, e2 simultaneously converge to a minimum value (0), the control FIR filter (C) and the control FIR filter (K) updating with the virtual error signals e1, e2 are also Since it converges to a constant value, the above equation becomes e=0.

From the above, if the virtual error signals e1 and e2 simultaneously converge to the minimum value without using the pre-measured value of the acoustic characteristic (G) during control, the sound pressure (e) at the error microphone position will also be minimized. It can be seen that it converges to a value

The "direct method using an FIR filter" and the "update of filter coefficients of the least mean square (LMS) algorithm" are described below.

First, a direct method using an FIR filter will be described with reference to FIG. As shown in FIG. 1, this active vibration noise reduction device uses a primary path transfer characteristic d^ indicating the transfer characteristic of the primary path and a secondary path transfer characteristic y^ indicating the transfer characteristic of the secondary path. The primary path is the path from the vibration noise source to the error signal detection means (error microphone, error microphone). The secondary path is a path from vibration noise canceling means (secondary sound source, speaker) to error signal detecting means.

ここで、ｅ１とｅ２が最小値（＝０）に収束すると、式（ｂ）、式（ｃ）より以下の連立方程式が成立する。

式（２）より、下式が成立する。

式（１）と式（３）より、下式が成立する。

ここで、式（ａ）に式（５）を代入すると、下式が成立し、"ｅ＝０"となる。

よって、直接法によれば、Ｈ＾とＣ＾の真値が未知であっても、ｅ１とｅ２が"０"に収束するとＨ＾とＣ＾の比が一定値（Ｈ＾とＣ＾が一定値）に収束し、制御フィルタ係数Ｗも最適値（＝－Ｈ／Ｃ）に収束するため、誤差信号ｅが最小になる。これが、直接法によって、Ｃの事前同定を必要とせず、制御中にＣが変化しても消音（又は制振）できる原理である。 According to the block diagram of the direct method shown in FIG. 1, the principle that C can be silenced even if C changes during control without prior identification of C is as follows.

Here, when e1 and e2 converge to the minimum value (=0), the following simultaneous equations are established from equations (b) and (c).

The following formula is established from the formula (2).

The following formula holds from formulas (1) and (3).

Here, substituting the formula (5) into the formula (a), the following formula holds and becomes "e=0".

Therefore, according to the direct method, even if the true values of H^ and C^ are unknown, when e1 and e2 converge to "0", the ratio of H^ and C^ is a constant value (H^ and C^ constant value), and the control filter coefficient W also converges to the optimum value (=-H/C), so the error signal e is minimized. This is the principle by which the direct method can silence (or dampen) even if C changes during control without prior identification of C.

Ｃ＾の更新は下式により表される。

Ｗの更新は下式により表される。

各更新式は、ＬＭＳアルゴリズムを用いてフィルタ係数を入力信号と誤差信号を基に誤差信号が最小となるように逐次更新するものである。ここで、各更新式におけるμは、正のスカラ量であり、サンプリングごとの適応フィルタのフィルタ係数の更新量を制御する（決定する）パラメータで、ステップサイズパラメータ（step size parameter）と呼ばれる。なお、一般的に、ステップサイズパラメータは正の定数である。 Next, the update of the filter coefficients by the LMS algorithm using the virtual error signals e1 and e2 by the direct method using the FIR filter will be described. The update of Ĥ is represented by the following equation.

The update of C is expressed by the following equation.

The update of W is represented by the following equation.

Each update formula uses the LMS algorithm to sequentially update the filter coefficients based on the input signal and the error signal so that the error signal is minimized. Here, μ in each update formula is a positive scalar quantity, a parameter that controls (determines) the update amount of the filter coefficients of the adaptive filter for each sampling, and is called a step size parameter. Note that the step size parameter is generally a positive constant.

ここで、時間ステップをｎ、上記３つのＦＩＲフィルタのタップ数（インパルス長）を"Ｎ"とする。基準信号をｘ（ｎ）とし、次式でＸ（ｎ）を定義する。なお、Ｘ（ｎ）は、基準信号の時系列信号ベクトルを表す。

一次経路モデル（推定値、フィルタ）、二次経路モデル（推定値、フィルタ）、制御フィルタのフィルタ係数は以下とする。

誤差信号ｅｎ＝ｅ（ｎ）で、実測値（スカラ）である。
二次音源の出力（スピーカ出力、制御フィルタ出力）をｕ（ｎ）とすると、次式で表される。

また、Ｕ（ｎ）を次式で表す。

参照信号をｒ（ｎ）とすると、次式で表される。

また、Ｒ（ｎ）を次式で表す。

式（ｂ）より、下式となる。

また、式（ｃ）より、下式となる。

From the above, e1 and e2 are expressed by the following equations.

Here, let n be the time step, and "N" be the number of taps (impulse length) of the three FIR filters. Let the reference signal be x(n) and define X(n) by the following equation. Note that X(n) represents the time-series signal vector of the reference signal.

The primary path model (estimated value, filter), the secondary path model (estimated value, filter), and the filter coefficients of the control filter are as follows.

The error signal, en=e(n), is the actual value (scalar).
Assuming that the output of the secondary sound source (speaker output, control filter output) is u(n), it is expressed by the following equation.

Also, U(n) is represented by the following equation.

Assuming that the reference signal is r(n), it is represented by the following equation.

Also, R(n) is represented by the following equation.

From the formula (b), the following formula is obtained.

Further, from the formula (c), the following formula is obtained.

Ｃ＾の更新は、式（７）と式（９）より次式で表される。

Ｗの更新は、式（８）と式（１０）より次式で表される。

The update of H^ is represented by the following equation from equations (6) and (9).

The update of C^ is represented by the following equation from equations (7) and (9).

The update of W is represented by the following equation from equations (8) and (10).

According to the direct method, there is no need to pre-identify the acoustic characteristics C of the spatial transfer path from the control sound source (speaker) to the error microphone, and it is possible to silence (or dampen) even if C changes during control. . However, the direct method requires three adaptive filters: a control filter (filter coefficient; W), a filter for the primary path model (filter coefficient Ĥ), and a filter for the secondary path model (filter coefficient Ĉ). .

As shown in Patent Document 1, when an FIR filter is used for each adaptive filter, when updating the filter coefficients of these three filters, the above-described update equations (A), (B), and (C) are used. Since it becomes a convolution operation so that In addition, for example, in the case of cancellation of vehicle interior vibration noise, it is necessary to increase the sampling frequency and increase the number of taps of the FIR filter in order to cope with sudden acceleration of the vehicle. becomes larger. For this reason, the active vibration noise reduction device requires a digital signal processor or the like with a large computing capacity, which causes the problem that the active vibration noise reduction device becomes expensive.

On the other hand, vibrational noise, such as engine roaring noise, generated in synchronization with the rotation of the engine output shaft (engine speed) is a periodic complex sound (harmonic compound tone). That is, the engine roaring sound is a vibration radiation sound generated by the excitation force generated by the engine rotation being transmitted to the vehicle body. In the case of a 4-cycle, 4-cylinder engine, torque fluctuations due to gas combustion that occur every 1/2 rotation of the engine output shaft generate excitation vibration with the engine as the base point, which causes vibration noise in the passenger compartment. Therefore, in the case of a 4-cycle 4-cylinder engine, vibration noise having a frequency twice as high as the engine speed, which is called a secondary component of rotation of the engine output shaft, dominates the engine booming noise. Therefore, by performing vibration noise control aiming at the dominant vibration noise, it is possible to expect an efficient noise reduction or damping effect. Focusing on this, we detect the frequency that causes vibration noise from the vibration noise source, specialize in vibration noise control of periodic and narrowband noise, and develop harmonic frequencies (vibration By using an adaptive notch filter (SAN type filter; Single frequency Adaptive Notch Filter) to muffle or damp vibration noise (dominant frequency that accounts for most of the noise), the sound muffling/damping effect can be efficiently achieved. The applicant has invented techniques that can be obtained (Patent Document 2, Patent Document 3, etc.). An active vibration noise reduction device using an adaptive notch filter does not require convolution calculations, and can be dealt with by simple four arithmetic operations, and therefore has the advantage of a very small calculation load.

An outline of an active vibration noise reduction device using an adaptive notch filter will be described below.

図３に示すように、"ｉ"を乗算することはπ／２（９０°）反時計回りに回転することになるからである。また、"－ｉ"をかけることは、９０°時計回りに回転することになるため、下式となる。
ｉ＊ｘｃ＝ｃｏｓ（θ＋π／２）＝－ｓｉｎ（θ）＝－ｘｓ
ｉ＊ｘｓ＝ｓｉｎ（θ＋π／２）＝ｃｏｓ（θ）＝ｘｃ
－ｉ＊ｘｃ＝ｃｏｓ（θ－π／２）＝ｓｉｎ（θ）＝ｘｓ
－ｉ＊ｘｓ＝ｓｉｎ（θ－π／２）＝－ｃｏｓ（θ）＝－ｘｃ
よって適応ノッチフィルタは図４に示す構成となる。なお、基準正弦波信号ｘｓ及び基準余弦波信号ｘｃは下式で表される。
ｘｃ＝ｃｏｓ（２πｆｔ）
ｘｓ＝ｓｉｎ（２πｆｔ） The adaptive notch filter with filter coefficients Ĥ shown in FIG. 2 can be viewed as multiplying the magnitude by C and retarding the phase. Frequency: f, 2πf [rad] between 1 [sec] and x [rad] at time t [sec] yields the following formula.
1:2πf=t:x
∴x=2πft
Assuming that C^ delays the phase by φ, the following formula is obtained.

This is because, as shown in FIG. 3, multiplying by "i" results in a .pi./2 (90.degree.) counterclockwise rotation. Also, since multiplying by "-i" means rotating clockwise by 90°, the following formula is obtained.
i*xc=cos(θ+π/2)=−sin(θ)=−xs
i*xs=sin(θ+π/2)=cos(θ)=xc
−i*xc=cos(θ−π/2)=sin(θ)=xs
−i*xs=sin(θ−π/2)=−cos(θ)=−xc
Therefore, the adaptive notch filter has the configuration shown in FIG. Note that the reference sine wave signal xs and the reference cosine wave signal xc are expressed by the following equations.
xc = cos(2πft)
xs=sin(2πft)

評価関数Ｊを最小にするフィルタ係数ｋ１（スピーカの制御フィルタである適応ノッチフィルタのフィルタ係数）を求めるのがＬＭＳアルゴリズムであり、具体的には、評価関数Ｊ（即ち、ｅ^２）をｋ１について偏微分した値（傾きΔ）で、フィルタ係数ｋ１を更新する。傾きΔを求めると下式となる。

評価関数すなわち２乗誤差の傾きΔが求められたので、この傾きΔをステップサイズパラメータμとして用い、評価関数すなわち２乗誤差（ｅ^２）が最小となる伝達特性（ｋ１）をＬＭＳアルゴリズムを用いた適応処理の更新式で求めると、下式となる。

Next, the LMS algorithm will be explained. The following equation holds for the error signal e shown in FIG.

The LMS algorithm obtains the filter coefficient k1 (the filter coefficient of the adaptive notch filter, which is the speaker control filter) that minimizes the evaluation function J. Specifically, the evaluation function J (that is, e ² ) is The filter coefficient k1 is updated with the partially differentiated value (slope Δ). When the slope Δ is obtained, the following formula is obtained.

