JP2009255664A - Active type noise control device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an active type noise control device capable of exerting a maximum noise reducing effect in any case of a phase of a control object noise and obtaining an optimum noise reducing effect, even when a control frequency is deviated from a frequency of actually generated control object noise. <P>SOLUTION: A phase of a control signal z[n] is directly updated in a phase correction means 11. Further, the control frequency is corrected in a control frequency correction means 12 to exert the maximum noise reducing effect. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to an active noise reduction device that actively reduces vibration noise generated from rotating equipment such as a vehicle engine.

  In a conventional active noise reduction device, a method of performing adaptive control using an adaptive notch filter is known (see, for example, Patent Document 1). FIG. 8 shows a configuration equivalent to the configuration of the conventional active noise reduction device described in Patent Document 1. In FIG.

In FIG. 8, the discrete calculation for realizing the active noise reduction device is executed in the discrete calculation processing unit 29. The engine speed detector 14 outputs a pulse train having a frequency proportional to the engine speed as an engine pulse P. For example, the engine pulse P is generated by taking out the output of the crank angle sensor. The control frequency detector 15 calculates and outputs a control frequency f based on the engine pulse P. The reference signal generation unit 16 has a sine wave table 17 that holds, in a memory, a sine value obtained by multiplying a sine value of each point obtained by dividing one cycle of a sine wave by 1 / √2 in a memory. Data is selected from 17, and a sine wave reference signal x1 [n] and a cosine wave reference signal x2 [n] whose frequency is equal to the control frequency f are generated and output. The reference signal generator 24 represents a sine wave reference signal correction value table 25 simulating the transfer characteristic value from the speaker 22 to the microphone 23 (the sine wave reference signal correction value at the frequency f [Hz] is represented as C1 [f]. ) And the cosine wave reference signal correction value table 26 (the cosine wave reference signal correction value at the frequency f [Hz] is expressed as C2 [f]), and the sine wave reference signal r1 [n] and the cosine wave reference. A signal r2 [n] is generated and output. The first one-tap digital filter 19 filters x1 [n] with a filter coefficient W1 [n] held therein to generate a sine wave control signal y1 [n]. The second 1-tap digital filter 20 filters the cosine wave reference signal x2 [n] with the filter coefficient W2 [n] held therein to generate the cosine wave control signal y2 [n]. The power amplifier 21 amplifies the control signal z [n] obtained by adding y1 [n] and y2 [n]. The speaker 22 outputs the output signal from the power amplifier 21 as noise canceling sound. The microphone 23 detects a sound generated as a result of interference between noise and noise canceling sound as an error signal ε [n]. Based on the sine wave reference signal r1 [n] and the error signal ε [n], the first adaptive control algorithm calculation unit 27 uses, for example, a filter coefficient W1 based on an LMS (Least Mean Square) algorithm which is a kind of steepest descent method. [N] is updated sequentially. Similarly, the second adaptive control algorithm calculation unit 28 sequentially updates the filter coefficient W2 [n] based on the cosine wave reference signal r2 [n] and the error signal ε [n]. The update formula for the coefficients W1 and W2 is
W1 [n] = W1 [n−1] −μ × r1 [n] × ε [n] (1)
W2 [n] = W2 [n−1] −μ × r2 [n] × ε [n] (2) Here, μ is a constant called a convergence coefficient, and is related to the time for which the coefficients W1 and W2 converge to the optimum value. By repeating the above-described processing at a predetermined cycle, noise can be reduced.
JP 2004-361721 A

However, in the conventional configuration, in order to avoid data overflow when calculating the control signal z [n] by combining the sine wave control signal y1 [n] and the cosine wave control signal y2 [n], the sine wave The table holds a value obtained by multiplying the sine value by 1 / √2. For this reason, z [n] has a problem that the maximum output can be obtained only when the phase of the noise to be controlled is 45 degrees, 135 degrees, 225 degrees, and 315 degrees, and the maximum noise reduction effect can always be exhibited. Furthermore, when the control frequency f deviates from the frequency of the control target noise that is actually generated due to an error in the frequency of the engine pulse P, the noise reduction effect is reduced.

