JPH08166788A - Active noise control device and active type vibration control device - Google Patents

Abstract

PURPOSE: To perform good noise reduction control and vibration reduction control by selecting the absolutely minimum reference signals out of plural signals indicating an occurrence state of noise and vibration. CONSTITUTION: A controller 20 performing noise reduction control is provided with a reference signal selecting section 21 selecting and outputting a signal being a main factor of a noise out of acceleration detecting signals Vt as a reference signal XK, a driving signal generation section 22 convoluting the reference signal XK and each filter coefficient Wkmi of a adaptive digital filter Wkm and adding them for each subscript (m) and generating and outputting a driving signal ym , an update reference signal operation section 23 operating an update reference signal rklm based on the reference signal Xk and a transfer function filter CΛlm , and a filter coefficient updating section 24 updating the filter coefficient Wkmi based on the update reference signal rklm and a residual noise signal el .

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an active noise control device for reducing noise by interfering a control sound generated from a control sound source with a noise transmitted from a noise source and a vibration transmitted from the vibration source. The present invention relates to an active vibration control device for reducing vibration by interfering with control vibration generated from a control vibration source, and particularly to a drive signal for driving a control sound source and a control vibration source based on a signal indicating a noise and vibration generation state. In an active noise control device and an active vibration control device adapted to generate a noise, it is possible to efficiently select a signal representing the generation state of the noise and vibration, and thus a good noise with a minimum necessary calculation load. This is to reduce the vibration and vibration.

[0002]

2. Description of the Related Art A conventional technique of this type is disclosed in, for example, Japanese Patent Laid-Open No. 4-43799. This conventional device is applied in order to reduce noise in a general room or an interior of an automobile, and includes a plurality of noise detectors that detect noise from a plurality of noise sources, and those noise detectors. A plurality of adaptive digital filters that adaptively control the noise signal detected by the detector, a plurality of speakers that reproduce the output of the adaptive digital filter, an error detector that detects noise at a predetermined listening point, and an error detection The calculation circuit that calculates independent noise systems for multiple noise sources based on the output of the detector, the extractor that extracts each noise signal by the output of the calculation circuit, and the output signal of the error detector are minimized And a signal processing circuit that performs signal processing as described above. The signal processing circuit controls the filter coefficient of the adaptive digital filter, and the output of the signal processing circuit is an arithmetic circuit. Is inputted, the arithmetic circuit was adapted to separate each of the noise signal a plurality of noise sources.

With such a structure, despite the noise coming from a plurality of noise sources at the listening point, each noise is separated / extracted and the noise control is performed independently to reduce the noise comprehensively. It was possible.

[0004]

Considering a case where the above-mentioned active noise control device is actually used,
It is necessary to detect the generation state of noise in a noise source by, for example, a noise detection microphone, but as a practical matter, the true noise source is not always known, and the number of true noise sources and the number of noise sources The relationship and size (contribution rate to noise in the space) are often unknown.

Consider, for example, an active noise control system for reducing road noise transmitted to the passenger compartment of a vehicle.
Road noise is caused by the vibration that is input from the road surface to the suspension side when the wheels pass through various irregularities on the road surface. Therefore, the true noise source is input to the tire from the road surface. Inferred to be "power". Road noise, which the passengers hear in the passenger compartment, occurs when the force that is the true noise source is transmitted to the vehicle body, vibrates the vehicle body panel, and is radiated as "sound" into the vehicle interior space. To do.

Therefore, in order to detect the noise generation state of a true noise source as some kind of physical quantity, it is necessary to detect vibration, noise or force at any position between the tire and the vehicle interior. . However, in this case,
In order to establish causality with a random signal such as road noise, the control sound (secondary sound) generated from the control sound source (speaker for noise control) is a microphone (error) for detecting residual noise at the listening point. It is necessary to obtain a signal (reference signal) serving as information of the noise source at a position farther in time of propagation than the time of propagation to the detection microphone). That is, if the noise generated as a sound in the vehicle interior space is detected and used as a signal representing the noise generation state of the noise source, it is presumed that the causality is not established.

Therefore, the vibration of the vehicle body is used as a signal representing the noise generation state of the noise source.
Vibration of the wheel in three degrees of freedom (x direction, y direction, and z direction orthogonal to each other) is considered as a reference signal candidate, but such a reference signal candidate is also not a true noise source. Since there is a correlation between the candidates of each reference signal, and the information of the true noise source is contained in duplicate in those reference signals, it is usually necessary to aggregate all the signals before the true noise source. All information will be obtained (however, depending on the number of measurement of reference signals and measurement positions, it may not be possible to obtain information on all true noise sources even if all reference signals are aggregated).