Since the evaluation function, that is, the slope Δ of the squared error is obtained, this slope Δ is used as the step size parameter μ, and the evaluation function, that is, the transfer characteristic (k1) that minimizes the squared error (e ² ), is obtained using the LMS algorithm. The following formula is obtained by using the update formula for adaptive processing.

二次経路の打消振動騒音推定値ｙは下式となる。

Next, the LMS algorithm using the adaptive notch filter will be described with reference to FIG. Multiplying the cos signal (xc) and the sin signal (xs) by "i" gives the following equation.

The negative vibration noise estimated value y of the secondary path is given by the following equation.

評価関数Ｊを最小にするフィルタ係数Ｗ０、Ｗ１を求めるのがＬＭＳアルゴリズムであり、具体的には、評価関数Ｊをスピーカの制御フィルタである適応ノッチフィルタのフィルタ係数Ｗ０、Ｗ１について偏微分した値をステップサイズパラメータとして用いて、フィルタ係数Ｗ０、Ｗ１を更新する。フィルタ係数Ｗ０、Ｗ１の更新式は下式で表される。

In FIG. 6, the canceling sound transfer characteristic estimated value Ĉ is represented by the following equation.
C ^ = C0 - iC1
W0 and W1 are updated based on the following equations.

The LMS algorithm obtains the filter coefficients W0 and W1 that minimize the evaluation function J. Specifically, the value obtained by partially differentiating the evaluation function J with respect to the filter coefficients W0 and W1 of the adaptive notch filter, which is the speaker control filter. is used as the step size parameter to update the filter coefficients W0, W1. An update formula for the filter coefficients W0 and W1 is expressed by the following formula.

FIG. 7 is a block diagram of the LMS algorithm using an adaptive notch filter based on XAT. In this example, filter coefficients W0 and W1 are represented by the following equations.

JP 2008-216375 A JP-A-2000-99037 Japanese Patent Application Laid-Open No. 2004-361721

However, since the control system using the direct method described in Patent Document 1 updates the three FIR filters during control, the amount of calculation is large compared to the control system using the conventional Filtered-X algorithm. There is a problem that convergence is slow. To address this problem, the control system of Patent Document 1 employs a method of using a high-speed FTF adaptive algorithm for initial convergence, and a highly stable LMS algorithm at the convergence stage. However, regarding the problem of computational load, the control system of Patent Document 1 is based on the assumption that an FIR filter with a large amount of computation is used. There is a problem of becoming

When the adaptive notch filters described in Patent Documents 2 and 3 are used in the direct method, it is important to optimally model the characteristics of the primary and secondary paths. If the model cannot be optimally modeled, the optimum reference signal for updating the filter coefficients of the adaptive notch filter as the control filter cannot be obtained, and for example, it may be difficult to sufficiently follow the rapid acceleration of the vehicle. However, there is a problem that a sufficient vibration noise control effect cannot be obtained.

In view of this background, the present invention uses an adaptive notch filter (SAN) with a small computational load in ANC (Active Noise Control) for silencing noise according to engine speed etc. Therefore, by constructing a control system (SAN filter direct method) that does not require prior identification of acoustic characteristics C and can follow changes in C during control, good noise reduction/control can be achieved even if large changes occur in C. An object of the present invention is to provide an inexpensive active vibration noise reduction device capable of achieving vibration performance.

To solve such problems, one embodiment of the present invention is an active noise and vibration reduction device (10) in which a reference sinusoidal signal having a frequency based on the frequency of the noise and vibration emitted from the noise and vibration source (xs) and a reference cosine wave signal (xc) as reference signals; and a first adaptive notch filter coefficient (W0) forming the real part of the adaptive notch filter coefficient (W). , a first adaptive notch control filter (31) that outputs a first control signal (uc) based on the reference cosine wave signal; and a second adaptive notch filter coefficient ( 31 ) that forms the imaginary part of the adaptive notch filter coefficient (W) W1) and outputting a second control signal (us) based on the reference sinusoidal signal; and adding the first control signal and the second control signal . vibratory noise canceling means (12) for outputting canceling vibratory noise (y) based on the first addition signal (u0) obtained by error signal detection means (11) for outputting an error signal (e) based on the difference from the output canceling vibration noise; and signal transmission from the vibration noise canceling means to the error signal detecting means with respect to the frequency of the reference signal. The first and second reference signals ( correction means (27) for generating r0, r1), said first correction filter being a first adaptive notch having a first correction filter coefficient (Ĉ0) forming the real part of said signal transfer characteristic (C). In the correction filters (41, 43), the second correction filters are respectively composed of second adaptive notch correction filters (42, 44) having a second correction filter coefficient (Ĉ1) forming the imaginary part of the signal transfer characteristic. be done .

The active vibration noise reduction device corrects the reference cosine wave signal and the reference sine wave signal using a third correction filter ( 51 ) and a fourth correction filter ( 53 ), respectively, to generate first and second vibration noise estimation signals. a first estimation signal generation means (28) for generating a vibration noise estimation signal (d̂) by adding the first vibration noise estimation signal and the second vibration noise estimation signal; and the reference cosine wave signal is corrected by the first correction filter ( 41 ) and the third adaptive notch control filter ( 71 ) having the first adaptive notch filter coefficient (W0), and the reference sine wave signal is combined with the second The second correction control signal corrected by the correction filter ( 42 ) and the third adaptive notch control filter ( 71 ) and the reference sine wave signal are combined with the first correction filter ( 43 ) and the second adaptive notch filter coefficient (W1 ), and the reference cosine wave signal corrected by the second correction filter ( 44 ) and the fourth adaptive notch control filter ( 73 ). second estimation signal generation means (27, 70) for generating a first canceling vibration noise estimation signal (ŷ2) by adding the fourth correction control signal and the vibration noise estimation signal (d̂); a first virtual error signal generating means (82) for generating a first virtual error signal (e'2) from the 1 canceling vibration noise estimation signal (y^2); and the first and second reference signals (r0, r1 ) and the first virtual error signal (e′2), the filter coefficients of the third and fourth adaptive notch control filters ( 71 , 73 ) are successively adjusted so that the first virtual error signal is minimized. and a first filter coefficient updating means (72, 74) for updating.

In the above configuration , the active vibration noise reduction device (10) includes a first adaptive notch filter ( 34) , a second adaptive notch filter ( 35) having said second adaptive notch filter coefficient (W1) and outputting a fourth control signal based on said reference cosine wave signal, said first addition signal ( u0) is corrected by a fifth adaptive notch correction filter ( 61 ) having the first correction filter coefficient (C^0), the fifth correction control signal, the third control signal and the fourth control signal and a sixth correction control signal obtained by correcting the second addition signal (u1) obtained by adding the a third estimation signal generating means (60) for generating a second canceling vibration noise estimation signal (ŷ1) by addition; the error signal (e), the vibration noise estimation signal (d̂) and the second canceling a second virtual error signal generating means (81) for generating a second virtual error signal (e'1) from the vibration noise estimation signal (y^1); filter coefficients of the fifth and sixth adaptive notch correction filters ( 61, 63 ) so as to minimize the second virtual error signal based on the third control signal, the fourth control signal, and the second virtual error signal; Second filter coefficient updating means (62, 64) for sequentially updating ( Ĉ0, Ĉ1) may be provided.

In the above configuration, the third correction filter ( 51 H^0) is a third adaptive notch correction filter, the fourth correction filter ( 53 H^1) is a fourth adaptive notch correction filter, and The third and fourth adaptive notch correction filters ( It is preferable to provide third filter coefficient updating means (52, 54) for sequentially updating the filter coefficients of 51, 53).

In the above configuration, the filter coefficients (C^0, C^1) of the fifth and sixth adaptive notch correction filters ( 61 , 63) are added to the filters of these fifth and sixth adaptive notch correction filters ( 61, 63 ) a normalization means (90) for multiplying the reciprocal of the square root of the sum of squares of the coefficients (Ĉ0, Ĉ1) to calculate first and second normalization filter coefficients; ) includes the first adaptive notch correction filter ( 41 ) having the first normalized filter coefficient (C^0) and the second adaptive notch correction filter (41 ) having the second normalized filter coefficient (C^1). 42 ) to generate the first and second reference signals (r0, r1) by correcting the reference cosine wave signal (xc) and the reference sine wave signal (xs), respectively.

In the above configuration, the filter coefficients (C^0, C^1) of the fifth and sixth adaptive notch correction filters ( 61 , 63) are added to the filters of these fifth and sixth adaptive notch correction filters ( 61, 63 ) normalizing means (90) for multiplying the reciprocals of the larger absolute values of the coefficients (Ĉ0, Ĉ1) to calculate third and fourth normalized filter coefficients; (27) comprises the first adaptive notch correction filter ( 41 ) having the third normalized filter coefficient (C^0) and the second adaptive notch correction having the fourth normalized filter coefficient (C^1) Preferably, the reference cosine signal and the reference sine signal are respectively corrected by a filter ( 42 ) to generate the first and second reference signals (r0, r1).

In the above configuration, the first, second, and third filter coefficient updating means (72, 74, 62, 64, 52, 54) control the amount of update of the filter coefficients of the adaptive notch filters that they update, respectively. The parameter (μ) may be determined based on the square root of the sum of squares of the filter coefficients just before the update.

In the above configuration, the first, second, and third filter coefficient updating means (72, 74, 62, 64, 52, 54) control the amount of update of the filter coefficients of the adaptive notch filters that they update, respectively. The parameter (μ) may be determined based on the larger absolute value of the filter coefficients immediately before updating.

As described above, according to the present invention, for example, an error microphone is placed on the headrest near the passenger's ear, and even if a large change in C occurs due to adjustment of the seat position or seat angle, or due to deterioration over time, C does not change. Active vibration noise control that does not impair the silencing performance is possible even if there is a change, and it is possible to greatly improve the silencing effect around the occupant's ears.

Block diagram of vibration noise reduction device by direct method using FIR filter Block diagram of adaptive notch filter Explanatory diagram of the principle of the adaptive notch filter Detailed block diagram of adaptive notch filter Illustration of LMS algorithm Illustration of LMS algorithm Block diagram of LMS with adaptive notch filter based on XAT Explanatory diagram of direct method using adaptive notch filter Block diagram of vibration noise reduction device optimally modeled by direct method using adaptive notch filter FIG. 1 is a configuration diagram showing a first application example of an active vibration noise reduction device according to the present invention; FIG. 2 is a configuration diagram showing a second application example of the active vibration noise reduction device according to the present invention; FIG. 3 is a configuration diagram showing a third application example of the active vibration noise reduction device according to the present invention; Functional block diagram of the active vibration noise reduction device according to the first embodiment Graph showing changes in assumed acoustic characteristics Graph showing the sound pressure level of the engine slamming sound by the active vibration noise reduction device according to the first embodiment in comparison with the control off and the conventional example Functional block diagram of the active vibration noise reduction device according to the second embodiment A graph showing the sound pressure level of the engine slamming sound when the step size parameter is fixed by the active vibration noise reduction device according to the second embodiment, in comparison with the control OFF and the first embodiment. A graph showing the sound pressure level of the engine slamming sound when the step size parameter is made variable by the active vibration noise reduction device according to the second embodiment, in comparison with the case where the control is turned off and the step size parameter is fixed.

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

Optimal modeling when using the adaptive notch filter in the direct method is done as follows.

