JP3517883B2 - Adaptive control device and active noise control device - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an adaptive control device capable of controlling even if the behavior of a controlled object is unknown or fluctuates, and the noise transmitted from a noise source is emitted from a control sound source. Regarding an active noise control device for reducing noise by interfering with control sound,
In a device including a digital filter with a variable filter coefficient that generates a signal for driving a control actuator or a control sound source, and an adaptive processing unit that updates the filter coefficient of this digital filter according to a predetermined adaptive algorithm, the state of the controlled object Even if the processing contents of the adaptive processing unit do not match the actual control target or the characteristics of the device due to changes or deterioration of the component parts of the device, the filter coefficient of the digital filter can quickly converge to the optimum value. It was done.

[0002]

2. Description of the Related Art As a conventional device of this type, there are those described in British Patent No. 2149614 and Japanese Patent Publication No. 1-501344. This conventional device is a noise reduction device applied to a cabin of an aircraft or a similar closed space, and a single noise source such as an engine located outside the closed space has a fundamental frequency f ₀ and its harmonics. It operates under the condition that noise including the waves f _{1 to} f _n is generated.

[0003] More specifically, includes a microphone for detecting a plurality of the installed sound pressure to a location within the closed space, and a plurality of loudspeakers for generating a control sound to the closed space, the frequency f ₀ of the noise source - Based on the f _n component, those frequencies f ₀ ~
The loudspeaker is driven by a signal having a phase opposite to that of the f _n component,
Therefore, a control sound having a phase opposite to that of the noise transmitted to the closed space is generated from the loudspeaker to cancel the noise.

Then, as a method of generating the control sound emitted from the loudspeaker, PROCEEDINGS OF THE IEEE, VOL.
63 PAGE 1692-1975, “ADAPTIVE NOISE CANSELLATION
: The "WIDROW LMS" algorithm described in "PRINCIPLES AND APPLICATIONS" is applied to multiple channels. The contents are based on the paper "A MULTIPLE ERROR LMS ALGORIT" by the inventor of the above patent.
HM AND ITS APPLICATIONTO THE ACTIVE CONTROL OF SOU
ND AND VIBRATION ”, IEEE TRANS.ACOUST., SPEECH, SIGN
AL PROCESSING, VOL.ASSP −35, PP.1432-1434, 1987.

That is, the LMS algorithm is one of the algorithms suitable for updating the filter coefficient of the adaptive digital filter, and is, for example, the so-called Filter.
In the red-X LMS algorithm, a filter (transfer function model) that represents the acoustic transfer characteristic from the loudspeaker to the microphone is set for all combinations of the loudspeaker and the microphone, and is a standard that represents the noise generation state of the noise source. A filter coefficient of a variable digital filter provided for each loudspeaker so that the value of a predetermined evaluation function based on the value obtained by processing the signal with the filter and the residual noise detected by each microphone is reduced. Have been updated.

[0006]

Certainly, in the conventional device as described above, the filter coefficient of the digital filter for generating the signal for driving the loudspeaker converges to the optimum value as the control progresses. Eventually, the noise in the closed space will be reduced, but if the acoustic transfer characteristics in the real space change due to changes in the environment such as temperature and humidity in the closed space, or deterioration of the loudspeaker or microphone, digital noise will change. There is a problem that the accuracy of the transfer function model required for the algorithm for updating the filter coefficient of the filter is relatively lowered, and it takes time to converge the filter coefficient of the digital filter, and the adaptive operation is delayed.

In order to solve such a problem, for example, a multidimensional map of a transfer function model is constructed with parameters such as temperature and humidity in a closed space and deterioration conditions of loudspeakers and microphones. A method of updating the transfer function model appropriately according to changes in the situation in the closed space can be considered, but with this, as the number of loudspeakers and microphones increases, the number of transfer function models to be stored as a map is increased. It becomes enormous and the storage capacity required becomes extremely large, which is not practical, and it is difficult to perfectly match a preset transfer function model with the actual transfer function of the acoustic space.

An identification sound is generated when a factor that affects the acoustic transfer characteristics between the loudspeaker and the microphone (for example, the temperature in the closed space, the open / closed state of windows, the number of people present, etc.) changes. To measure the acoustic transfer characteristics,
A method of updating the transfer function model on the basis of this is also conceivable, but this has a drawback in that the identification sound is heard by a person existing in the space when performing identification, which causes discomfort.

The present invention has been made by paying attention to the unsolved problems of the prior art as described above, and does not cause a problem such as an extreme increase in storage capacity and an unpleasant feeling to humans. Even if the processing content of the adaptive processing unit does not match the actual control target or device characteristics due to changes in the status of the control target or deterioration of the components of the device, the filter coefficient of the digital filter can be quickly set to the optimum value. An object of the present invention is to provide an adaptive processing device and an active noise control device that can be converged.

[0010]

In order to achieve the above object, the invention according to claim 1 is a digital filter having a variable filter coefficient for generating a signal for driving a control actuator according to a predetermined reference signal, and a predetermined filter. In the adaptive control device, comprising: an adaptive processing unit that updates the filter coefficient of the digital filter so that the value of the evaluation function of ## EQU1 ## matches the target value. and the optimum value convergence determining means for determining, from the filter coefficient of the digital filter when the optimum value convergence determining means is determined to have converged to the optimum value
Shortest convergence path to the optimum value can be seen, and, determine the filter coefficients of the digital filter does not converge to the optimum value
Based on the actual convergence path that the actual convergence path before
The correction means for correcting the processing content of the adaptive processing means is provided so as to approach the shortest convergence path .