If all the candidates of the reference signal are used for noise reduction control, for example, in the above example, if there is one speaker, then 4 (wheels) × 3 (degrees of freedom; excluding rotational degrees of freedom). = 12, 12 adaptive digital filters are required. Normally, a plurality of speakers are required to correspond to the sound flow in the entire vehicle interior, so if speakers are arranged at the four corners of the vehicle interior space, for example, 4 (wheels) ×
3 (degree of freedom) × 4 (speaker) = 48, and 48 adaptive digital filters are required.

The increase in the number of adaptive digital filters means that the number of multiplications and additions in the convolution operation for generating the drive signal is increased correspondingly, and that the update processing of the filter coefficient of the adaptive digital filter is increased. The amount of calculation will increase accordingly.
In other words, an increase in the number of reference signals used for control leads to an increase in the calculation load on the controller, which requires a high-speed processing and thus an expensive calculation processor.

For such a problem, for example, only a signal having a high correlation with the noise in the vehicle compartment is selected from a plurality of reference signal candidates, and noise reduction control is performed based on the selected reference signal. It is also possible to carry out. In particular,
It is conceivable to calculate the coherence between each of the plurality of reference signal candidates and the residual noise signal and select the reference signal having the large coherence.

However, if the reference signal is selected only on the basis of the magnitude of such coherence, when there are a plurality of true noise sources, a large amount of information of a part of these true noise sources is included. I chose only the reference signal,
Information on other true noise sources may be discarded. In order to solve such a problem, coherence between the residual noise signal and some selected reference signals (also referred to as multiple coherence. It is the coherence with the noise signal, and the duplication of information is taken into consideration by the calculation.), And the minimum required combination of reference signals that maximizes the multiple coherence is found, and the found reference is found. It is also possible to perform noise reduction control based on the signal. However, even if this multiple coherence is calculated, the work of selecting the reference signal is a trial and error process, and the number of reference signals used for noise reduction control must be determined by a trial and error process. Then, for many combinations, multiple
The coherence must be calculated, and the calculation load becomes too large. For example, if the candidate of the reference signal as mentioned above is 12, since the their reference signals to each other, there is a combination of _{12 C 1 + 12 C 2 +} 12 C 3 + ... + 12 C 12 = 4270 ways, this If is selected at the start of the noise reduction control, then 4270 multiple coherence calculations are required.

When the calculation load becomes excessive, the time until the actual noise reduction control is executed for selecting the reference signal becomes long, and the noise in the vehicle compartment cannot be reduced during that time, and the occupant cannot be reduced. Can be uncomfortable with. Further, such a problem is not limited to the active noise control device whose noise is to be reduced, but is the active vibration control device which reduces the vibration which is only the difference between the sound and the transmission medium. The same is true.

The present invention has been made by paying attention to the unsolved problems of the conventional active noise control device and active vibration control device as described above. An object of the present invention is to provide an active noise control device and an active vibration control device capable of selecting a minimum reference signal required for performing noise reduction control and vibration reduction control from among reference signal candidates.

[0014]

In order to achieve the above object, an active noise control system according to the invention of claim 1 is
A control sound source capable of generating a control sound that interferes with noise emitted from a noise source, and a plurality of noise generation state detection means for detecting the generation state of the noise and outputting it as a noise generation state signal,
Reference signal selection means for selecting a part of the plurality of noise generation state signals output from the plurality of noise generation state detection means as a reference signal, and noise after the interference is reduced based on the reference signal. Active control means for controlling the control sound source, and the reference signal selection means selects the noise generation state signal which is a main factor of the noise as the reference signal.

In order to achieve the above object, an active noise control device according to a second aspect of the present invention includes a control sound source capable of generating a control sound that interferes with noise emitted from a noise source, and a control sound source after the interference. Residual noise detection means for detecting residual noise and outputting it as a residual noise signal, a plurality of noise generation state detection means for detecting the noise generation state and outputting it as a noise generation state signal, and a plurality of noise generation state detection means Reference signal selecting means for selecting a part of the plurality of noise generation state signals to be output as a reference signal, and the control sound source for reducing the noise after the interference based on the reference signal and the residual noise signal. And active control means for controlling,
The reference signal selection means selects the noise generation state signal, which is the main factor of the noise, as the reference signal.

According to a third aspect of the present invention, in the active noise control device according to the second aspect of the invention, the active control means drives the control sound source by filtering the reference signal. An adaptive digital filter for generating a signal; and an adaptive processing means for updating the filter coefficient of the adaptive digital filter according to an adaptive algorithm so as to reduce the noise after the interference based on the reference signal and the residual noise signal. I did it.

According to a fourth aspect of the invention, in the active noise control device according to the first to third aspects of the invention, the reference signal selecting means selects a virtual noise source based on the noise generation state signal. Virtual noise source calculation means for calculating,
Correlation calculation means for calculating a correlation between the virtual noise source and the plurality of noise generation state signals, and the noise generation state signal, which is the main factor, based on the calculation result of the correlation calculation means, the reference signal I chose to choose it.