一次経路モデル（推定値、フィルタ）、二次経路モデル（推定値、フィルタ）、制御フィルタのフィルタ係数は以下とする。

ｘｃに対する適応フィルタをＷ０（ｎ）、ｘｓに対する適応フィルタをＷ１（ｎ）とする。
誤差信号ｅｎ＝ｅ（ｎ）で、実測値（スカラー）である。 Assuming that the time step is n (time; t) and the reference signal at that time is x(n), the following equation is obtained. Two orthogonal reference signals x(xc, xs) are used in the SAN filter.

The primary path model (estimated value, filter), the secondary path model (estimated value, filter), and the filter coefficients of the control filter are as follows.

Let the adaptive filter for xc be W0(n) and the adaptive filter for xs be W1(n).
The error signal, en=e(n), is the actual value (scalar).

一次経路の基準信号をＸとすると、振動騒音推定信号ｄ＾（ｎ）は次式で表される。ここで、振動騒音源では振動騒音が発生しているから、振動騒音の基準信号はｃｏｓ信号（ｘｃ）とする。

なお、マイクに到達する（入力される）騒音（騒音信号）；ｄ（ｎ）は、次式で表される。

打消振動騒音推定信号ｙ＾（ｎ）、及び、参照信号ｒ０（ｎ）、ｒ１（ｎ）は次式で表される。

仮想誤差信号ｅ１（ｎ）、ｅ２（ｎ）は次式で表される。

As shown in FIG. 8, if the secondary sound source output (speaker output, control filter output) is u0(n), then u0(n) is expressed by the following equation.

Assuming that the primary path reference signal is X, the vibration noise estimation signal d̂(n) is expressed by the following equation. Here, since vibration noise is generated at the vibration noise source, the reference signal of the vibration noise is assumed to be the cos signal (xc).

The noise (noise signal) that reaches (inputs) the microphone; d(n) is represented by the following equation.

The canceling vibration noise estimation signal ŷ(n) and the reference signals r0(n) and r1(n) are represented by the following equations.

Virtual error signals e1(n) and e2(n) are represented by the following equations.

Referring to the direct method using the FIR filter, updating of the filter coefficients by the LMS algorithm using the virtual error signals e1 and e2 in the direct method using the SAN filter will be described.

評価関数Ｊ１を最小にするフィルタ係数Ｈ０、Ｈ１（振動騒音源からマイクまで（一次経路）の伝達特性を示す適応ノッチフィルタのフィルタ係数）を求めるのがＬＭＳアルゴリズムであり、具体的には、下式に示すように、評価関数Ｊ１をフィルタ係数Ｈ０、Ｈ１について偏微分した値をステップサイズパラメータとして用いて、フィルタ係数Ｈ０、Ｈ１をそれぞれ更新する。

ここで、μｈ０、μｈ１；ステップサイズパラメータ、である。累乗の微分公式は以下の通りである。

Updating of ĥ, ie H0 and H1, is done based on the following equations.

The LMS algorithm obtains the filter coefficients H0 and H1 (filter coefficients of the adaptive notch filter indicating the transfer characteristics from the vibration noise source to the microphone (primary path)) that minimize the evaluation function J1. As shown in the formula, the values obtained by partially differentiating the evaluation function J1 with respect to the filter coefficients H0 and H1 are used as step size parameters to update the filter coefficients H0 and H1.

where μh0, μh1; step size parameters. The differential formula for exponentiation is:

ここで、μｃ０、μｃ１；ステップサイズパラメータ、である。 Referring to the direct method using the FIR filter, updating of the filter coefficients by the LMS algorithm using the virtual error signals e1 and e2 in the direct method using the SAN filter will be described. Updating C^, ie, updating C0 and C1, is performed based on equation (11) above. The LMS algorithm obtains the filter coefficients C0 and C1 (filter coefficients of the adaptive notch filter indicating the transfer characteristics from the speaker to the microphone (secondary path)) that minimize the evaluation function J1. , the filter coefficients C0 and C1 of the adaptive notch filter are updated using the values obtained by partially differentiating the evaluation function J1 with respect to the filter coefficients C0 and C1 of the adaptive notch filter as step size parameters.

where μc0, μc1; step size parameters.

評価関数Ｊ２を最小にするフィルタ係数Ｗ０、Ｗ１（制御フィルタをなす適応ノッチフィルタのフィルタ係数）を求めるのがＬＭＳアルゴリズムであり、具体的には、下式に示すように、評価関数Ｊ２を適応ノッチフィルタのフィルタ係数Ｗ０、Ｗ１について偏微分した値をステップサイズパラメータとして用いて、フィルタ係数Ｗ０、Ｗ１を更新する。

ここで、μｗ０、μｗ１；ステップサイズパラメータ、である。 W0 and W1 are updated based on the following equations.

The LMS algorithm obtains filter coefficients W0 and W1 (filter coefficients of an adaptive notch filter forming a control filter) that minimize the evaluation function J2. Specifically, as shown in the following equation, the evaluation function J2 is adapted. A value obtained by partially differentiating the filter coefficients W0 and W1 of the notch filter is used as a step size parameter to update the filter coefficients W0 and W1.

where μw0, μw1; step size parameters.

具体的には、式（Ｉ）～式（VII）より、ＳＡＮフィルタを用いた直接法により最適にモデル化した振動騒音低減システムのブロック図が図９として示される。 Based on FIG. 1, which is a block diagram of a direct method using a general FIR filter, FIG. 9 is a block diagram showing a vibration noise reduction system optimally modeled by a direct method using a SAN filter. Two virtual error signals, three filter coefficient update formulas for adaptive notch filters, canceling vibration noise estimation signal, and vibration noise estimation signal are defined as follows.

Specifically, FIG. 9 shows a block diagram of a vibration noise reduction system optimally modeled by a direct method using a SAN filter from equations (I) to (VII).

Ｃ＾が小さいと、フィルタＷの更新に用いる参照信号（ｒ０、ｒ１）が小さくなり、Ｗの収束が遅くなる。更に、Ｃ＾の更新には、Ｗの出力も用いているためＣ＾自身の収束も遅くなる。一方、Ｃ＾が大きい周波数帯域では、ＷとＣ＾の収束が速くなるが、毎回の更新量が大きいため最適値に確実に収束することができないため制御が不安定になりやすい傾向がある。 Since the filter C^ corresponds to an estimated value (secondary path model) of the acoustic characteristics (signal transfer characteristics) from the speaker to the error microphone, its magnitude changes with frequency.

If C^ is small, the reference signals (r0, r1) used to update the filter W are small, and W converges slowly. Furthermore, since the output of W is also used to update C^, the convergence of C^ itself slows down. On the other hand, in a frequency band where Ĉ is large, W and Ĉ converge quickly, but since the amount of update each time is large, it is impossible to reliably converge to the optimum value, and control tends to become unstable.

By normalizing the magnitude of C^ and updating the filter coefficients based only on the phase of C^, the stability of the control is secured and the convergence performance is improved regardless of the size of C^. A direct method using a SAN filter is provided.

Note that "normalization" means normalization of a vector, and is to set the magnitude of the vector to "1" while maintaining the direction of the vector.

Ｃ＾（ｎ＋１）の振幅（大きさ）｜Ｃ＾（ｎ＋１）｜は、次式となる。

正規化後のＣ＾（ｎ＋１）をＣ'＾（ｎ＋１）とすると、Ｃ'＾（ｎ＋１）は次式となる。

正規化されたＣ'＾（ｎ＋１）で、制御フィルタのフィルタ係数（Ｗ（ｎ＋１））の更新と、次の二次経路モデルのフィルタ係数（Ｃ＾（ｎ＋２））の更新を行う。 Normalization of C^ is performed by the following equation.

The amplitude (magnitude) |C^(n+1)| of C^(n+1) is given by the following equation.

If Ĉ(n+1) after normalization is C′̂(n+1), C′̂(n+1) is given by the following equation.

The normalized C′(n+1) updates the filter coefficients of the control filter (W(n+1)) and the next secondary path model filter coefficients (C(n+2)).

As an alternative to the above normalization, it is also possible to use the larger absolute value of C0̂ and C1̂ to reduce the amount of calculation.

Next, variable step size parameters will be described.

Updating the filter coefficients starts from a preset initial value (a small value, such as "0"), so the filter coefficients are small at the start. It is necessary to increase the amount of each update. To increase the update amount, the step size parameter μ is set large. However, if μ is increased, convergence to the optimum value cannot be reliably achieved, so control tends to become unstable, and there is a trade-off between convergence speed and stability.

Focusing on the fact that the filter coefficient is a small value at the beginning of the update, but increases as it approaches the optimum value, the value of the step size parameter is made variable according to the size of the filter coefficient to improve control stability. To provide a direct method using a SAN filter that improves convergence performance while ensuring . The step size parameter can be varied by multiplying the step size parameter of the update formula for each adaptive notch filter by the reciprocal of the amplitude of each adaptive notch filter immediately before updating. Alternatively, the step size parameter of the update formula for each adaptive notch filter is multiplied by the reciprocal of the larger absolute value of the two filter coefficients of each adaptive notch filter immediately before updating. Each update equation with a fixed step size parameter is:

A specific method for varying the step size parameter of the adaptive notch filter update formula by multiplying the reciprocal of the amplitude of each adaptive notch filter immediately before updating is as follows.

各更新式のステップサイズパラメータは、次式となる。

可変ステップサイズパラメータを用いた各更新式は、次式となる。

The amplitude (magnitude) |Ĥ(n)|, |Ĉ(n)|, and |W(n)| of each adaptive notch filter is given by the following equations.

The step size parameter of each update formula is given by the following formula.

Each update formula using the variable step size parameter becomes the following formula.

The step size parameter of each adaptive notch filter update formula is multiplied by the reciprocal of the larger absolute value of the two filter coefficients of each adaptive notch filter immediately before updating. Street.

各更新式のステップサイズパラメータは、次式となる。

可変ステップサイズパラメータを用いた各更新式は、次式となる。

The amplitude (magnitude) |Ĥ(n)|, |Ĉ(n)|, and |W(n)| of each adaptive notch filter is given by the following equations.

The step size parameter of each update formula is given by the following formula.

Each update formula using the variable step size parameter becomes the following formula.

Thus, in the active vibration noise reduction device, the reference signal generating means generates the reference sine wave signal xs and the reference cosine wave signal xc having frequencies based on the vibration noise frequency generated from the vibration noise source as reference signals. The first adaptive notch control filter W0 outputs a first control signal uc based on the reference cosine wave signal xc, and the second adaptive notch control filter W1 outputs a second control signal us based on the reference sine wave signal xs. The vibrating noise canceling means outputs canceling vibrating noise based on a first addition signal u0 obtained by adding the first control signal uc and the second control signal us. The error signal detection means outputs an error signal e based on the difference between the vibration noise generated from the vibration noise source and the canceling vibration noise output from the vibration noise cancellation means. The correcting means converts the reference cosine wave signal xc and the reference cosine wave signal xc and the reference The sinusoidal signal xs is corrected to generate first and second reference signals r0, r1, respectively.