  The present invention can exhibit the maximum noise reduction effect regardless of the phase of the noise to be controlled, and the optimum noise reduction effect even when the control frequency f deviates from the frequency of the noise to be controlled actually generated. It is an object of the present invention to provide an active noise control device obtained.

  The active noise control device of the present invention generates a control frequency detection means for detecting the frequency of noise to be controlled due to a noise source, and a sine wave having the same frequency as the control frequency determined by the control frequency detection means. A sine wave generating means, a one-tap digital filter to which the sine wave from the sine wave generating means is input, and an output signal from the one-tap digital filter is input to cause interference with noise to be controlled due to the noise source Control signal generating means for outputting a control signal for detecting the error signal, and error signal detection for detecting an error signal resulting from interference between the control signal output from the control signal generating means and noise to be controlled due to the noise source Means, coefficient updating means for updating the filter coefficient of the one-tap digital filter, and phase correcting means for correcting the phase of the sine wave generating means And a control frequency correcting means for correcting the control frequency of the control signal detecting means, wherein the coefficient updating means and the phase correcting means update the filter coefficient and the phase, respectively, using the error signal. The noise updating unit is configured to reduce noise to be controlled due to the noise source. In particular, the coefficient updating unit includes the error signal and a phase characteristic of a transfer characteristic from the control signal generating unit to the error signal detecting unit at a control frequency. The phase correction means is configured to update the filter coefficient based on the error signal and the phase characteristic of the transfer characteristic from the control signal generation means to the error signal detection means at the control frequency. The phase of the sine wave generation means is determined, and the control frequency correction means is configured to determine the phase correction amount calculated by the phase correction means. The control frequency correction unit is configured to correct the control frequency based on the product value, and the sine wave generation unit is configured to determine a read position of the sine wave table that holds one cycle of the discretized sine value as a predetermined value. Further, the reading position is moved by a phase correction amount determined by the phase correction means.

  In the active noise control device of the present invention, since the signal synthesis process does not occur when calculating the control signal z [n], the value held in the sine wave table does not need to be multiplied by 1 / √2. The sine value itself may be used. Therefore, the effect of being able to output the control signal z [n] at the maximum value whatever the phase of the noise to be controlled is obtained. In addition, when the control frequency f deviates from the frequency of the noise to be controlled that is actually generated, the effect of being able to realize a more optimal noise reduction effect by correcting the control frequency f in a direction closer to the noise frequency. Is obtained.

(Embodiment 1)
Hereinafter, an active noise reduction apparatus according to Embodiment 1 of the present invention will be described with reference to the drawings.

  FIG. 1 is a block diagram of an active noise reduction apparatus according to Embodiment 1 of the present invention.

In FIG. 1, an engine speed detector 1 outputs a pulse train having a frequency proportional to the speed of an engine as a noise source mounted on a vehicle as an engine pulse P. The control frequency detection means 2 has a control frequency correction amount fcomp [n] [Hz] inside, and is based on the predicted control frequency fep [n] [Hz] calculated from the engine pulse P and the control frequency correction amount fcomp [n]. Then, the control frequency f [n] [Hz] is calculated. The sine wave table 3 holds a sine value at each point obtained by equally dividing one cycle of the sine wave in a memory. The sine wave generating means 4 has a phase correction amount Δθ [n] [point] inside, and updates a pointer p [n] [point] indicating the current position of the sine wave table. The 1-tap digital filter 5 has a filter coefficient W [n] therein and outputs a control signal z [n] using the pointer p [n] and the sine wave table 3. The power amplifier 6 amplifies the control signal z [n]. The speaker 7 as the control signal generating means outputs the output signal from the power amplifier 6 as noise canceling sound. The microphone 8 serving as the error signal detection means detects a sound generated as a result of interference between the control target noise generated due to engine vibration and the noise canceling sound as an error signal ε [n]. The phase characteristic table 9 holds a value obtained by converting the phase characteristic value of the transfer characteristic from the speaker 7 to the microphone 8 into the relative point movement amount of the sine wave table 3 for each frequency. The coefficient updating means 10 uses the pointer p [n], the control frequency f [n], the sine wave table 3, the phase characteristic table 9, and the error signal ε [n] to filter the filter coefficient W [ n]. The phase correction unit 11 uses the pointer p [n], the control frequency f [n], the sine wave table 3, the phase characteristic table 9, and the error signal ε [n], and the phase correction amount Δθ of the sine wave generation unit 4. [N] is determined. The control frequency correction unit 12 determines the control frequency correction amount fcomp [n] of the control frequency detection unit 2 using the phase correction amount Δθ [n] determined by the phase correction unit 11. The discrete arithmetic processing unit 13 is configured by software.