According to a second aspect of the present invention, in the above-mentioned first aspect of the invention, the correction means is the actual convergence process.
It has a gain adjusting means for changing the update gain of the filter coefficient of the digital filter so that the path approaches the shortest convergence path . And the invention of claim 3 is the same as the invention of claim 1 or claim 2,
The correction means does not correct the processing content of the adaptive processing means when the filter coefficient of the digital filter changes when the optimum value convergence determination means determines that the optimum value has converged to the optimum value.

Further, in the invention described in claim 4, in the invention described in any one of claims 1 to 3, the correction means is a filter of the digital filter when the optimum value convergence judgment means judges that the optimum value has converged to the optimum value. A direction determined by the coefficient and the filter coefficient of the digital filter at the start of control,
To eliminate the deviation between the filter coefficient of the digital filter that has not converged to the optimum value and the direction determined by the filter coefficient of the digital filter at the start of control.
Then, the processing content of the adaptive processing means is corrected.

On the other hand, in order to achieve the above object, the invention according to claim 5 detects a control sound source capable of generating a control sound in a space where noise is transmitted from the noise source, and a noise generation state of the noise source. Noise generating state detecting means for outputting as a reference signal, a digital filter having a variable filter coefficient for generating a signal for driving the control sound source according to the reference signal, and a residual noise detecting residual noise at a predetermined position in the space. In an active noise control device comprising noise detection means and adaptive processing means for updating the filter coefficient of the digital filter so that the value of a predetermined evaluation function based on the residual noise is reduced, Optimal value convergence determining means for determining whether or not the filter coefficient has converged to an optimal value, and when it is determined that the optimal value convergence determining means has converged to an optimal value. Shortest convergence path to the optimum value can be seen from the filter coefficients of the digital filters, and, based on the actual convergence path seen from the filter coefficient of the digital filter does not converge to the optimum value, the actual yield
Correction means for correcting the processing content of the adaptive processing means so that the bundle path approaches the shortest convergence path .

According to the invention of claim 6, in the invention of claim 5, the adaptive processing means has a transfer function model corresponding to an acoustic transfer characteristic between the control sound source and the residual noise detecting means, The means is that the actual convergence path is
The transfer function model is corrected so as to approach the shortest convergence path . The invention according to claim 7 is
In the invention described in claim 6, the correction means is a transfer function correction filter arranged in series with the transfer function model included in the adaptive processing means, the transfer characteristic variable filter, and the optimum value convergence determination means converges to an optimum value. Based on the filter coefficient of the digital filter and the filter coefficient of the digital filter that has not converged to the optimum value when it is determined that
As the actual convergence path approaches the shortest convergence path,
Transfer characteristic control means for changing the transfer characteristic of the transfer function correction filter.

Further, in the invention described in claim 8, in the invention described in claim 6, the correction means is a transfer function correction filter having a variable transfer characteristic arranged in series with a digital filter, and an optimum value convergence determination means. Based on the filter coefficient of the digital filter when it is determined that has converged to the optimum value and the filter coefficient of the digital filter that has not converged to the optimum value, the actual convergence path is the shortest.
Transfer characteristic control means for changing the transfer characteristic of the transfer function correction filter so as to approach the convergent path .

On the other hand, in the invention described in claim 9, in the invention described in claim 5, the correction means is configured to perform the actual convergence process.
It has a gain adjusting means for changing the update gain of the filter coefficient of the digital filter so that the path approaches the shortest convergence path . The invention according to claim 10 is the invention according to any one of claims 5 to 9,
The correction means does not correct the processing content of the adaptive processing means when the filter coefficient of the digital filter changes when the optimum value convergence determination means determines that the optimum value has converged to the optimum value.

According to an eleventh aspect of the present invention, in the invention according to the fifth to tenth aspects, the correction means is a filter of the digital filter when the optimum value convergence determination means determines that the optimum value has converged to the optimum value. The deviation between the direction determined by the coefficient and the filter coefficient of the digital filter at the start of control and the direction determined by the filter coefficient of the digital filter immediately after the start of control and the filter coefficient of the digital filter at the start of control is eliminated. The processing contents of the adaptive processing means are corrected.

[0018]

According to the first aspect of the present invention, when the digital filter generates a signal for driving the control actuator in response to a predetermined reference signal, the control actuator is driven by this signal and predetermined control is executed. Immediately after the start of control, the filter coefficient of the digital filter does not always converge to the optimum value, and therefore it cannot be said that good control is always executed by the control actuator.

However, since the adaptive processing means updates the filter coefficient of the digital filter so that the predetermined evaluation function coincides with the target value, the filter coefficient of the digital filter gradually converges to the optimum value, and the control actuator operates. Good control will be performed. That is,
As a result of the adaptive processing means appropriately updating the filter coefficient of the digital filter, even if the optimum filter coefficient of the digital filter is unknown before the control is executed, it is possible to finally obtain the filter coefficient that allows good control.

Since the shortest convergence path from the filter coefficient of the digital filter that converges to the optimum value to the optimum value can be known, and the actual convergence path can be known from the filter coefficient of the digital filter that has not converged to the optimum value. Based on the filter coefficient of the digital filter when the optimum value convergence determination means determines that the optimum value has converged to the optimum value and the filter coefficient of the digital filter that has not converged to the optimum value, the actual convergence path for the shortest convergence path. You can see the deviation.

Therefore, the shortest convergence path to the optimum value which can be known from the filter coefficient of the digital filter when the optimum value convergence judgment means judges that the optimum value has converged to the optimum value, and the filter coefficient of the digital filter which has not converged to the optimum value. Understand from
Based on the actual convergence path , the correction means causes the processing content of the adaptive processing means to approach the convergence path where the actual convergence path is the shortest.
As a result of such correction, the convergence path of the filter coefficient of the digital filter by the subsequent processing of the adaptive processing means becomes closer to the shortest convergence path as compared with the convergence path before correction.