The first estimation signal generating means corrects the reference cosine wave signal xc and the reference sine wave signal xs by the third correction filter Ĥ0 and the fourth correction filter Ĥ1, respectively, to generate the first and second vibration noise estimation signals. Then, the first vibration noise estimation signal and the second vibration noise estimation signal are added to generate the vibration noise estimation signal d̂. The second estimation signal generation means generates a first correction control signal obtained by correcting the reference cosine wave signal xc with the first correction filter Ĉ0 and the first adaptive notch control filter W0, and the reference sine wave signal xs with the second correction filter C A second correction control signal corrected by ^1 and a first adaptive notch control filter W0, and a third correction control signal obtained by correcting the reference sine wave signal xs by a first correction filter C^0 and a second adaptive notch control filter W1. , the reference cosine wave signal xc is added to the fourth correction control signal corrected by the second correction filter Ĉ1 and the second adaptive notch control filter W1 to generate the first canceling vibration noise estimation signal ŷ. The first virtual error signal generating means generates a first virtual error signal e2 from the vibration noise estimation signal d̂ and the first canceling vibration noise estimation signal ŷ. The first filter coefficient updating means updates the first and second adaptive notch control filters so that the first virtual error signal e2 is minimized based on the first and second reference signals r0 and r1 and the first virtual error signal e2. The filter coefficients of W0 and W1 are successively updated.

The first adaptive notch control filter W0 outputs a third control signal based on the reference sine wave signal xs, and the second adaptive notch control filter W1 outputs a fourth control signal based on the reference cosine wave signal xc. The first correction filter Ĉ0 is the first adaptive notch correction filter Ĉ0, and the second correction filter Ĉ1 is the second adaptive notch correction filter Ĉ1. The third estimated signal generating means adds a fifth correction control signal obtained by correcting the first addition signal u0 by the first adaptive notch correction filter Ĉ0, the third control signal and the fourth control signal. A sixth correction control signal obtained by correcting the second addition signal u1 obtained by the second adaptive notch correction filter C^1 is added to generate a second canceling vibration noise estimation signal y^. The second virtual error signal generating means generates a second virtual error signal e1 from the error signal e, the vibration noise estimation signal d̂, and the second canceling vibration noise estimation signal ŷ. The second filter coefficient updating means minimizes the second virtual error signal e1 based on the first control signal uc, the second control signal us, the third control signal, the fourth control signal, and the second virtual error signal e1. The filter coefficients of the first and second adaptive notch correction filters Ĉ0 and Ĉ1 are successively updated as follows.

The third correction filter Ĥ0 is the third adaptive notch correction filter Ĥ0, and the fourth correction filter Ĥ1 is the fourth adaptive notch correction filter Ĥ1. The third filter coefficient updating means performs third and fourth adaptive notch corrections based on the reference sine wave signal xs, the reference cosine wave signal xc, and the second virtual error signal e1 so as to minimize the second virtual error signal e1. The filter coefficients of the filters Ĥ0 and Ĥ1 are successively updated.

The normalization means multiplies the filter coefficients of the first and second adaptive notch correction filters by the reciprocal of the square root of the sum of squares of the filter coefficients of the first and second adaptive notch correction filters to obtain first and second normalizations. Calculate the filter coefficients. The correction means corrects the reference cosine wave signal xc and the reference sine wave signal xs by a first adaptive notch correction filter having a first normalization filter coefficient and a second adaptive notch correction filter having a second normalization filter coefficient. A first and a second reference signal r0, r1 are generated.

The normalization means converts the filter coefficients of the first and second adaptive notch correction filters Ĉ0 and Ĉ1 to the absolute values of the filter coefficients of these first and second adaptive notch correction filters Ĉ0 and Ĉ1. The reciprocal of the larger value may be multiplied to calculate the third and fourth normalized filter coefficients. In this case, the correction means is adapted to generate the reference cosine wave signal xc and the reference sine wave signal xs based on a first adaptive notch correction filter having a third normalization filter coefficient and a second adaptive notch correction filter having a fourth normalization filter coefficient. are respectively corrected to generate the first and second reference signals r0 and r1.

The first, second, and third filter coefficient updating means set the step size parameter μ for controlling the amount of update of the filter coefficients of the adaptive notch filters to be updated, based on the square root of the sum of squares of the filter coefficients immediately before updating. decide.

The first, second, and third filter coefficient updating means set the step size parameter μ for controlling the update amount of the filter coefficients of the adaptive notch filters to be updated to the larger absolute value of the filter coefficients immediately before updating. may be determined based on

Next, first to third application examples of the active vibration noise reduction device 10 according to the present invention will be described with reference to FIGS. 10 to 12. FIG. In these examples, the active vibration noise reduction device 10 is applied to the vehicle 1 .

As shown in FIG. 10, a vehicle 1 is equipped with an engine 2 as a drive source. The active vibration noise reduction device 10 includes an error microphone 11, which is a vibration noise detection unit that detects noise in the vehicle interior 3, and a canceling sound that generates a canceling sound having a phase opposite to that of the noise as a control sound for canceling the noise. It has a speaker 12 as a generating means and an active vibration noise control section 13 . The error microphones 11 are attached, for example, to the ceiling above the front seats and above the rear seats. The speakers 12 may be the speakers 12 of an audio system, which are door speakers mounted on the front and rear doors. The error microphone 11 functions as error signal detection means for detecting, as an error signal e, the offset error between the noise from the engine 2 which is the vibration noise source and the canceling sound from the speaker 12 . Vehicle information such as the engine speed and vehicle speed and the error signal e detected by the error microphone 11 are supplied to the active vibration noise control unit 13 . Based on the vehicle information and the error signal e, the active vibration noise control unit 13 generates a control signal u0 (first addition signal) for driving the speaker 12, and cancels the noise generated by the speaker 12. By controlling, the engine noise (engine roaring noise) transmitted to the occupant due to the vibration of the engine 2 is reduced. In this case, the active vibration noise control section 13 functions as an active noise control section.

The active vibration noise reduction device 10 shown in FIG. 11 includes an error microphone 11 for detecting noise in the vehicle interior 3, and an offsetting vibration of the opposite phase to the vibration of the engine 2, which is the cause of the noise, for canceling the vibration. It has a vibration actuator 14 which is a generating part for canceling vibration, and an active vibration noise control part 13 . The error microphone 11 is the same as that of the active vibration noise reduction device 10 shown in FIG. The vibration actuator 14 is configured to apply the generated offset vibration to the engine 2, and is configured by, for example, an active engine mount. Vehicle information such as the engine speed and vehicle speed and the error signal e detected by the error microphone 11 are supplied to the active vibration noise control unit 13 . Based on the vehicle information and the error signal e, the active vibration noise control unit 13 generates a control signal u0 for driving the vibration actuator 14, and controls the offset vibration generated in the vibration actuator 14. , reduce the vibration of the engine 2, and reduce the engine noise (engine roaring noise) transmitted to the occupant due to the engine vibration. In this case, the active vibration noise control section 13 functions as an active vibration control section.

An active vibration noise reduction device 10 shown in FIG. It has a vibration actuator 14 that generates vibration and an active vibration noise control section 13 . The vibration sensor 15 is attached to the engine 2, and is an error signal detector that detects, as an error signal e, an error vibration that is a combination of the engine vibration generated by the rotation of the engine 2 and the offset vibration given to the engine 2 by the vibration actuator 14. act as a means. The vibration actuator 14 is similar to that of the active vibration noise reduction device 10 shown in FIG. Vehicle information such as engine speed and vehicle speed and an error signal e detected by the vibration sensor 15 are supplied to the active vibration noise control unit 13 . Based on the vehicle information and the error signal e, the active vibration noise control unit 13 generates a control signal u0 for driving the vibration actuator 14, and controls the offset vibration generated in the vibration actuator 14. , reduce engine vibration, and reduce engine noise (engine roaring noise) transmitted to the occupant due to the vibration of the engine 2 . Also in this case, the active vibration noise control section 13 functions as an active vibration control section.

Thus, the active vibration noise reduction device 10 according to the present invention can be applied in various aspects. Other than these examples, for example, a motor is mounted as a drive source instead of the engine 2, and the active vibration noise reduction device 10 is configured to reduce the vibration noise of the motor that is the source of the vibration noise. good too. Alternatively, the active vibration noise reduction device 10 may be configured to reduce drive system noise transmitted to the occupant due to vibration noise of drive system rotating bodies such as a propeller shaft and a drive shaft when the vehicle 1 is running. good. That is, the active vibration noise reduction device 10 reduces the vibration noise of the engine 2 or drive system that generates periodic narrow-band vibration noise due to the rotational motion of the rotating body.

In each embodiment described below, the vehicle 1 includes an engine 2 as a drive source, the active vibration noise reduction device 10 includes an error microphone 11 as a vibration noise detection unit, a speaker 12 as canceling sound generation means, It is assumed that the active vibration noise control section 13 functions as an active noise control section.

<<First embodiment>>
First, a first embodiment of the present invention will be described with reference to FIGS. 13 to 15. FIG. FIG. 13 is a functional block diagram of the active vibration noise reduction device 10 according to the first embodiment. As shown in FIG. 13, an engine/drive system signal X is supplied to the active vibration noise control section 13 . The engine/drive system signal X may be an engine pulse synchronized with a vibration frequency such as the rotation frequency of the output shaft of the engine 2, or a drive system rotation pulse that transmits the driving force of the engine 2 to the wheels. The engine/drive system signal X is not limited to this. Any operation-related information relating to the operation of a drive source or drive system that is a source of vibration noise may be used. The active vibration noise control section 13 includes a reference signal generation section 21 that generates a reference signal x (xc, xs) based on the engine/drive system signal X. FIG.

In the reference signal generator 21 , the frequency estimating circuit 22 estimates the frequency f of the vibration noise d that becomes noise in the vehicle interior 3 from the engine/driving system signal X. Specifically, the frequency estimation circuit 22 estimates the frequency f of the vibration noise d based on the engine/drive system signal X by referring to a map or the like. The estimated frequency f is supplied to the cosine wave generator circuit 23 and the sine wave generator circuit 24 . The cosine wave generating circuit 23 generates a reference cosine wave signal xc, which is a reference signal x synchronized with the vibration noise d generated from the engine 2 and the drive system due to the rotation of the engine 2, based on the supplied frequency f. do. The sine wave generating circuit 24 generates a reference sine wave signal xs, which is a reference signal x synchronized with the vibration noise d, based on the supplied frequency f. That is, the reference signal generation unit 21 does not generate the reference signal x (xc, xs) by detecting the frequency f from the physical quantity of the vibration noise d detected by the microphone or the vibration sensor, but rather A reference signal x(xc, xs) having a frequency f of the vibration noise d estimated based on the operation-related information is generated. The reference signal x (xc, xs) generated by the reference signal generation unit 21 is supplied to the control signal generation unit 25, the reference signal correction unit 26, the reference signal generation unit 27 (correction means), and the vibration noise estimation signal generation unit 28 (second 1 estimated signal generating means).

The control signal generator 25 is a notch filter that generates a control signal u0 by filtering the reference signal x(xc, xs), and has an adaptive notch filter coefficient W expressed using one complex number. there is This adaptive notch filter coefficient W indicates the circuit characteristics of the control signal generator 25 . The control signal generator 25 generates a first adaptive notch control filter 31 having a first adaptive notch filter coefficient W0 that is the real part of the adaptive notch filter coefficient W, a second adaptive notch filter coefficient that is the imaginary part of the adaptive notch filter coefficient W, and It has a second adaptive notch control filter 32 with W1 and an adder 33 . The reference cosine wave signal xc is fed to a first adaptive notch control filter 31 and filtered using the first adaptive notch filter coefficient W0, and the reference sine wave signal xs is fed to a second adaptive notch control filter 32 to be filtered by a second Filtered using adaptive notch filter coefficients W1. The first control signal uc output from the first adaptive notch control filter 31 and the second control signal us output from the second adaptive notch control filter 32 are added by the adder 33 to become the control signal u0. . The control signal generation unit 25 constitutes a part of the reference signal correction unit 26, and the reference signal x (xc, xs) is corrected by the circuit characteristics (adaptive notch filter coefficient W) of the control signal generation unit 25 to generate the control signal u0 (first addition signal).