  Next, a specific operation of this apparatus will be described. Calculation of control frequency f [n], update of pointer p [n], generation of control signal z [n], detection of error signal ε [n], update of filter coefficient W [n], and phase correction amount Δθ [n ] And control frequency correction amount fcomp [n] are all executed in the same cycle, and each represents a value after n cycles. In the following description, the cycle is T (seconds).

  The control frequency detection means 2 first generates an interrupt at each rising edge of the engine pulse P, for example, measures the time between the rising edges, and calculates the predicted control frequency fep [n] based on the measurement result. Next, the control frequency f [n] is calculated according to the equation (3) based on the predicted control frequency fep [n] and the control frequency correction amount fcomp [n].

f [n] = fep [n] + fcomp [n] (3)
The sine wave table 3 divides one cycle of the sine wave into N equal parts and holds a value obtained by discretizing the sine value of each point with a predetermined bit in a memory. When an array in which sine values from the 0th point to the (N−1) th point are discretized with b bits and stored is represented by s [m] (0 ≦ m <N), the relational expression (4) is established.

s [m] = int (2 ^b-1 × sin (360 × m / N)) (4)
Here, int (x) represents the integer part of x, and the unit of the angle of the sine function is [degree]. For example, a graph and a table of s [m] when N = 3000 are shown in FIG. 2 and (Table 1), respectively.

  The sine wave generating means 4 updates the pointer p [n] according to the equation (5) based on the control frequency f [n] and the phase correction amount Δθ [n].

p [n] = (p [n−1] + (f [n] × N × T) + Δθ [n])
mod N (5)
However, “x mod y” represents the remainder when x is divided by y. Here, if N and T are selected in advance such that N × T = 1, the multiplication of N and T in Equation (5) is not necessary. Furthermore, considering that “f [n] + Δθ [n] <N” normally, Equation (5) can be rewritten as Equation (6).

When (p [n−1] + f [n] + Δθ [n] <N):
p [n] = p [n−1] + f [n] + Δθ [n]
When (p [n−1] + f [n] + Δθ [n] ≧ N):
p [n] = p [n−1] + f [n] + Δθ [n] −N (6)
In the following description, it is assumed that N × T = 1 holds.

  The 1-tap digital filter 5 generates the control signal z [n] by Expression (7).

z [n] = W [n] × s [p [n]] (7)
The phase characteristic table 9 is an array c [k] obtained by converting the phase characteristic value (example of graph: FIG. 3) of the transfer characteristic from the speaker 7 to the microphone 8 into the relative point movement amount of the sine wave table 3. Is stored in the memory (k is the frequency [Hz]). When the phase characteristic value at k [Hz] is phase [k] [degree], the relational expression (8) is established.

c [k] = int (N × phase [k] / 360) (8)
For example, the state of c [k] when N = 3000 and the range of the control target noise frequency is from 30 Hz to 100 Hz is shown in (Table 2).

  The coefficient updating means 10 updates the filter coefficient W [n] according to Expression (9) and Expression (10), for example, by an LMS (Least Mean Square) algorithm which is a kind of steepest descent method.