On the other hand, the invention according to claim 5 is the active noise control device according to claim 2 or 3, wherein the reference signal selecting means selects a virtual noise source based on the noise generation state signal. Virtual noise source calculation means for calculating,
Virtual noise source identification means for identifying a main virtual noise source based on the correlation between the virtual noise source and the residual noise signal, and calculating the correlation between the identified virtual noise source and the plurality of noise generation state signals Correlation calculation means is provided, and the noise generation state signal, which is the main factor, is specified as the reference signal based on the calculation result of the correlation calculation means.

According to a sixth aspect of the present invention, in the active noise control device according to the fifth aspect of the invention, the reference signal selecting means sets the number of the selected reference signals to the specified virtual signal. The number is the same as the number of noise sources. The invention according to claim 7 is the active noise control device according to claim 5 or 6, wherein the virtual noise source identification means determines the correlation value between the virtual noise source and the residual noise signal. A virtual noise source of a combination whose sum is a predetermined value or more is specified as the main virtual noise source.

Further, the invention according to claim 8 is the active noise control device according to any of claims 4 to 7, wherein the reference signal selecting means has a large correlation value calculated by the correlation calculating means. The noise generation state signal is sequentially selected as the reference signal. The invention according to claim 9 is the active noise control device according to any one of claims 4 to 8, wherein the reference signal selecting means transmits between the virtual noise source and the noise generation state signal. A determinant calculating means for calculating the determinant of a function matrix is provided,
In consideration of the determinant, the noise generation state signal, which is the main factor, is selected as the reference signal.

According to a tenth aspect of the present invention, in the active noise control device according to the second to ninth aspects, the reference signal selecting means gives priority to a frequency band having a high level of the residual noise signal. Then, the reference signal is selected. On the other hand, in order to achieve the above object, an active vibration control device according to an invention of claim 11 is a control vibration source capable of generating a control vibration that interferes with a vibration generated from a vibration source, and the generation of the vibration. A plurality of vibration generation state detection means for detecting a state and outputting as a vibration generation state signal, and a reference signal for selecting a part of the plurality of vibration generation state signals output from the plurality of vibration generation state detection means as a reference signal Selection means and active control means for controlling the control vibration source so as to reduce the vibration after the interference based on the reference signal, wherein the reference signal selection means is a main factor of the vibration. The vibration generation state signal is selected as the reference signal.

Similarly, in order to achieve the above object, an active vibration control device according to the invention of claim 12 is a control vibration source capable of generating a control vibration that interferes with a vibration emitted from a vibration source, Residual vibration detection means for detecting the residual vibration after the interference and outputting it as a residual vibration signal, a plurality of vibration generation state detection means for detecting the generation state of the vibration and outputting it as a vibration generation state signal, and a plurality of these vibration generation Reference signal selection means for selecting a part of the plurality of vibration generation state signals output from the state detection means as a reference signal, and vibration after the interference is reduced based on the reference signal and the residual vibration signal. Active control means for controlling the controlled vibration source, wherein the reference signal selection means selects the vibration generation state signal, which is a main factor of the vibration, as the reference signal. It was.

According to a thirteenth aspect of the present invention, in the active vibration control device according to the twelfth aspect of the invention,
The active control means filters the reference signal to generate a drive signal for driving the controlled vibration source, and the post-interference vibration is reduced based on the reference signal and the residual vibration signal. And adaptive processing means for updating the filter coefficient of the adaptive digital filter according to the adaptive algorithm.

According to a fourteenth aspect of the present invention, in the active vibration control device according to the above eleventh to thirteenth aspects, the reference signal selecting means sets a virtual vibration source based on the vibration generation state signal. A virtual vibration source calculation unit for calculating and a correlation calculation unit for calculating a correlation between the virtual vibration source and the plurality of vibration generation state signals are provided, and the main factor is based on a calculation result of the correlation calculation unit. The vibration generation state signal is selected as the reference signal.

On the other hand, the invention according to claim 15 is the active vibration control device according to claim 12 or 13, wherein the reference signal selecting means selects a virtual vibration source based on the vibration generation state signal. Virtual vibration source calculation means for calculating, virtual vibration source identification means for identifying a main virtual vibration source based on the correlation between the virtual vibration source and the residual vibration signal,
Correlation calculation means for calculating the correlation between the virtual vibration source and the plurality of vibration generation state signals, and the vibration generation state signal, which is the main factor, based on the calculation result of the correlation calculation means, the reference signal. I chose to choose it.