The reference signal correction unit 26 includes the above-described control signal generation unit 25, and further includes a first adaptive notch filter 34 having coefficients obtained by inverting the polarity of the first adaptive notch filter coefficient W0, and a second adaptive notch filter coefficient. A second adaptive notch filter 35 with W 1 and an adaptive notch filter with an adder 36 . The reference cosine wave signal xc is supplied to the first adaptive notch filter 34 and filtered using the polarity-inverted value of the first adaptive notch filter coefficient W0. The reference sinusoidal signal xs is applied to a second adaptive notch filter 35 and filtered using a second adaptive notch filter coefficient W1 . The third control signal output from the first adaptive notch filter 34 and the fourth control signal output from the second adaptive notch filter 35 are added by the adder 36 so that the reference signal x(xc, xs) is The control signal u1 (second addition signal) corrected by the circuit characteristics (adaptive notch filter coefficient W) of the control signal generator 25 is obtained.

The control signal u<b>0 output from the control signal generator 25 is converted into an analog signal by the D/A converter 37 and supplied to the speaker 12 . Based on the supplied control signal u0, the speaker 12 generates a canceling sound (control sound) for canceling noise generated from the engine 2 and the driving system, which are noise sources.

In the reference signal generator 27, a canceling sound transfer characteristic estimated value Ĉ, which is an estimated value of the acoustic characteristic C of the canceling sound from the speaker 12 to the error microphone 11, is set. The canceling sound transfer characteristic estimated value Ĉ is a value provided by the first adaptive notch correction filter section 60, which will be described later. The canceling sound transfer characteristic estimated value C^ is based on a function that sets the transfer characteristic (amplitude characteristic and phase characteristic) from the speaker 12 to the error microphone 11, and uses one complex number obtained for the frequency f of the canceling sound. and has a real part Ĉ0 (first correction filter coefficient) and an imaginary part Ĉ1 (second correction filter coefficient).

In the reference signal generator 27, the reference cosine wave signal xc is input to the first correction filter 41 having the real part Ĉ0 of the canceling sound transfer characteristic estimated value Ĉ as a coefficient. The reference sine wave signal xs is input to a second correction filter 42 having, as a coefficient, the imaginary part Ĉ1 of the canceling sound transfer characteristic estimated value Ĉ. Further, the reference sine wave signal xs is input to the first correction filter 43 having the real part Ĉ0 of the canceling sound transfer characteristic estimated value Ĉ as a coefficient. The reference cosine wave signal xc is input to a second correction filter 44 having coefficients obtained by inverting the polarity of the imaginary part Ĉ1 of the canceling sound transfer characteristic estimated value Ĉ.

The reference cosine wave signal xc is filtered in a first correction filter 41 using the real part C^0 of the canceling sound transfer characteristic estimate C^. The reference sinusoidal signal xs is filtered in a second correction filter 42 using the imaginary part Ĉ1 of the canceling sound transfer characteristic estimate Ĉ. The output of the first correction filter 41 and the output of the second correction filter 42 are added by the adder 45, so that the reference signal x(xc, xs) is corrected by the canceling sound transfer characteristic estimated value Ĉ to obtain the first It becomes the reference signal r0. The reference cosine wave signal xc is also filtered in the first correction filter 43 using the real part Ĉ0 of the canceling sound transfer characteristic estimated value Ĉ. The reference sinusoidal signal xs is filtered in the second correction filter 44 using a value obtained by inverting the polarity of the imaginary part C^1 of the canceling sound transfer characteristic estimated value C^. The output of the first correction filter 43 and the output of the second correction filter 44 are added by the adder 46, so that the reference signal x(xc, xs) is corrected by the canceling sound transfer characteristic estimated value Ĉ. 2 becomes the reference signal r1.

The vibration noise estimation signal generator 28 is a so-called SAN filter (Single frequency Adaptive Notch filter). In the vibration noise estimation signal generation unit 28, the transfer characteristic estimated value H^ that is an estimated value of the transfer characteristic H of the noise (that is, the noise propagation path) from the engine 2 and the drive train that are the noise source to the error microphone 11 is generated. A small value such as 0 is set in advance as an initial value. The transfer characteristic estimated value H^ is expressed using one complex number obtained for the frequency f of the vibration noise d based on a function that sets the transfer characteristic (amplitude characteristic and phase characteristic) from the noise source to the error microphone 11. and has a real part H^0 (third correction filter coefficient (third adaptive notch correction filter coefficient)) and an imaginary part H^1 (fourth correction filter coefficient (fourth adaptive notch correction filter coefficient)) there is The transfer characteristic estimated value H^ is not a physical quantity obtained by directly measuring the vibration frequency of the noise source, but is generated from the reference signal x generated based on the operation-related information of the engine 2 and drive system.

In the vibration noise estimation signal generation unit 28, the reference cosine wave signal xc is generated by the third adaptive notch correction filter 51 having the coefficient of the real part H^0 of the transfer characteristic estimated value H^ and the filters of the third adaptive notch correction filter 51. It is input to a filter coefficient updating unit 52 (third filter coefficient updating means) that adaptively updates coefficients. The reference sine wave signal xs includes a fourth adaptive notch correction filter 53 having the imaginary part H^1 of the transfer characteristic estimated value H^ as a coefficient, and a filter coefficient for adaptively updating the filter coefficient of the fourth adaptive notch correction filter 53. It is input to the updating unit 54 (third filter coefficient updating means). The third adaptive notch correction filter 51 and the fourth adaptive notch correction filter 53 correspond to the frequency of the reference signal x. It is a correction filter corresponding to the transfer characteristic and an adaptive notch correction filter whose filter coefficients are adaptively updated. The filter coefficient updating unit 52 and the filter coefficient updating unit 54 will be described later in detail.

The reference cosine signal xc is filtered in a third adaptive notch correction filter 51 using the real part Ĥ0 of the transfer characteristic estimate Ĥ. The reference sinusoidal signal xs is filtered in a fourth adaptive notch correction filter 53 using the imaginary part Ĥ1 of the transfer characteristic estimate Ĥ. The first vibration noise estimation signal output from the third adaptive notch correction filter 51 and the second vibration noise estimation signal output from the fourth adaptive notch correction filter 53 are added by the adder 55 and reach the error microphone 11. Vibration noise estimation signal d^, which is an estimated value of vibration noise d, is obtained. That is, the vibration noise estimation signal generator 28 generates the vibration noise estimation signal d^ in the error microphone 11 based on the reference signal x(xc, xs).

The control signal u0 and the control signal u1 output from the reference signal correction section 26 are supplied to the first adaptive notch correction filter section 60 (third estimation signal generation means). The first adaptive notch correction filter unit 60 is an SAN filter, and a small value such as 0 is preset in the first adaptive notch correction filter unit 60 as an initial value of the canceling sound transfer characteristic estimated value C^. . In the first adaptive notch correction filter unit 60, the control signal u0 is a fifth adaptive notch correction filter 61 having the real part C^0 of the canceling sound transfer characteristic estimated value C^ as a coefficient, and It is input to a filter coefficient updating unit 62 (second filter coefficient updating means) that adaptively updates the filter coefficients. The control signal u1 is a sixth adaptive notch correction filter 63 having the imaginary part Ĉ1 of the canceling sound transfer characteristic estimated value Ĉ as a coefficient, and a filter coefficient for adaptively updating the filter coefficient of the sixth adaptive notch correction filter 63. It is input to the updating section 64 (second filter coefficient updating means). The fifth adaptive notch correction filter 61 and the sixth adaptive notch correction filter 63 are correction filters and adaptive notch correction filters whose filter coefficients are adaptively updated. The filter coefficient updating unit 62 and the filter coefficient updating unit 64 will be described later in detail.

The control signal u0 is filtered in the fifth adaptive notch correction filter 61 using the real part C^0 of the canceling sound transfer characteristic estimate C^. The control signal u1 is filtered in the sixth adaptive notch correction filter 63 using the imaginary part Ĉ1 of the canceling sound transfer characteristic estimated value Ĉ. The first correction control signal output from the fifth adaptive notch correction filter 61 and the second correction control signal output from the sixth adaptive notch correction filter 63 are added by the adder 65 and reach the error microphone 11. It becomes the first estimated value y^1 of the canceling vibration noise y (second canceling vibration noise estimation signal). That is, the first adaptive notch correction filter unit 60 generates the first estimated value y^1 of the canceling sound reaching the error microphone 11 based on the control signal u0 and the control signal u1.

The first and second reference signals r0 and r1 output from the reference signal generation section 27 are supplied to the adaptive notch control filter section 70 (second estimation signal generation means). The adaptive notch control filter unit 70 is a SAN filter, and the adaptive notch control filter unit 70 has an initial value of adaptive notch filter coefficients W (W0, W1) indicating the circuit characteristics of the control signal generation unit 25, for example 0. is preset to a small value. In the adaptive notch control filter unit 70, the first reference signal r0 is a third adaptive notch control filter 71 having a first adaptive notch filter coefficient W0 forming the real part of the adaptive notch filter coefficient W, and the third adaptive notch control filter It is input to a filter coefficient updating unit 72 (first filter coefficient updating means) that adaptively updates the filter coefficient of 71 . The second reference signal r1 adaptively updates the fourth adaptive notch control filter 73 having the second adaptive notch filter coefficient W1 forming the real part of the adaptive notch filter coefficient W and the filter coefficients of the fourth adaptive notch control filter 73. It is input to the filter coefficient updating unit 74 (first filter coefficient updating means). The filter coefficient updating unit 72 and the filter coefficient updating unit 74 will be described later in detail.

The first reference signal r0 is filtered in the third adaptive notch control filter 71 using the first adaptive notch filter coefficient W0. The second reference signal r1 is filtered in the fourth adaptive notch control filter 73 using the second adaptive notch filter coefficient W1. The output of the third adaptive notch control filter 71 and the output of the fourth adaptive notch control filter 73 are added by an adder 75 to obtain a second estimated value y^2 of the canceling vibration noise y in the error microphone 11 (first canceling vibration noise estimation signal). That is, the adaptive notch control filter unit 70 generates the second estimated value y^2 of the canceling sound reaching the error microphone 11 based on the first and second reference signals r0 and r1.

Adaptive notch filter coefficients W (W 0 , W 1 ) adaptively updated in adaptive notch control filter section 70 are provided to control signal generating section 25 . That is, the adaptive notch filter coefficients W (W0, W1) set in the control signal generation unit 25 are not fixed values, but the same values as those sequentially updated by the filter coefficient updating unit 72 and the filter coefficient updating unit 74 are adapted. The real part W0 and the imaginary part W1 of the notch filter coefficient W are adaptively set.

The error microphone 11 is generated by the noise in the vehicle interior 3, that is, the vibration noise d having a certain frequency f, which is mainly generated by the engine 2 and the drive system and reaches the error microphone 11, and the error microphone 11 generated by the speaker 12. A noise that is a cancellation error synthesized with the canceling vibration noise y arriving at is detected as an error signal e. It should be noted that the noise detected by the error microphone 11 includes not only the noise resulting from the canceling error, but also the noise other than the engine 2 and the driving system. The error signal e is converted into a digital signal by the A/D converter 76 and supplied to the virtual error signal generator 80 .