When (0 ≦ p [n] + c [f [n]] <N):
r1 [n] = s [p [n] + c [f [n]]]
When (p [n] + c [f [n]] ≦ N):
r1 [n] = s [p [n] + c [f [n]] − N]
When (p [n] + c [f [n]] <0):
r1 [n] = s [p [n] + c [f [n]] + N] (9)
W [n] = W [n-1]-[mu] 1 * r1 * [epsilon] [n] (10)
Here, μ1 is a parameter for determining the convergence speed of W [n] in the steepest descent method, and the larger the value, the faster the convergence. When the calculation result of Expression (10) is a negative number, the sign of W [n] is inverted, and p [n] is advanced by 180 degrees according to Expression (11).

When (p [n] <N / 2): p [n] = p [n] + N / 2
When (p [n] ≧ N / 2): p [n] = p [n] −N / 2 (11)
Thereby, W [n] can always be a positive value, and calculation of a phase correction amount Δθ [n], which will be described later, is simplified.

  The phase correction unit 11 calculates the phase correction amount Δθ [n] according to the equations (12) and (13).

When (0 ≦ p [n] + c [f [n]] + N / 4 <N):
r2 [n] = s [p [n] + c [f [n]] + N / 4]
When (p [n] + c [f [n]] + N / 4 ≦ N):
r2 [n] = s [p [n] + c [f [n]] + N / 4-N]
When (p [n] + c [f [n]] + N / 4 <0):
r2 [n] = s [p [n] + c [f [n]] + N / 4 + N] (12)
Δθ [n] = − μ2 × r2 [n] × ε [n] (13)
Here, μ2 is a parameter that determines the degree of phase correction amount determination, and the larger the value, the faster the convergence.

  The control frequency correction unit 12 first updates the cumulative value Δθaccum [n] of the phase correction amount Δθ [n] determined by the phase correction unit 11 according to the equation (14).

Δθaccum [n] = Δθaccum [n−1] + Δθ [n] (14)
Next, when the cumulative number of phase correction amounts reaches the specified number Naccum, the control frequency correction amount fcomp [n] is updated and Δθaccum [n] is reset to zero.

  As a method for determining the control frequency correction amount fcomp [n], for example, there is a method based on a calculation expression such as Expression (15).

fcomp [n] = Δθaccum ÷ Naccum (15)
According to the above procedure, the filter coefficient W [n] is converged to a constant value, the phase correction amount Δθ [n] is converged to 0, and the control frequency correction amount fcomp [n] is converged to a constant value. Can be reduced.

  Here, the mechanism by which the noise of the control frequency is reduced by the phase correction amount calculation formula according to Formula (13) will be described. In the noise control apparatus described in the conventional example, the filter coefficients W1 [n] and W2 [n] are sequentially updated based on an LMS (Least Mean Square) algorithm. The update formula was as follows.

W1 [n] = W1 [n−1] −μ × r1 [n] × ε [n] (1)
W2 [n] = W2 [n−1] −μ × r2 [n] × ε [n] (2)
As described above, in general, the sine wave reference signal r1 [n] and the cosine wave reference signal r2 [n] have a product of the sine wave signal and cosine wave signal of the noise frequency to be reduced and the error signal ε [n]. Is used. This utilizes the orthogonality of the sine wave and cosine wave, and the sine wave reference signal r1 [n] and cosine wave reference are included in the error signal ε [n] in the long-term sequential update (ie, n → ∞). The product of the same frequency components as the frequency of the signal r2 [n] is accumulated, and the accumulated value of the products of the other frequency components is zero. Therefore, W1 [n] and W2 [n] reduce the same frequency component as the frequency of the sine wave reference signal r1 [n] and the cosine wave reference signal r2 [n] in the error signal ε [n]. The coefficient is updated, and finally the average of W1 [n] and W2 [n] when the same frequency component as the frequency of the sine wave reference signal and the cosine wave reference signal in the error signal becomes zero. The coefficient update is 0, and W1 [n] and W2 [n] converge. In other words, it can be said that W1 [n] and W2 [n] adjust the amplitude and phase of the control signal z [n] so that the noise to be controlled is reduced. The amplitude and phase of the control signal are
Amplitude: √ (W1 [n] ² + W2 [n] ² )
Phase: tan ⁻¹ (W2 [n] / W1 [n])
It becomes.