Further, if the deviation of the actual convergence path from the shortest convergence path is large and the update gain of the filter coefficient of the digital filter is large, the deviation between them may become larger as the filter coefficient is updated. Therefore, as in the invention according to claim 2, as the correction means, the update gain of the filter coefficient of the digital filter is set.
There is a gain adjusting means for changing the actual convergence path so that it approaches the shortest convergence path . For example, if the deviation of the actual convergence path from the shortest convergence path is large, the deviation of the convergence path will be reduced if the update gain is reduced. Can be prevented from expanding.

When the optimum value convergence determination means determines that the filter coefficient of the digital filter has converged to the optimum value, the shortest convergence path cannot be accurately determined. If the processing content of the adaptive processing means is not corrected as in the invention, it is possible to avoid that the processing content of the adaptive processing means fluctuates based on unclear information.

Then, from the filter coefficient of the digital filter when the optimum value convergence determining means determines that it has converged to the optimum value and the filter coefficient of the digital filter at the start of control (that is, the initial value of the filter coefficient), the shortest The direction of the convergence path is determined, and the actual convergence path is determined from the filter coefficient of the digital filter that has not converged to the optimum value (for example, the filter coefficient immediately after the start of control) and the filter coefficient of the digital filter at the start of control. Therefore, if the correction means corrects the processing content of the adaptive processing means so that the deviation between the directions is eliminated , the shortest convergence path and the actual convergence path can be obtained. The correction is performed so that the gap between the two is eliminated .

On the other hand, according to the invention of claim 5, in accordance with the reference signal output from the noise generation state detecting means,
Since a digital filter with a variable filter coefficient generates a signal for driving the control sound source, the control sound source generates a control sound that is correlated with the noise generated in the noise source. The noise is not necessarily reduced because it does not necessarily converge to the value.

However, in order to reduce the value of the predetermined evaluation function based on the residual noise detected by the residual noise detecting means,
Since the adaptive processing means updates the filter coefficient of the digital filter, the filter coefficient gradually converges to the optimum value, the control sound emitted from the control sound source cancels the noise, and the noise in the space is reduced. The shortest convergence path from the filter coefficient of the digital filter that converges to the optimum value to the optimum value is known, and the actual convergence path is known from the filter coefficient of the digital filter that has not converged to the optimum value. However, when the optimum value convergence determination means determines that the optimum value has converged to the optimum value, the shortest convergence path to the optimum value can be found from the filter coefficient of the digital filter, and the actual filter coefficient of the digital filter that has not converged to the optimum value can be found. Based on the convergence path , the processing contents of the adaptive processing means are adjusted so that the actual convergence path approaches the shortest convergence path.
Results urchin correction, the subsequent processing by the adaptive processing unit once again the optimum value is changed in accordance with the example the variation of the frequency of the noise is carried out, the convergence path of the filter coefficient of the digital filter is compared with a convergence path before correction Becomes closer to the shortest convergence path, and the filter coefficient converges to the optimum value relatively quickly.

Here, when the adaptive processing means has a transfer function model corresponding to the acoustic transfer characteristic between the control sound source and the residual noise detecting means, the main cause of the delay in the adaptive operation of the filter coefficient of the digital filter is actually That is, the acoustic transfer characteristics of the above change, and the accuracy of the transfer function model relatively decreases. Therefore, if the correction means corrects the transfer function model included in the adaptive processing means so that the actual convergence path approaches the shortest convergence path, the correction means for the shortest convergence path The deviation of the actual convergence path is accurately reflected in the correction of the processing content of the adaptive processing means.

Therefore, the transfer characteristic control means converges the frequency characteristic of the transfer function correction filter arranged in series with the transfer function model to the optimum value by the optimum value convergence determination. The actual convergence path is determined based on the filter coefficient of the digital filter when it is determined that
As a result of the change so as to approach the short convergence path, the deviation between the shortest convergence path and the actual convergence path is accurately reflected in the correction of the processing content of the adaptive processing means.

Similarly, the transfer characteristic control means of the transfer function correction filter arranged in series with the digital filter for generating the signal for driving the control sound source according to the reference signal. Frequency characteristics are actually converged
Is appropriately changed so as to approach the shortest convergent path, the deviation between the shortest convergent path and the actual convergent path is accurately reflected in the correction of the processing content of the adaptive processing means. In the invention according to claim 9, as a result of the gain adjusting means as the correcting means changing the update gain of the filter coefficient of the digital filter, as in the invention according to claim 2, for example, the shortest When the deviation of the actual convergence path with respect to the convergence path is large, the update gain can be reduced to prevent the deviation of the convergence path from increasing.

Further, in the invention described in claim 10, as in the invention described in claim 3, it is avoided that the processing content of the adaptive processing means changes based on unclear information. To be In the invention described in claim 11, as in the invention described in claim 4, the filter coefficient of the digital filter when the optimum value convergence determining means determines that the optimum value has converged to the optimum value and the control coefficient at the start of control The direction of the shortest convergence path that is determined from the filter coefficient of the digital filter (that is, the initial value of the filter coefficient), the filter coefficient of the digital filter that has not converged to the optimum value (for example, the filter coefficient immediately after the start of control), and at the start of control. Since the correction means corrects the processing contents of the adaptive processing means so as to eliminate the deviation from the direction of the actual convergence path determined by the filter coefficient of the digital filter of, the shortest convergence path and the actual convergence path. The correction is performed so that the deviation of

[0031]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a diagram showing an overall configuration of a first embodiment of the present invention, which is an active system for reducing muffled noise transmitted from an engine 4 as a noise source into a vehicle interior 6 as a space. The present invention is applied to the mold noise control device 1.