The vibration noise estimation signal d̂ in the error microphone 11 output from the vibration noise estimation signal generation section 28 is also supplied to the virtual error signal generation section 80 . Also, the first estimated value y^1 and the second estimated value y^2 of the canceling vibration noise y reaching the error microphone 11, which are output from the first adaptive notch correction filter section 60 and the adaptive notch control filter section 70, are also , is supplied to the virtual error signal generator 80 .

The virtual error signal generator 80 generates an apparent virtual error signal e′ (second virtual error signal e′1 and first virtual error signal e '2) is generated. Specifically, the virtual error signal generator 80 includes a second virtual error signal generator 81 that generates the second virtual error signal e′1 and a first virtual error signal generator 81 that generates the first virtual error signal e′2. and a generator 82 .

In the second virtual error signal generator 81 , the error signal e is supplied to the adder 83 . Also, the vibration noise estimation signal d̂ in the error microphone 11 is supplied to the adder 83 after the polarity is inverted by the first polarity inverting circuit 84 . Further, the first estimated value ŷ1 of the canceling vibration noise y is supplied to the adder 83 after being inverted in polarity by the second polarity inverting circuit 85 . The adder 83 generates a second virtual error signal e'1 by adding the three supplied values. The second virtual error signal e′1 is supplied to the vibration noise estimation signal generation section 28 and the first adaptive notch correction filter section 60 .

In the first virtual error signal generator 82 , the vibration noise estimation signal d̂ in the error microphone 11 is supplied to the adder 86 . A second estimate ŷ2 of the canceling vibratory noise y is also provided to the adder 86 . Adder 86 generates a first virtual error signal e'2 by adding the two supplied values. The first virtual error signal e′ 2 is supplied to the adaptive notch control filter section 70 .

ここで、ｒ：参照信号（基準余弦波信号ｘｃ、基準正弦波信号ｘｓで構成）、＊：フィルタリング計算（ＳＡＮフィルタでは複素数の掛け算に相当）、ｎ：サンプリング時刻、である。 The second virtual error signal e'1 and the first virtual error signal e'2 generated in the virtual error signal generator 80 can be expressed by the following equations.

Here, r: reference signal (consisting of reference cosine wave signal xc and reference sine wave signal xs), *: filtering calculation (corresponding to multiplication of complex numbers in SAN filter), n: sampling time.

In the vibration noise estimation signal generation unit 28, the filter coefficient updating unit 52 uses the reference cosine wave signal xc and the second virtual error signal e′1 to minimize the second virtual error signal e′1 using the LMS algorithm. The filter coefficient (Ĥ0) of the third adaptive notch correction filter 51 is calculated so that The filter coefficient updating unit 52 performs coefficient calculation for the third adaptive notch correction filter 51 at each sampling time, and updates the filter coefficient (Ĥ0) of the third adaptive notch correction filter 51 to the calculated value. In addition, the filter coefficient updating unit 54 uses the reference sine wave signal xs and the second virtual error signal e′1 to perform the fourth adaptation so that the second virtual error signal e′1 is minimized using the LMS algorithm. A filter coefficient (Ĥ1) of the notch correction filter 53 is calculated. The filter coefficient updating unit 54 performs coefficient calculation for the fourth adaptive notch correction filter 53 at each sampling time, and updates the filter coefficient (Ĥ1) of the fourth adaptive notch correction filter 53 to the calculated value. That is, the vibration noise estimation signal generator 28 serves as an updating unit that updates the transfer characteristic estimation value Ĥ.

In the first adaptive notch correction filter unit 60, the filter coefficient updating unit 62 uses the control signal u0 and the second virtual error signal e′1 to minimize the second virtual error signal e′1 using the LMS algorithm. The filter coefficient (Ĉ0) of the fifth adaptive notch correction filter 61 is calculated as follows. The filter coefficient updating unit 62 performs coefficient calculation for the fifth adaptive notch correction filter 61 at each sampling time, and updates the filter coefficient (Ĉ0) of the fifth adaptive notch correction filter 61 to the calculated value. Further, the filter coefficient updating unit 64 uses the control signal u1 and the second virtual error signal e′1 to perform sixth adaptive notch correction so that the second virtual error signal e′1 is minimized using the LMS algorithm. A filter coefficient (Ĉ1) of the filter 63 is calculated. The filter coefficient updating unit 64 performs coefficient calculation for the sixth adaptive notch correction filter 63 at each sampling time, and updates the filter coefficient (Ĉ1) of the sixth adaptive notch correction filter 63 to the calculated value. That is, the first adaptive notch correction filter unit 60 serves as an updating unit that updates the canceling sound transfer characteristic estimated value Ĉ.

In the adaptive notch control filter unit 70, the filter coefficient updating unit 72 uses the first reference signal r0 and the first virtual error signal e′2 to minimize the first virtual error signal e′2 using the LMS algorithm. The first adaptive notch filter coefficient W0 of the third adaptive notch control filter 71 is calculated so that The filter coefficient updating unit 72 performs coefficient calculation for the third adaptive notch control filter 71 at each sampling time, and updates the first adaptive notch filter coefficient W0 of the third adaptive notch control filter 71 to the calculated value. In addition, the filter coefficient updating unit 74 uses the second reference signal r1 and the first virtual error signal e′2 to perform the fourth adaptation so that the first virtual error signal e′2 is minimized using the LMS algorithm. A second adaptive notch filter coefficient W1 of the notch control filter 73 is calculated. The filter coefficient updating unit 74 performs coefficient calculation of the fourth adaptive notch control filter 73 at each sampling time, and updates the second adaptive notch filter coefficient W1 of the fourth adaptive notch control filter 73 to the calculated value. That is, the adaptive notch control filter section 70 serves as an updating section that updates the adaptive notch filter coefficient W representing the circuit characteristics of the control signal generating section 25 .

The first adaptive notch filter coefficient W0 and the second adaptive notch filter coefficient W1 updated by the adaptive notch control filter unit 70 are provided to the control signal generation unit 25 as described above, and the first adaptive notch control filter 31 The first adaptive notch filter coefficient W0 and the second adaptive notch filter coefficient W1 of the second adaptive notch control filter 32 are successively updated.

As a result, the reference cosine wave signal xc and the reference sine wave signal xs filtered by the control signal generator 25 are optimized. Vibrational noise d, which is the periodic noise of , is canceled, and indoor noise is reduced.

ここで、μ：それぞれの適応フィルタ係数の更新量を調整するためのステップサイズパラメータ、である。 The filter coefficients (H^, C^, W) of these adaptive notch filters (28, 60, 70) are obtained by the LMS algorithm as follows using the virtual error signal e'(e'1,e'2): Updated.

where μ is a step size parameter for adjusting the update amount of each adaptive filter coefficient.

上式（１３）より、下式（１４）が導出される。

また、上式（１４）及び上式（１２）より、下式（１５）が導出される。

ここで、／：複素数の割り算、である。
上式（１４）及び上式（１５）を連立すると、下式（１６）となる。

誤差マイク１１の位置における音圧を示す誤差信号ｅは、下式で表される。
ｅ_ｎ＝ｄ_ｎ＋ｙ_ｎ＝ｒ_ｎ＊Ｈ_ｎ＋ｒ_ｎ＊Ｗ_ｎ＊Ｃ_ｎ
この式に上式（１６）を代入すると、ｅ＝０であることがわかる。 When the second virtual error signal e'1 and the first virtual error signal e'2 converge to the minimum value (0) by the above adaptive updating, the following simultaneous equations are established.

The following formula (14) is derived from the above formula (13).

Further, the following formula (15) is derived from the above formulas (14) and (12).

where /: division of complex numbers.
Simultaneously combining the above equations (14) and (15) yields the following equation (16).

An error signal e indicating the sound pressure at the position of the error microphone 11 is represented by the following equation.
e _n =d _n +y _n =r _n *H _n +r _n *W _n *C _n
Substituting the above equation (16) into this equation reveals that e=0.

Therefore, according to the active vibration noise reduction device 10, even if the true values of the transfer characteristic estimated value H^ and the canceling sound transfer characteristic estimated value C^ are unknown, the second virtual error signal e′1 and the first When the virtual error signal e'2 converges to 0, the ratio of the transfer characteristic estimated value H^ and the canceling sound transfer characteristic estimated value C^ converges to a constant value, and the filter coefficient is supplied to the control signal generator 25, which is a control filter. The adaptive notch filter coefficient W of the adaptive notch control filter unit 70 to be provided is also guaranteed to converge to the optimum value -H/C, and the sound pressure (error signal e) in the error microphone 11 is minimized. This is because the active vibration noise reduction device 10 does not require prior identification of the transfer characteristics (acoustic characteristics C) of the canceling sound from the speaker 12 to the error microphone 11, and the acoustic characteristics C of the canceling sound do not change during control. It means that it operates on the principle that even if it occurs, it can be silenced.

Next, functions and effects confirmed for the active vibration noise reduction device 10 according to the embodiment will be described. FIG. 14 is a graph showing changes in assumed acoustic characteristics C in the active vibration noise reduction device 10 shown in FIG. As shown in FIG. 14, in the frequency band (100 Hz to 150 Hz) corresponding to the engine speed of 3000 to 4500 RPM, the acoustic characteristic C changes from the initial characteristic indicated by the solid line to the current characteristic indicated by the broken line. It is assumed that there is a difference between a certain canceling sound transfer characteristic estimated value Ĉ and the actual acoustic characteristic C.

Under such conditions, when the active vibration noise control unit 13 according to the embodiment executes noise reduction control, the sound pressure level of the engine slamming sound is reduced as shown in FIG. FIG. 15 shows the sound pressure levels when the control is turned off, when the stability improvement control is performed by the conventional method using the stabilization coefficient α, and when the control is performed according to the first embodiment of the present invention. As shown in FIG. 15, in the engine speed range of 3000 to 4500RPM where the actual acoustic characteristic C changes, the control performance of the conventional example deteriorates greatly, and around 3800RPM, a noise increase of about 15dB occurs. there is On the other hand, in the present invention, it is possible to follow changes in the actual acoustic characteristics C during control, and even if the actual acoustic characteristics C change greatly, a large deterioration in performance does not occur, and noise is reduced by about 10 dB. is made of. In the area where the acoustic characteristic C does not change, the performance equivalent to that of the conventional example is realized. As for the initial convergence, the present invention is slower than the conventional example, but the convergence time is very short.

As described above, in the active vibration noise control unit 13, the reference signal generation unit 21 controls the engine 2/drive system based on the engine/drive system signal X, which is information related to the operation of the engine 2/drive system that is the source of the vibration noise. estimates the frequency f of the vibration noise d emitted by and generates a reference signal x(xc, xs) synchronized with the vibration noise d. Also, the virtual error signal generator 80 generates a virtual error signal e′ using the error signal e and the vibration noise estimation signal d reaching the error microphone 11 . Then, the vibration noise estimation signal generator 28 successively updates the filter coefficients using an adaptive algorithm using the virtual error signal e'. Therefore, the active vibration noise control unit 13 must accurately identify the transfer characteristic H of the vibration noise d transmitted from the vibration noise source without arranging a microphone or a vibration sensor for detecting the reference signal x. Also, when the transfer characteristic H of the vibration noise d changes, the noise can be reduced by the canceling vibration noise y. In addition, since no microphone or vibration sensor is required, the configuration of the active vibration noise control unit 13 is simplified, and noise is not mixed into the reference signal x (xc, xs), so good noise reduction performance is achieved. It is feasible.