On the other hand, in the present invention, the filter coefficient W [n] corresponding to the amplitude of the control signal z [n] is updated by a conventional method, while the phase of the control signal z [n] is controlled by the control signal z [n]. The phase correction amount Δθ [n] is updated by directly calculating the product “r2 [n] × ε [n]” of the orthogonal component advanced by 90 degrees and the error signal ε [n]. 4 and 5 show how the phase correction direction is determined. When the sign of r2 [n] × ε [n] is positive, the phase of the control signal z [n] is corrected in the negative direction as shown in FIG. On the other hand, when the sign of r2 [n] × ε [n] is negative, the control signal z [n] is phase-corrected in the positive direction as shown in FIG. That is, the control signal z [n] is corrected to the error signal ε by correcting the phase of the control signal z [n] in the direction opposite to the sign of the orthogonal component of the control signal z [n] as in Expression (13). It is updated so as to approach a signal having a phase opposite to that of [n], and the effect of reducing the control target noise is increased.

Next, a mechanism by which the control frequency correction amount fcomp [n] can be calculated correctly using Expression (15) will be described. When the frequency of the control signal z [n] deviates from the frequency of the control target noise that is actually generated, the adaptive notch filter tracks its own phase characteristic θ (t) in an attempt to approach the latter frequency ( An appropriate time is set to t = 0). That is,
Noise (t) = Rnoise x sin (360 x fnoise x t)
Control signal (t) = − Rctrl × sin (360 × fctrl × t + θ (t))
... (16)
The adaptive notch filter is 360 × fnoise × t = 360 × fctrl × t + θ (t) (17)
The θ (t) is continuously updated so that From equation (18), the frequency deviation fdiff between the control signal and the noise is
fdiff = fnoise-fctrl
= Θ (t) ÷ (360 × t) (18)
It can be expressed as. Further, when t = Naccum × T,
fdiff = θ (Naccum × T) ÷ (360 × Naccum × T)
= (N × θ (Naccum × T) ÷ 360) ÷ Naccum (19)
Here, “N × θ (Naccum × T) ÷ 360” is nothing but the phase characteristic after Naccum × T [seconds] of the adaptive notch filter expressed in points in the sine wave table. Therefore, it can be said that the calculation formula of the control frequency correction value of the above equation (15) is appropriate.

  On the other hand, a method of updating the control frequency correction amount fcomp [n] based on the equation (20) is also conceivable.

When (Δθaccum ≧ X):
fcomp [n] = fcomp [n−1] + Δf
When (Δθaccum ≦ −X):
fcomp [n] = fcomp [n−1] −Δf
(-X <Δθaccum <X):
fcomp [n] = fcomp [n−1] (20)
The above calculation formula is an update formula that does not use division, and is effective when an arithmetic unit that does not have a division instruction is used. However, the method for calculating the control frequency correction amount fcomp [n] according to the equation (20) requires a longer time to reach the optimum correction value than the method according to the equation (15).

  Note that the larger the Naccum value, the more stable the control frequency can be corrected, but the longer it takes to reach the optimum correction value.

Here, the present invention and the method described in Patent Document 1 are compared from the viewpoint of noise reduction effect. In the method described in Patent Document 1, the sine wave control signal y1 [n] and the cosine wave control signal y2 [n] are combined to calculate the control signal z [n] to avoid data overflow. In the wave table, it was necessary to hold a value obtained by multiplying the sine value by 1 / √2. For this reason, the control signal z [n] has a problem that it can output the maximum output only when the phase of the noise to be controlled is 45 degrees, 135 degrees, 225 degrees, and 315 degrees, and cannot always exert the maximum noise reduction effect. It was. An example of this situation is shown in FIG. The control signal z [n] can be output at the maximum value in FIG. 6 (A), but cannot be output at the maximum value in FIGS. 6 (B) and 6 (C). On the other hand, in the present invention, when the control signal z [n] is calculated, signal synthesis processing does not occur. Therefore, the value held in the sine wave table does not need to be multiplied by 1 / √2, and the sine value itself. It doesn't matter. Therefore, the control signal z [n] can be output at the maximum value regardless of the phase of the noise to be controlled. An example of this situation is shown in FIG. In any case of FIGS. 7A to 7C, the control signal z [n] can be output at the maximum value.