First, the structure will be described. The vehicle body 3 is supported by front wheels 2a, 2b, rear wheels 2c, 2d, and suspensions interposed between the wheels 2a to 2d and the vehicle body 3. The vehicle shown in FIG. 1 is a so-called front-mounted engine front-wheel drive vehicle in which the front wheels 2a and 2b are rotationally driven by an engine 4 arranged in the front part of the vehicle body 3. The engine 4 is provided with a crank angle sensor 5 as a noise generation state detecting means.
Supplies a reference signal x, which is a pulse signal corresponding to the crank angle of the engine 4 (for example, in the case of a reciprocating 4-cylinder, one pulse every 180 degrees) to the controller 10.

In the vehicle interior 6 of the vehicle body 3, a control actuator is installed.
A loudspeaker 7a as a chute or a control sound source,
7b, 7c and 7d are front seats S₁, S₂And rear seat
Seat S ₃, S_FourPlaced on the doors facing each of the
There is. Furthermore, each seat S₁~ S_FourAt the headrest position
Are microphones 8a to 8 as residual noise detecting means.
There are two h each, and these microphones
Residual noise signal measured as sound pressure by Rophone 8a-8h
e₁~ E₈Are supplied to the controller 10.

Then, the controller 10 executes the arithmetic processing described later based on the reference signal x supplied from the crank angle sensor 5 and the residual noise signals e _{1 to} e ₈ supplied from the microphones 8a to 8h. The drive signals y _{1 to} y ₄ are output to the loudspeakers 7a to 7d so that the loudspeakers 7a to 7d generate a control sound for canceling the muffled sound transmitted to the vehicle interior 6.

The controller 10 is, as shown in FIG.
An A / D converter 11 for converting the reference signal x supplied from the crank angle sensor 5 into a digital value, and a digital filter 12 to which the reference signal x converted into a digital value by the A / D converter 11 is input. , An adaptive digital filter 13 as a filter filter variable digital filter to which the reference signal x is input, and a microphone 8a.
An A / D converter 15a~15h for converting the residual noise signal e ₁ to e ₈ amplified by the supplied and amplifier 14a~14h to a digital value from ～8H, processed signal r _lm which is filtered by the digital filter 12 And A / D
A microprocessor 16 for executing a predetermined arithmetic processing based on the residual noise signals e _{1 to} e ₈ converted into digital values by the converters 15a to 15h to update the filter coefficient of the adaptive digital filter 13, and the adaptive digital filter 13 output from the drive signal y ₁ ~y ₄ is converted into an analog value and supplies the amplifier 18a~18d driving the loudspeaker 7a to 7d D / a converter 17a~1
7d.

Here, the digital filter 12 is
Of the speaker 7a-7d and the microphone 8a-8h
The transfer function between two models in the form of a finite impulse response function
Digital filter C as a transfer function model
_lm(L = 1, 2, ..., L, m = 1, 2, ..., M)
(In this embodiment, M = 4) loudspeakers 7a to 7
d and L (L = 8 in this embodiment) microphones
8a-8h for all combinations (L × M)
And the reference signal x is converted to those digital filters C _lmso
The filtered value r_lmIs generated and output.

On the other hand, the adaptive digital filter 13 has M adaptive filter filters W _m with variable filter coefficients corresponding to the number of loudspeakers 7a to 7d, and the reference signal x is adapted to the adaptive digital filter W _m. The drive signals y _{1 to} y ₄ are generated and output by performing the filtering process with. Then, the microprocessor 16 receives the processed signal r _lm supplied from the digital filter 12,
According to the residual noise signals e _{1 to} e ₈ supplied from the microphones 8a to 8h, the filter coefficient of each adaptive digital filter W _m of the adaptive digital filter 13 is set to LM.
Update based on S algorithm.

Here, the LMS algorithm is one of the algorithms suitable for updating the filter coefficient of the adaptive digital filter, and is the l-th (l = 1, 2, ...,).
L: L = 8) the residual noise signal detected by the microphone is e _l (n), the residual noise signal detected by the l-th microphone when no control sound is generated from the loudspeaker is e _l (n), m-th (m = 1, 2, ..., M: M =
J-th digital filter C _lm of the transfer function between the loudspeaker and the l-th microphone modeled in the form of a finite impulse response function 4) (j = 0,1,2,
, I _C −1: I _C is a constant), and the filter coefficient is C _lmj ,
X (n) is the reference signal, and m is the reference signal x (n)
The i-th (i = 0, 1, 2, ..., I _K −1: I _{K of} the adaptive digital filter W _m for driving the th loudspeaker
Is a constant) and the filter coefficient is W _mi , Is established.

The terms with (n) are all sump
Represents the sampled value at ring time n, and I_CIs
Filter C_lmNumber of taps (filter order), I_KIs adapted
Digital filter W_mBy the number of taps (filter order)
is there. In the above formula (1), “ΣW” on the right side_mix (n-j-
i) ”is the reference signal x to the adaptive digital filter.
Output y when (n) is input_m(N) is represented by “ΣC
_lmj{ΣW_mix (n-j-i)} ”is the m-th Lau
Signal y input to the speaker _m(N) controls from there
Transfer function C is output to the space as sound_lmThrough the l-th
It shows the signal when it reaches the icrophone, and
ΣC_lmj{ΣW_mix (n−j−i)} ”is the l-th term
Is the signal reaching the microphone added up?
The sum of control sounds reaching the l-th microphone
It represents.