Further, in this embodiment, the vibration noise estimation signal generator 28 is not an FIR filter but an SAN filter, and successively updates the filter coefficients using an adaptive algorithm. Therefore, the noise reduction performance is secured even for the vibration noise d whose characteristics are constantly changing, and the amount of calculation is small, and an expensive processor with high processing performance is not required, so the noise reduction performance is good and the active vibration noise It is possible to construct the reduction device 10 at low cost.

In the active vibration noise control unit 13, the reference signal correction unit 26 corrects the reference signal x (xc, xs) with the adaptive notch filter coefficient W indicating the circuit characteristics of the control signal generation unit 25, and the control signal u0 is The first adaptive notch correction filter unit 60 corrects the control signal u0 with the canceling sound transfer characteristic estimated value Ĉ to generate the first estimated value ŷ1 of the canceling vibratory noise y in the error microphone 11. Further, the reference signal generator 27 corrects the reference signal x (xc, xs) with the canceling sound transfer characteristic estimated value Ĉ to generate the reference signal r (r0, r1) and provides the reference signal r (r0, r1) to the control signal generator 25 . An adaptive notch control filter unit 70 having adaptive notch filter coefficients W (W0, W1) corrects the reference signal r with the adaptive notch filter coefficients W to obtain a second estimate y of the canceling vibration noise y in the error microphone 11. Generate ^2. Then, the virtual error signal generation unit 80 further uses the first estimated value y^1 and the second estimated value y^2 of the canceling vibration noise y to generate the virtual error signal e′, and the first adaptive notch correction filter unit 60 and adaptive notch control filter unit 70 successively update the corresponding filter coefficients using an adaptive algorithm using the virtual error signal e'.

Specifically, in the virtual error signal generation unit 80, the second virtual error signal generation unit 81 generates the first A second virtual error signal e′1 is generated, and a first virtual error signal generator 82 generates a first virtual error signal Generate e'2. Then, the vibration noise estimation signal generator 28 updates the filter coefficient based on the reference signal x(xc, xs) and the second virtual error signal e′1, and the first adaptive notch correction filter 60 updates the control signal u0 and the second The filter coefficients are updated based on the second virtual error signal e'1, and the adaptive notch control filter unit 70 updates the filter coefficients based on the reference signal r(r0, r1) and the first virtual error signal e'2.

Therefore, even if a large change occurs in the transfer characteristic (acoustic characteristic C) of the canceling sound from the speaker 12 to the error microphone 11 during the control, the three adaptive notch filters (28, 60, 70) generate the virtual error signal e'. Good noise reduction performance is achieved by successively updating the filter coefficients using an adaptive algorithm using . That is, by performing noise reduction control using the above-described control method in which the active vibration noise control unit 13 adaptively updates the coefficients of the SAN filter according to the virtual error signal e', the acoustic characteristic C can be identified without prior identification, and the acoustic characteristic It is possible to realize an active vibration noise reduction device 10 that exhibits good noise reduction performance by following changes in the acoustic characteristics C during control even if C changes significantly. In addition, since the active vibration noise control unit 13 uses an adaptive notch filter composed of a SAN filter instead of an FIR filter, the amount of calculation can be reduced, and a high-performance processor is not required. A device 10 is realized.

Further, in the present embodiment, even if the acoustic characteristics C are greatly changed by adjusting the position or angle of the seat, noise reduction control is possible without deteriorating the noise reduction performance. It is possible to place it on the headrest of a car, etc., and it is possible to greatly improve the silencing effect around the passenger's ears.

Since the noise source is the engine 2 which is the drive source of the vehicle 1 or a rotating body included in the drive system, the frequency f of the vibration noise d is narrow band, and the active vibration noise reduction device 10 reduces the vibration noise d. can be reliably reduced.

<<Second embodiment>>
Next, a second embodiment of the present invention will be described with reference to FIGS. 16 to 18. FIG. In addition, the same code|symbol is attached|subjected to the same or similar element as 1st Embodiment, and the overlapping description is abbreviate|omitted.

FIG. 16 is a functional block diagram of the active vibration noise reduction device 10 according to the second embodiment. As shown in FIG. 16, the active vibration noise reduction device 10 of this embodiment differs from that of the first embodiment in that it further includes a phase extractor 90 . A specific description will be given below.

In the control method performed by the active vibration noise control unit 13 of the first embodiment, the first adaptive notch correction filter unit 60 uses an estimated value (cancellation Since it corresponds to the sound transfer characteristic estimated value C^), the magnitude of the filter coefficient changes depending on the frequency f. When the canceling sound transfer characteristic estimated value Ĉ is small, the first and second correction reference signals r0 and r1 used for updating the adaptive notch control filter section 70 that provides the filter coefficients to the control signal generation section 25 are small , The convergence of the response notch control filter section 70 is delayed. Furthermore, since the output of the reference signal correction unit 26 including the control signal generation unit 25 having the adaptive notch filter coefficient W is also used for updating the first adaptive notch correction filter unit 60, the first adaptive notch correction filter The convergence of the section 60 itself is also slowed down. On the other hand, in the frequency band where the canceling sound transfer characteristic estimated value C^ is large, the convergence of the adaptive notch control filter unit 70 and the first adaptive notch correction filter unit 60 is fast, but the amount of update each time is large, so that it becomes unstable. tends to be easy.

Therefore, in order to improve the convergence performance of the control method of the first embodiment, the active vibration noise control unit 13 of the present embodiment provides a canceling sound transfer characteristic estimation value that does not depend on the magnitude of the canceling sound transfer characteristic estimated value C^. A phase extractor 90 is further provided to update the filter coefficients using the phase information of the value Ĉ.

ここで、「｜｜」は複素数の振幅を表す。また、計算量軽減のために、打消音伝達特性推定値Ｃ＾の振幅の代わりに、打消音伝達特性推定値Ｃ＾の実部Ｃ＾０と虚部Ｃ＾１とのうち絶対値の大きい方が用いられてもよい。

As shown in the following equations, the vibration noise estimation signal generation unit 28, the first adaptive notch correction filter unit 60, and the adaptive notch control filter unit 70 use the same equations as in the first embodiment to obtain filter coefficients (H^, C^, W), the phase extractor 90 normalizes the canceling sound transfer characteristic estimated value Ĉ. That is, the phase extractor 90 obtains the real part C^ of the canceling sound transfer characteristic estimated value C^ by obtaining the reciprocal of the square root sum of the real part C^0 and the imaginary part C^1 of the canceling sound transfer characteristic estimated value C^. 0 and the imaginary part Ĉ1 are multiplied to calculate the first and second normalization filter coefficients, respectively.

Here, "||" represents the amplitude of the complex number. In order to reduce the amount of calculation, instead of the amplitude of the canceling sound transfer characteristic estimated value C^, the real part C^0 and the imaginary part C^1 of the canceling sound transfer characteristic estimated value may be used.

The reference signal generation unit 27 corrects the reference signal x (xc, xs) using the canceling sound transfer characteristic estimated value C ^ normalized by the phase extraction unit 90 using the above equation to generate the reference signal r (r0, r1 ), and the adaptive notch control filter unit 70 uses this reference signal r (r0, r1) to generate a second estimated value y^2 of the canceling vibratory noise y. Also, when the first adaptive notch correction filter unit 60 updates the canceling sound transfer characteristic estimated value C^ in the next sample, the canceling sound transfer characteristic estimated value C^ normalized by the phase extracting unit 90 is used for the update. I do.

By performing such control by the active vibration noise control unit 13, the active vibration noise reduction device 10 of the second embodiment exhibits a higher noise reduction performance than that of the first embodiment. Specifically, the first adaptive notch correction filter unit 60 uses the phase information of the first adaptive notch correction filter unit 60 corresponding to the estimated value of the transfer characteristic (acoustic characteristic C) of the canceling sound from the speaker 12 to the error microphone 11. The convergence performance is improved compared to the control method of the first embodiment by the control of updating the filter coefficients by using . Details of the noise reduction performance will be described later.

Since each adaptive notch filter (28, 60, 70) adaptively updates its respective filter coefficients (H^, C^, W) starting from a preset initial value (a small number, such as 0), To quickly converge from the initial value to the optimum value, it is necessary to increase the amount of update each time. For this purpose, the step size parameter μ should be set large. However, when the step size parameter μ is set large, the adaptation process tends to become unstable, and there is a trade-off between convergence speed and stability.

Therefore, each adaptive notch filter (28, 60, 70) changes the step size parameter μ according to the size of the filter coefficient in order to improve the initial convergence speed compared to the control method of the first embodiment. good. Each adaptive notch filter (28, 60, 70) updates the filter coefficients (Ĥ, Ĉ, W) using the same equations as in the first embodiment, but the respective update equations are Add a calculation that multiplies the step size parameter μ by the reciprocal of the filter amplitude.

By multiplying the step size parameter μ by the reciprocal of the filter amplitude, each step size parameter μ is increased at the beginning of the adaptation process, and the convergence speed is increased. When the filter coefficients (Ĥ, Ĉ, W) of each adaptive notch filter (28, 60, 70) converge, the step size parameter μ also decreases and converges to a constant value. This improves the initial convergence of the adaptation process without compromising stability.

Also, for computational complexity reduction, each adaptive notch filter (28, 60, 70) uses the first adaptive notch filter coefficient W0 ( real part) and the second adaptive notch filter coefficient W1 (imaginary part), the real part H^0 and the imaginary part H^1 of the transfer characteristic estimate H^, and the real part C^ of the canceling sound transfer characteristic estimate C^ The larger absolute value of 0 and the imaginary part Ĉ1 may be used.

同様に、第１適応ノッチ補正フィルタ部６０は更新用のステップサイズパラメータμ_Ｃに対し、適応ノッチ制御フィルタ部７０は更新用のステップサイズパラメータμ_Ｗに対し、それぞれ上記式のように最大値を制限する。 Additionally, each adaptive notch filter (28, 60, 70) may limit the maximum value of the step-size parameter μ to improve initial convergence speed while ensuring minimum stability. Taking the calculation of the update step size parameter _μH of the vibration noise estimation signal generator 28 as an example, the following equation is obtained.

Similarly, the first adaptive notch correction filter unit 60 and the adaptive notch control filter unit 70 set the update step size parameter _μC and the update step size parameter _μW , respectively, to the maximum value limit.

By the active vibration noise control unit 13 performing control with the step size parameter μ variable in this way, the active vibration noise reduction device 10 of the second embodiment can be compared with the first embodiment with the step size parameter μ fixed. Demonstrates higher noise reduction performance compared to the case of

Next, functions and effects confirmed with respect to the active vibration noise control unit 13 according to the present embodiment will be described. As in the first embodiment, it is assumed that the acoustic characteristic C shown in FIG. 14 changes. FIG. 17 is a graph showing the sound pressure level of the engine stuffy sound in this case. FIG. 17 shows sound pressure levels under the control off, the control of the first embodiment, and the control of the second embodiment (fixing the step size parameter μ). As shown in FIG. 17, in the engine speed range of 3000 to 4500 RPM where the acoustic characteristic C changes, the active vibration noise reduction device 10 of the second embodiment has the same acoustic characteristic C as in the first embodiment. It can follow changes and achieves a noise reduction effect of 10 dB or more.

Also, in the active vibration noise reduction device 10 of the second embodiment, the convergence performance is improved in all frequency bands compared to the first embodiment. In particular, on the low rotational speed side (low frequency side), the active vibration noise reduction device 10 of the second embodiment exhibits an improvement in noise reduction performance of 5 dB or more as compared to that of the first embodiment. From the above, the effectiveness of the active vibration noise reduction device 10 of the second embodiment can be confirmed.