  Further, in the method described in Patent Document 1, when the control frequency f deviates from the frequency of the control target noise that is actually generated due to an error in the frequency of the engine pulse P, the noise reduction effect is reduced. There was a problem. On the other hand, in the present invention, the predicted control frequency fep [n] calculated based on the engine pulse P is corrected so as to be close to the frequency of the control target noise that is actually generated, thereby further reducing noise more optimally. The effect can be realized.

  In the present invention, by preparing a plurality of control frequency detection means 2, sine wave generation means 4, 1-tap digital filter 5, coefficient update means 10, phase correction means 11 and control frequency correction means 12, respectively. It is also possible to mute multiple order components of the control target noise.

  The active noise reduction device according to the present invention can exhibit the maximum noise reduction effect regardless of the phase of the control target noise, and is optimal even when the control frequency deviates from the frequency of the control target noise actually generated. It is useful as an active noise reduction device that can achieve a significant noise reduction effect.

The block diagram for demonstrating the active noise reduction apparatus in Embodiment 1 of this invention Graph showing an example of a sine wave table in the active noise reduction device The graph which shows the example of the phase characteristic value of the transfer characteristic from the speaker to the microphone in the active noise reduction device The figure which shows the mode of the phase correction of the control signal in the active noise reduction apparatus (the 1) The figure which shows the mode of the phase correction of the control signal in the active noise reduction apparatus (the 2) Block diagram showing magnitude of control signal output in conventional active noise reduction device The block diagram which shows the magnitude | size of the output of the control signal in Embodiment 1 of this invention Block diagram showing the configuration of a conventional active noise reduction device

Explanation of symbols

1 Engine speed detector (noise source)
2 Control frequency detection means 3 Sine wave table 4 Sine wave generation means 5 1 tap digital filter 6 Power amplifier 7 Speaker (control signal generation means)
8 Microphone (error signal detection means)
DESCRIPTION OF SYMBOLS 9 Phase characteristic table 10 Coefficient update means 11 Phase correction means 12 Control frequency correction means 13 Discrete arithmetic processing part

Claims (5)

Control frequency detection means for detecting the frequency of noise to be controlled due to a noise source, sine wave generation means for generating a sine wave having the same frequency as the control frequency determined by the control frequency detection means, and the sine wave generation A 1-tap digital filter to which a sine wave from the means is input, and a control signal generation for outputting a control signal for receiving an output signal from the 1-tap digital filter and causing interference with noise to be controlled caused by the noise source Means, an error signal detecting means for detecting an error signal resulting from interference between the control signal output from the control signal generating means and noise to be controlled caused by the noise source, and a filter of the one-tap digital filter Coefficient updating means for updating the coefficient, phase correcting means for correcting the phase of the sine wave generating means, and control of the control signal detecting means Control frequency correction means for correcting the wave number, and the coefficient update means and the phase correction means update the filter coefficient and the phase, respectively, using the error signal, thereby controlling the noise source. An active noise control device configured to reduce noise to be generated. 2. The coefficient updating unit is configured to update the filter coefficient based on the error signal and a phase characteristic of a transfer characteristic from a control signal generating unit to an error signal detecting unit at a control frequency. Active noise control device. The phase correction unit is configured to determine a phase of the sine wave generation unit based on the error signal and a phase characteristic of a transfer characteristic from the control signal generation unit to the error signal detection unit at a control frequency. The active noise control apparatus according to claim 1. 2. The active noise control apparatus according to claim 1, wherein the control frequency correction unit is configured to correct a control frequency of the control frequency correction unit based on a cumulative value of the phase correction amount calculated by the phase correction unit. . The sine wave generating means moves the reading position of the sine wave table that holds one cycle of the discretized sine value at a predetermined period, and further moves the reading position by the phase correction amount determined by the phase correcting means. The active noise control apparatus according to claim 1, wherein the active noise control apparatus is configured to cause the noise.

Images