Next, the evaluation function Je is And

The LMS algorithm finds the filter coefficient W _mi that minimizes the evaluation function Je. Specifically, the filter coefficient W _{mi is} a value obtained by partially differentiating the evaluation function Je with respect to each filter coefficient W _mi. To update. Therefore, from equation (2) above, From equation (1) above, Therefore, if the right side of the equation (4) is set to r _lm (n−i), the filter coefficient is updated by the following equation (5).

[0042] Here, the convergence coefficient α as the update gain is a coefficient relating to the speed at which the filter converges optimally and its stability.

That is, the microprocessor 16 updates the filter coefficient W _mi of the adaptive digital filter W _m based on the above equation (5). Further, in this embodiment, each digital filter C _lm constituting the digital filter 12 is set to an order larger than the order I _C necessary for the operation of the LMS algorithm, and the filter coefficient C _lmj can be changed by the microprocessor 16. Therefore, the lead / lag of each digital filter C _lm can be arbitrarily adjusted.

Here, it is assumed that the noise generated in the engine 4 has a single frequency and the filter order of the adaptive digital filter W _m is the second order (I _K = 2), and the adaptive digital filter W in this case is filter coefficients of _m W _m0, W _m1
The state of convergence to the optimum value of is described with reference to FIG. Note that FIG. 3 is an orthogonal coordinate system in which the filter coefficient at the start of control is the origin, and the coordinates corresponding to the optimum value are (W _m00 , W
_m10 ).

That is, if each digital filter C _lm accurately represents the acoustic transfer characteristic between the loudspeakers 7a to 7d and the microphones 8a to 8h, the convergence path of the filter coefficients W _m0 and W _m1 starting from the origin is , A, the shortest route that goes straight to the optimum point (W _m00 , W _m10 ) should be taken. However, changes in conditions such as temperature and humidity in the passenger compartment 6, increase / decrease in occupants, or the loudspeaker 7
a to 7d and microphones 8a to 8h are deteriorated,
Loudspeakers 7a-7d and microphones 8a-8
The acoustic transfer characteristic between h varies and each digital filter C
_When the accuracy of _lm relatively decreases, the transfer function represented by the digital filter C _lm has a phase error with respect to the transfer function of the actual acoustic space, and the optimum point (W _m00 , W
It is affected by the wrong optimum points (W _m00 ', W _m10 ') different from _m10 ), and as shown by B, the optimum convergence path A
While drawing a curved trajectory that deviates greatly from the optimum point (W
_m00 , W _m10 ), and even if the convergence is made to the optimum value, it takes a relatively long time to reach it.

Therefore, in the present embodiment, the microprocessor 16 executes the arithmetic processing described later to obtain the phase error between the transfer function represented by the digital filter C _lm and the actual transfer function in the acoustic space, and the phase error is obtained. The filter coefficient C _lmj of the digital filter C _lm is set so that the error is eliminated.
Shall be changed. FIG. 4 is a flow chart showing the outline of the processing executed in the controller 10. The operation of this embodiment will be described below with reference to FIG.

First, in step 001, it is determined whether or not the filter coefficient W _mi has converged to the optimum value. In this step 001, for example, when the update amount of the filter coefficient W _mi based on the above equation (5) does not exceed the predetermined value a predetermined number of times or more, the filter coefficient W _mi converges to the optimum value. To determine. When it is determined in step 001 that the filter coefficient W _mi has not yet converged to the optimum value, the process proceeds to step 002, the filter coefficient W _mi is updated based on the equation (5), and then step 00
3 proceeds to generate a drive signal y ₁ ~y ₄ outputs to the loudspeaker 7a to 7d.

[0048] Then, the control sound into the passenger compartment 6 from loudspeakers 7a~7d occurs immediately after start of control is not necessarily the filter coefficient W _mi of the adaptive digital filter W _m is converged to an optimum value Therefore, it cannot be said that the muffled sound transmitted to the vehicle interior 6 is necessarily reduced. However, as a result of repeatedly executing the processing of steps 001 to 003, each filter coefficient W _mi converges toward the optimum value, and the muffled sound transmitted into the vehicle interior 6 is transmitted to the loudspeaker 7.
It is canceled by the control sounds emitted from a to 7d, and the noise in the vehicle interior 6 is reduced.

Then, the filter coefficient W_miConverges to the optimal value
Then, the determination in step 001 is “YES”.
To step 004, the filter section at that time
Number, that is, the optimum value (W_m00, W_m10) Is remembered. Na
For convenience of explanation, the adaptive digital filter W_mThe fi
Ruta order is quadratic. Then move to step 005
And again from the initial value, the adaptive digital filter W _mThe fi
Coefficient W_miIs updated a predetermined number of times, and immediately after the start of control
Adaptive digital filter W in_mFilter coefficient (W
_m01, W_m11).

This filter coefficient (W _m01 , W _m11 ) is
Since the point is near the origin of the actual convergence path B shown in FIG. 3, the direction from the origin to the filter coefficient (W _m01 , W _m11 ) is incorrect from the origin to the optimum point (W _m00 ', W _m00 ).
There is no big difference even if it agrees with the direction toward _m10 '). The direction from the origin to the optimum point (W _m00 , W _m10 ) and the wrong optimum point (W _m00 ', W _m from the origin)
The deviation from the direction toward _m10 ') is due to the phase error between the transfer function represented by the digital filter C _lm and the actual transfer function in the acoustic space, and _therefore the deviation from those directions. If the digital filter C _lm is corrected based on, the phase error between transfer functions can be reduced.