FIG. 18 is a graph showing the sound pressure level of the engine slamming sound when the active vibration noise control unit 13 makes the step size parameter μ variable. FIG. 18 shows sound pressure levels under the control off, the control of the second embodiment (fixed step size parameter μ), and the control of the second embodiment (variable step size parameter μ). As shown in FIG. 18, in the engine speed range of 3000 to 4500 RPM where the acoustic characteristic C changes, the active vibration noise reduction device 10 of the second embodiment that performs control with a variable step size parameter μ A change in the characteristic C can be followed, and a silencing effect of 10 dB or more is realized.

Also, the active vibration noise reduction device 10 of the second embodiment, which performs control with a variable step size parameter μ, achieves improved convergence performance compared to the case where the step size parameter μ is fixed. In particular, in the area up to 2000 RPM at the beginning of the adaptation process, the noise reduction performance is improved by about 10 dB compared to the case where the step size parameter μ is fixed. From the above, it can be confirmed that the initial convergence speed of the adaptive update is improved compared to the control method of the first embodiment by normalizing the step size parameter μ of the adaptive update.

Further, in the present embodiment, the active vibration noise control unit 13 further includes a phase extraction unit 90 that extracts the phase of the filter coefficient corresponding to the canceling sound transfer characteristic estimated value C^, and the reference signal generation unit 27 The reference signal x(xc, xs) is corrected by its phase rather than by the canceling sound transfer characteristic estimate C^. Therefore, the influence of the amplitude characteristic of the canceling sound transfer characteristic estimated value Ĉ on the filter coefficient update amount is reduced, and the convergence performance of the adaptive update is improved as compared with the control method of the first embodiment. That is, the canceling sound transfer characteristic estimated value Ĉ consists of an amplitude component and a phase component. The canceling sound transfer characteristic estimated value C^ has a frequency band in which the amplitude component changes greatly with the change of the frequency f, and the canceling sound transfer characteristic estimated value C^ changes greatly in this frequency band. Therefore, in this frequency band, the update amount of the filter coefficients becomes large, and there is a possibility that the convergence performance of the adaptive update may deteriorate. In this embodiment, the phase (component) is extracted from the canceling sound transfer characteristic estimated value C^, and the reference signal x is corrected with this phase, thereby suppressing the update amount of the filter coefficient and improving the convergence performance of the adaptive update. do.

In addition, the vibration noise estimation signal generation unit 28, the first adaptive notch correction filter unit 60, and the adaptive notch control filter unit 70 adjust the magnitude of the filter coefficient update amount for each sample with respect to the step size parameter μ is normalized by multiplying by the inverse of the adaptive notch filter amplitude, and the normalized step-size parameter μ is used to update the corresponding filter coefficients (Ĥ, Ĉ, W). Therefore, the filter coefficient update amount for each sample is automatically adjusted, and the initial convergence performance of adaptive updating is improved compared to the control method of the first embodiment without impairing control stability.

Although the specific embodiments have been described above, the present invention is not limited to the above embodiments and can be widely modified. For example, in the above embodiment, the active vibration noise reduction device 10 has the configuration shown in FIG. 10 as an example, but it may have the configuration shown in FIG. 11 or 12 . In this case, it can be explained by replacing the canceling sound with canceling vibration. In addition, the specific configurations, arrangements, quantities, formulas, procedures, etc. of each member and part can be changed as appropriate within the scope of the present invention. Also, the above embodiments can be combined as appropriate. On the other hand, not all of the components shown in the above embodiments are essential, and can be selected as appropriate.

2 Engine (vibration noise source)
10 active vibration noise reduction device 11 error microphone (error signal detection means)
12 speaker (canceling sound generating means)
13 Active vibration noise control unit 14 Vibration actuator (canceling sound generating means)
15 vibration sensor (error signal detection means)
21 reference signal generation unit 25 control signal generation unit 26 reference signal correction unit 27 reference signal generation unit (correction means)
28 Vibration noise estimation signal generation unit (first estimation signal generation means)
60 first adaptive notch correction filter unit (third estimated signal generating means)
70 adaptive notch control filter unit (second estimated signal generating means)
80 virtual error signal generator 81 second virtual error signal generator 82 first virtual error signal generator 90 phase extractor C Acoustic characteristics (transmission characteristics of canceling sound (from speaker 12 to error microphone 11))
C^ Canceling sound transfer characteristic estimated value C^0 Real part of canceling sound transfer characteristic estimated value C^ (first correction filter coefficient)
C^1 Imaginary part of canceling sound transfer characteristic estimated value C^ (first correction filter coefficient)
H Vibration noise d (from engine 2 to error microphone 11) transfer characteristic H^ transfer characteristic estimated value H^0 real part of transfer characteristic estimated value H^ (third correction filter coefficient)
H^1 Imaginary part of transfer characteristic estimate H^ (third correction filter coefficient)
W adaptive notch filter coefficient (circuit characteristics of reference signal generator 21)
W0 first adaptive notch filter coefficient (real part of adaptive notch filter coefficient W)
W1 second adaptive notch filter coefficient (imaginary part of adaptive notch filter coefficient W)
X engine/drive system signal r reference signal r0 first reference signal r1 second reference signal d vibration noise d^ vibration noise estimation signal e error signal e' virtual error signal e'1 second virtual error signal e'2 first virtual Error signal f Frequency x Reference signal xc Reference cosine wave signal xs Reference sine wave signal u0 Control signal (first addition signal)
u1 control signal (second addition signal)
y Cancellation vibration noise (cancellation sound reaching the error microphone 11 from the speaker 12)
y^1 First estimated value of canceling vibration noise y (second canceling vibration noise estimation signal)
y^2 Second estimated value of canceling vibration noise y (first canceling vibration noise estimation signal)

Claims (6)

reference signal generation means for generating a reference sine wave signal and a reference cosine wave signal having frequencies based on the frequency of vibration noise generated from the vibration noise source;
a first adaptive notch control filter having a first adaptive notch filter coefficient that is the real part of the adaptive notch filter coefficient and outputting a first control signal based on the reference cosine wave signal;
a second adaptive notch control filter having a second adaptive notch filter coefficient forming an imaginary part of the adaptive notch filter coefficient and outputting a second control signal based on the reference sine wave signal;
vibration noise cancellation means for outputting cancellation vibration noise based on a first addition signal obtained by adding the first control signal and the second control signal;
error signal detection means for outputting an error signal based on the difference between the vibration noise generated from the vibration noise source and the canceled vibration noise output from the vibration noise cancellation means;
The reference cosine wave signal and the reference sine wave signal are corrected by a first correction filter and a second correction filter corresponding to signal transmission characteristics from the vibration noise cancellation means to the error signal detection means with respect to the frequency of the reference signal. correction means for generating first and second reference signals by
The first correction filter is a first adaptive notch correction filter having a first correction filter coefficient forming the real part of the signal transfer characteristic, and the second correction filter is a second correction filter coefficient forming the imaginary part of the signal transfer characteristic. In an active vibration noise reduction device each configured with a second adaptive notch correction filter having
The reference cosine wave signal and the reference sine wave signal are respectively corrected by a third correction filter and a fourth correction filter to obtain first and second vibration noise estimation signals, and the first vibration noise estimation signal and the second vibration noise estimation signal are obtained. a first estimated signal generation means for generating a vibration noise estimation signal by adding the vibration noise estimation signal;
a first correction control signal obtained by correcting the reference cosine wave signal with the first correction filter and a third adaptive notch control filter having first adaptive notch filter coefficients ; A second correction control signal corrected by a third adaptive notch control filter, and a third correction control by correcting the reference sine wave signal by a fourth adaptive notch control filter having the first correction filter and the second adaptive notch filter coefficient. and a fourth correction control signal obtained by correcting the reference cosine wave signal with the second correction filter and the fourth adaptive notch control filter to generate a first canceling vibration noise estimation signal. means and
a first virtual error signal generating means for generating a first virtual error signal from the vibration noise estimation signal and the first canceling vibration noise estimation signal;
sequentially updating the filter coefficients of the third and fourth adaptive notch control filters based on the first and second reference signals and the first virtual error signal so that the first virtual error signal is minimized; 1 filter coefficient updating means ;
a first adaptive notch filter having the first adaptive notch filter coefficient and outputting a third control signal based on the reference sinusoidal signal;
a second adaptive notch filter having the second adaptive notch filter coefficients and outputting a fourth control signal based on the reference cosine wave signal;
Obtained by adding a fifth correction control signal obtained by correcting the first addition signal with a fifth adaptive notch correction filter having the first correction filter coefficient, the third control signal, and the fourth control signal. A third estimation for generating a second canceling vibration noise estimation signal by adding a sixth correction control signal obtained by correcting the second addition signal obtained by correcting with a sixth adaptive notch correction filter having the second correction filter coefficient a signal generating means;
a second virtual error signal generating means for generating a second virtual error signal from the error signal, the vibration noise estimation signal, and the second canceling vibration noise estimation signal;
Based on the first control signal, the second control signal, the third control signal, the fourth control signal, and the second virtual error signal, the fifth and the second virtual error signals are minimized. and second filter coefficient updating means for sequentially updating filter coefficients of the sixth adaptive notch correction filter . In the active vibration noise reduction device according to claim 1 ,
The third correction filter is a third adaptive notch correction filter, the fourth correction filter is a fourth adaptive notch correction filter, and
Filter coefficients of the third and fourth adaptive notch correction filters are successively set based on the reference sine wave signal, the reference cosine wave signal, and the second virtual error signal so as to minimize the second virtual error signal. An active vibration noise reduction device, comprising a third filter coefficient updating means for updating. In the active vibration noise reduction device according to claim 1 or claim 2 ,
The filter coefficients of the fifth and sixth adaptive notch correction filters are multiplied by the reciprocals of the square root sums of the filter coefficients of the fifth and sixth adaptive notch correction filters to obtain first and second normalization filter coefficients. A normalization means for calculating
The correction means is configured to convert the reference cosine wave signal and the reference sine wave signal by the first adaptive notch correction filter having the first normalization filter coefficients and the second adaptive notch correction filter having the second normalization filter coefficients. are corrected to generate the first and second reference signals. In the active vibration noise reduction device according to claim 1 or claim 2 ,
The filter coefficients of the fifth and sixth adaptive notch correction filters are multiplied by the reciprocal of the larger absolute value of the filter coefficients of the first and second adaptive notch correction filters to obtain the third and fourth normalization filters. A normalization means for calculating coefficients,
The correcting means is configured to convert the reference cosine wave signal and the reference sine wave signal by the first adaptive notch correction filter having the third normalization filter coefficient and the second adaptive notch correction filter having the fourth normalization filter coefficient. are corrected to generate the first and second reference signals. In the active vibration noise reduction device according to claim 2 ,
The first, second, and third filter coefficient updating means set a step size parameter for controlling the amount of update of the filter coefficients of the adaptive notch filters to be updated, based on the square root of the sum of squares of the filter coefficients immediately before update. An active vibration noise reduction device, characterized in that: In the active vibration noise reduction device according to claim 2 ,
The first, second, and third filter coefficient updating means set a step size parameter for controlling the update amount of the filter coefficients of the adaptive notch filters to be updated to the larger absolute value of the filter coefficients immediately before updating. An active vibration noise reduction device, characterized in that the determination is made based on the

Images