Therefore, the _{process proceeds} to step 006, and the optimum point (W _m00 , W _m00 ,
The angle Δθ between the direction toward W _m10 ) and the direction from the origin toward the filter coefficient (W _m01 , W _m11 ) immediately after the start of control is calculated. However, the filter coefficient at the origin (at the start of control) is 0. _{^{Δθ = cos -1 {(W m00}} W m01 + W m10 W m11) / ((W m00 2 + W m10 2) 1/2 (W m01 2 + W m11 2) 1/2)} ...... (6) Then, step 007, the filter coefficient C of the digital filter C _lm is advanced or delayed by an amount corresponding to the angle Δθ obtained in step 006.
_{Adjust lmj} accordingly.

That is, as shown in FIG. 5, if the phase error (∝ angle Δθ) at an arbitrary frequency is known, the phase error in the entire frequency range can be estimated, so that the digital filter C _lm can eliminate the phase error. Adjust the phase.
As a result of correcting the digital filter C _lm in this way, the phase error between the transfer function represented by the digital filter C _lm and the transfer function of the actual acoustic space becomes small, and thereafter, for example, with the fluctuation of the frequency of noise. When the optimum value changes and the filter coefficient W _mi of the adaptive digital filter W _m is updated again, the filter coefficient W _mi passes through the vicinity of the shortest convergence path as shown in A of FIG. The value converges to the value, and quick adaptive operation becomes possible.

The main cause of the delay in the adaptive operation of the adaptive digital filter W _m is that the digital filter 12 is changed as a result of the change in the acoustic transfer characteristics between the loudspeakers 7a to 7d and the microphones 8a to 8h in the real space.
Therefore, if the configuration is such that the filter coefficient C _lmj of the digital filter C _lm is appropriately changed as in the present embodiment, the change of the acoustic transfer characteristics can be controlled accurately. Can be reflected in.

Moreover, since it is not necessary to obtain a plurality of transfer functions in advance in consideration of changes in the acoustic transfer characteristics and store the transfer functions, the storage capacity is not particularly increased, which is practical. Since it is not necessary to generate the identification sound, the occupant is not uncomfortable. Note that step 005
The normal noise reduction control of steps 002 and 003 is stopped when the processing of step 00 is executed.
The processing of 4 to 007 is performed, for example, in steps 002 and 003.
There is no particular problem because the deterioration of the noise reduction control can be suppressed to an extremely low level by setting the control content to be executed once or twice for 100 loops.

In this embodiment, the digital filter is used.
C_lmRequired filter order for LMS algorithm operation
Na degree I_CBy setting each digit to a higher degree than
Le Filter C_lmTo be able to arbitrarily adjust the lead / lag of
However, for example, the transfer function initially set is exp
Convolve (Δθ (ω)) (ω is the frequency where the phase error occurs)
The transfer function is recalculated by inserting the
Digital filter C _lmProceed by replacing
The delay may be adjusted arbitrarily.

Here, in this embodiment, the digital filter 12, the microprocessor 16 and the step 002 are used.
The processing of step 003 constitutes the adaptive processing means, the processing of step 001 corresponds to the optimum value convergence determination means, and the processing of steps 004 to 007 constitutes the correction means. FIGS. 6 to 8 are views showing a second embodiment of the present invention. The present invention relates to an active noise control device for reducing muffled noise transmitted from an engine into a vehicle compartment as in the first embodiment. Is applied.

That is, in this embodiment, as shown in FIG.
A transfer function correction filter 20 having a frequency characteristic as shown in FIG. 7 and a variable frequency characteristic is interposed between the digital filter 12 and the microprocessor 16 in a serial relationship with the digital filter 12. The frequency characteristic of the transfer function correction filter 20 can be controlled by the microprocessor 16.

This transfer function correction filter 20 is shown in FIG.
(A), the phase between the predetermined frequency f _a ~f _b can be controlled to an arbitrary value, and is a filter having the characteristic of not amplitude was varied as shown in FIG. 7 (b) . That is, by supplying the processed signal r _lm output from the digital filter 12 to the microprocessor 16 through the transfer function correction filter 20, the frequency f
Only the phase of the processed signal r _lm between _{a and} f _b changes, and as a result, the processing content of the digital filter 12 is corrected.

FIG. 8 is a flow chart showing an outline of processing in the controller 10. The operation of this embodiment will be described below with reference to FIG. Since the processing of steps 001 to 006 shown in FIG. 8 is the same as that of the first embodiment, the duplicated description will be omitted. That is, step 0
When the angle Δθ is obtained in 06, the process _proceeds to step 010, and the processed signal r output from the digital filter C _{lm is} output.
The filter coefficient of the transfer function correction filter is changed so that the phase of _{lm leads} or lags by an amount corresponding to the angle Δθ. In step 010, the frequencies f _a and f _b shown in FIG. 7A are also changed according to the noise frequency.

As a result, the phase error between the transfer function represented by the digital filter C _lm and the transfer function correction filter 20 and the transfer function of the actual acoustic space becomes small, so that the same effect as the first embodiment can be obtained. can get. Here, in this embodiment, the microprocessor 16 and step 007 are used.
The transfer characteristic control means is constituted by the processing of.

FIG. 9 is a diagram showing a third embodiment of the present invention. In this embodiment as well, similar to the first embodiment, the active noise for reducing the muffled noise transmitted from the engine into the vehicle compartment is used. The present invention is applied to a control device. That is, this embodiment is the same as the above-mentioned second embodiment except that the transfer function correction filter 20 is arranged in series with the adaptive digital filter 13 instead of the next stage of the digital filter 12. .

Even with such a configuration, the drive signals y _{1 to} y ₄
Is corrected according to the phase error between the transfer function represented by the digital filter C _lm and the transfer function of the actual acoustic space, so that the same effect as the first embodiment can be obtained. FIG. 10 is a diagram showing a fourth embodiment of the present invention. This embodiment is also applied to an active noise control device for reducing muffled noise transmitted from the engine into the vehicle compartment as in the first embodiment. The invention is applied.

FIG. 10 is a flow chart showing the outline of the processing in the controller 10.
Since the processing of 006 is the same as that of the first embodiment, the duplicated description will be omitted. That is, in this embodiment, when the angle Δθ is obtained in step 006, step 011
Then, the convergence coefficient α is corrected according to the magnitude of the angle Δθ.

For example, a large angle Δθ means that
Since the phase error between the transfer function represented by the digital filter C _lm and the transfer function of the actual acoustic space is large, if the convergence coefficient α is large, the bulge of the path B shown in FIG. 3 is further increased. Since it takes more time to converge to the optimum value, it is desirable to reduce the convergence coefficient α.

On the contrary, the small angle Δθ means that the phase error between the transfer function represented by the digital filter C _lm and the transfer function in the actual acoustic space is small, so that the convergence coefficient α is increased. Then, the adaptive operation to the optimum value can be performed in a shorter time. Here, in this embodiment, the processing of step 011 corresponds to the gain adjusting means.

Although not shown in particular, both effects can be obtained by executing both the correction of the convergence coefficient α and the correction of the digital filter C _lm described in the first to third embodiments. . FIG. 11 is a diagram showing a fifth embodiment of the present invention, and the configuration of the present embodiment is the same as that of the above-mentioned first embodiment except that the processing of step 012 is performed before step 004.

That is, in this embodiment, if it is determined "YES" in step 001, the process proceeds to step 012,
It is determined whether or not the optimum value of the filter coefficient W _mi of the adaptive digital filter W _m is fluctuating, and only when it is determined that the optimum value is not fluctuating, the processing from step 004 is executed. The process after step 004 is executed as an interrupt process for every 100 loops with respect to the normal noise reduction control to correct the digital filter C _lm , but as shown in the above formula (6), Since the phase error is obtained based on the optimum value of the coefficient W _mi, the optimum value of the filter coefficient W _mi changes,
For example, even if the digital filter C _lm is corrected in a situation where the temperature inside the vehicle compartment 6 changes sharply immediately after the air conditioner is operated, a good effect cannot be expected.

Therefore, if it is determined in step 012 that the optimum value fluctuates, the process returns to step 002 and the normal noise reduction control is executed.
Other functions and effects are similar to those of the first embodiment. In each of the above embodiments, after it is determined that the filter coefficient W _mi has converged to the optimum value, the filter coefficient W _mi is updated again from the initial value in step 005 several times, and the filter coefficient (W _m01 , W _m11 ), but when the situation change is not so drastic, the filter coefficient W _mi immediately after the start of the noise reduction control immediately before the processing of step 004 and thereafter is stored, and The angle Δθ may be calculated based on the filter coefficient W _mi ,
By doing so, there is an advantage that the processing of step 005 becomes unnecessary and the correction processing of the digital filter 12 can be performed in parallel with the normal noise reduction control.

In each of the above embodiments, the case where the active noise control device according to the present invention is applied to the device for reducing the noise transmitted to the vehicle interior 6 has been described. May be a device for reducing the above. Further, in each of the above embodiments, the correction process is performed based on the angle Δθ obtained by the above equation (6), but the present invention is not limited to this, and for example, the route B shown in FIG.
The correction process may be performed on the basis of the curvature, the area surrounded by the paths A and B, and the like.

In each of the above-described embodiments, the case where the muffled noise transmitted from the engine 4 into the vehicle interior 6 is reduced has been described, but the noise that can be reduced by the present invention is not limited to this. Instead, for example, pickup signals for suspension vibration that correlates with road noise, pickup signals for wind noise near door mirrors, pickup signals for case vibration of differential gears and transmission gears (such as noise caused by case vibration of the driving force transmission system). Correlation signals) and pulse signals (signals that correlate with noise due to meshing of differential gears and transmission) according to rotation of the output shaft of the vehicle speed measurement transmission can be reduced It is also possible to use a device for

Further, in the above embodiment, the case where the adaptive control device according to the present invention is applied to the active noise control device for a vehicle is explained, but the application target of the present invention is to a device for reducing noise. The present invention is not limited to this, and other adaptive control systems can be applied. Then, in the above embodiment, four loudspeakers 7a to
7d and eight microphones 8a to 8h as residual noise detecting means are provided, but the number of these is arbitrary.

Further, in the above embodiment, the LMS algorithm in the time domain is applied as the algorithm for updating the filter coefficient W _mi of the adaptive digital filter W _m , but the present invention is not limited to this and, for example, in the frequency domain. LMS algorithm etc. may be applied.

[0073]

As described above, according to the present invention,
The processing contents of the adaptive processing means based on the shortest convergence path to the optimum value found from the filter coefficient of the digital filter when converged to the optimum value and the actual convergence path found from the filter coefficient before convergence to the optimum value. The actual convergence path is the shortest
Since the configuration is such that the correction is performed so as to approach the convergence path of the above, it is adapted based on the change in the condition of the control target and the deterioration of the component parts of the device without causing the trouble of increasing the necessary storage capacity and causing discomfort to humans. Even if the processing content of the processing unit does not match the actual control target or the characteristics of the device, the filter coefficient of the digital filter can be quickly converged to the optimum value.

[Brief description of drawings]

FIG. 1 is a diagram showing an overall configuration of a first embodiment.

FIG. 2 is a block diagram showing a configuration of a controller in the first embodiment.

FIG. 3 is a diagram showing how filter coefficients of an adaptive digital filter are adapted.

FIG. 4 is a flowchart showing an outline of processing in the first embodiment.

FIG. 5 is a diagram illustrating a relationship between a phase error and an angle Δθ.

FIG. 6 is a block diagram showing a configuration of a second embodiment.

FIG. 7 is a diagram showing frequency characteristics of a transfer function correction filter.

FIG. 8 is a flowchart showing an outline of processing in the second embodiment.

FIG. 9 is a block diagram showing a configuration of a third embodiment.

FIG. 10 is a flowchart showing an outline of processing in a fourth embodiment.

FIG. 11 is a flowchart showing an outline of processing in the fifth embodiment.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 Active noise control device (adaptive control device) 4 Engine (noise source) 5 Crank angle sensor (noise generation state detection means) 6 Cabin (space) 7a-7d Loudspeaker (control actuator, control sound source) 8a-8d Microphone (Residual noise detection means) 10 Controller 12 Digital filter (adaptive processing means, transfer function model) 13 Adaptive digital filter (digital filter with variable filter coefficient) 16 Microprocessor (adaptive processing means) 20 Transfer function correction filter

Claims (11)

(57) [Claims] 1. A digital filter having a variable filter coefficient for generating a signal for driving a control actuator according to a predetermined reference signal, and a filter coefficient of the digital filter so that the value of a predetermined evaluation function matches a target value. In an adaptive control device including an adaptive processing unit for updating, an optimum value convergence determining unit that determines whether or not the filter coefficient of the digital filter has converged to an optimum value, and the optimum value convergence determining unit sets the optimum value to an optimum value. The maximum that can be known from the filter coefficient of the digital filter when it is determined that it has converged.
The shortest convergence path to an appropriate value, and the actual fact that can be understood from the filter coefficient of the digital filter that has not converged to the optimum value
Based on the convergence path, the actual convergence path the shortest
An adaptive control device for correcting the processing content of the adaptive processing means so as to approach the convergent path . 2. The correction means is such that the actual convergence path is the
2. The adaptive control device according to claim 1, further comprising gain adjusting means for changing the update gain of the filter coefficient of the digital filter so as to approach the shortest convergence path . 3. The correction means does not correct the processing content of the adaptive processing means when the filter coefficient of the digital filter changes when the optimum value convergence determination means determines that the optimum value has converged to the optimum value. 1 or the adaptive control device according to claim 2. 4. The correction means converges to an optimum value and a direction determined by the filter coefficient of the digital filter when the optimum value convergence determination means determines that the optimum value has converged to the optimum value and the filter coefficient of the digital filter at the start of control. 4. The processing content of the adaptive processing means is corrected so that the deviation between the filter coefficient of the digital filter which is not performed and the direction determined by the filter coefficient of the digital filter at the start of control is eliminated. An adaptive control device according to claim 1. 5. A control sound source capable of generating a control sound in a space in which noise is transmitted from a noise source, a noise generation state detecting means for detecting a noise generation state of the noise source and outputting it as a reference signal, and the reference signal. A digital filter having a variable filter coefficient for generating a signal for driving the control sound source according to
Residual noise detecting means for detecting residual noise at a predetermined position in the space, and adaptive processing means for updating the filter coefficient of the digital filter so that the value of a predetermined evaluation function based on the residual noise is reduced. In an active noise control device provided with, an optimum value convergence determination means for determining whether or not the filter coefficient of the digital filter has converged to an optimum value, and when it is determined that this optimum value convergence determination means has converged to an optimum value. most apparent from the filter coefficients of the digital filter
The shortest convergence path to an appropriate value, and the actual fact that can be understood from the filter coefficient of the digital filter that has not converged to the optimum value
Based on the convergence path, the actual convergence path the shortest
And a correction unit that corrects the processing content of the adaptive processing unit so as to approach the convergence path of the active noise control device. 6. The adaptive processing means has a transfer function model corresponding to the acoustic transfer characteristic between the control sound source and the residual noise detection means, and the correction means has the actual convergence path with the shortest convergence.
The active noise control device according to claim 5 , wherein the transfer function model is corrected so as to approach the bundle path . 7. The correction means includes a transfer function correction filter arranged in series with a transfer function model included in the adaptive processing means and having a variable transfer characteristic, and an optimum value convergence judgment means when it judges that the optimum value converges to an optimum value. on the basis of the filter coefficient of the digital filter in the filter coefficient and the optimum value it does not converge to the digital filter, the actual convergence path before
7. The active noise control device according to claim 6 , further comprising a transfer characteristic control unit that changes a transfer characteristic of the transfer function correction filter so as to approach the shortest convergence path . 8. The correction means comprises a transfer function correction filter having a variable transfer characteristic arranged in series with a digital filter,
Based on the filter coefficient of the digital filter when the optimum value convergence determination means determines that it has converged to the optimum value and the filter coefficient of the digital filter that has not converged to the optimum value, the actual convergence path is the shortest convergence. On the route
The active noise control device according to claim 6 , further comprising a transfer characteristic control unit that changes a transfer characteristic of the transfer function correction filter so as to approach . 9. The correction means is such that the actual convergence path is the
As approached, the shortest convergence path, active noise control apparatus according to claim 5, further comprising a gain adjustment means for changing the update gain of the filter coefficient of the digital filter. 10. The correction means does not correct the processing content of the adaptive processing means when the filter coefficient of the digital filter changes when the optimum value convergence determination means determines that the optimum value has converged to the optimum value. The active noise control device according to any one of claims 5 to 9. 11. The correcting means determines the direction determined by the filter coefficient of the digital filter when the optimum value convergence determining means determines that the optimum value has converged to the optimum value and the filter coefficient of the digital filter at the start of control, and immediately after the start of control. 11. The processing content of the adaptive processing means is corrected so that there is no deviation between the filter coefficient of the digital filter and the direction determined by the filter coefficient of the digital filter at the start of control. Active noise control device.