According to a sixteenth aspect of the present invention, in the active type vibration control device according to the fifteenth aspect of the invention,
The reference signal selection means sets the number of the selected reference signals to be the same as the number of the specified virtual vibration sources. The invention according to claim 17 is the active vibration control device according to claim 15 or 16, wherein the virtual vibration source identification means is a correlation value between the virtual noisy vibration source and the residual vibration signal. A virtual vibration source of a combination in which the sum of is equal to or larger than a predetermined value is specified as the main virtual vibration source.

Furthermore, the invention according to claim 18 is the active vibration control device according to any one of claims 14 to 17, wherein the reference signal selecting means has a large correlation value calculated by the correlation calculating means. The vibration generation state signal is sequentially selected as the reference signal. Also,
The invention according to claim 19 is the active vibration control device according to any one of claims 14 to 18, wherein the reference signal selecting means is a transfer function matrix between the virtual vibration source and the vibration generation state signal. The determinant calculation means for calculating the determinant is provided, and the vibration generation state signal which is the main factor is selected as the reference signal in consideration of the determinant.

According to a twentieth aspect of the invention, in the active vibration control device according to the twelfth to nineteenth aspects of the invention, the reference signal selecting means gives priority to a frequency band having a high level of the residual vibration signal. Then, the reference signal is selected.

[0029]

In the invention according to claim 1 or 2,
Since the reference signal selection means selects the signal that is the main factor of noise from the plurality of noise generation state signals as the reference signal, the number of reference signals used for control in the active control means becomes the minimum necessary. In this case, unlike the case where the reference signal is selected based on the coherence between the noise generation state signal and the noise, etc., the total sum of the information contained in the selected reference signal is a part of the true noise source. A bias toward noise sources can be avoided.

The difference between the invention according to claim 1 and the invention according to claim 2 is that the former active control means is configured to execute feedforward control based on at least a reference signal. The latter active control means is configured to execute the feedforward based on the reference signal and the feedback control based on the residual noise signal.

Further, in the invention according to claim 3, the active control means in the invention according to claim 2 has an adaptive digital filter and an adaptive processing means, so that adaptive control is executed here. That is, the reference signal selected by the reference signal selecting means is filtered by the adaptive digital filter to generate a drive signal, and the control sound source is driven by the drive signal, while the adaptive processing means controls the reference signal and the residual noise signal. Based on and, the filter coefficient of the adaptive digital filter is updated according to the adaptive algorithm.After the filter coefficient of the adaptive digital filter has converged to some extent, the control sound emitted from the control sound source cancels noise and is transmitted. The level of noise generated is reduced.

The inventions of claims 4 to 10 are more specific than the inventions of claims 1 to 3, but in order to facilitate understanding of the contents, theoretical The operation of each invention will be explained by weaving various backgrounds. First, the noise generation state signals output by the N noise generation state detection means are sampled M times to create a vector (N rows and 1 column vector) composed of N data blocks including M data. And Then, it is assumed that the spectrum matrix [X] for each frequency is created by performing FFT (Fast Fourier Transform) on the vector for each data block. Then, the cross power spectrum of each component of the spectrum matrix [X] is calculated to calculate the cross power matrix [S _XX ] for each frequency. The calculation formula of the cross power matrix [S _XX ] is as shown in the following formula (1).

[S _XX ] = [X] [X] ^h (1) Here, [] ^h means transposition of the conjugate complex matrix. Here, if N = 12 as in the above-described example, the specific content of a certain frequency component of the cross power matrix [S _XX ] is as in the following expression (2).

[0034]

[Equation 1]

[0035] Note that C _nm means the cross power spectrum of the nth channel and the mth channel, and C _nn is the cross power spectrum of the nth channel and the nth channel (that is, the auto power spectrum of the nth channel).
Means On the other hand, the spectrum matrix [X] is the transfer function [H] between the true noise source and the noise generation state detecting means.
And the force [F] generated by the true noise source can be expressed as in the following equation (3).

[X] = [H] [F] (3) Therefore, [S _XX ] = [X] [X] ^h = [H] [F] [F] ^h [H] ^h = [H ] [S _FF ] [H] ^h (4) That is, the cross power matrix [S _FF ] of the force [F] is transformed (similar transformation) according to the above equation (4) using the regular matrix [H]. For example, the cross power matrix [S
_XX ] is required.

Since the regular matrix [H] generally has a full rank, the rank of the cross power matrix [S _XX ] becomes equal to the rank of the cross power matrix [S _FF ]. Therefore, the number of true noise sources that are independent of each other can be obtained by examining the rank of the cross power matrix [S _XX ]. That is, the eigenvalue analysis of the cross power matrix [S _XX ] is performed,
The eigenvalue analysis means that a certain matrix is transformed to create a diagonal matrix. Therefore, the eigenvalue of [S _XX ] = [U] [S ′ _XX ] [U] ^h (5) A diagonal matrix [S ' _XX ] (Principal Component: Principal Comonent) of λ may be created. [U] is a matrix composed of eigenvectors {φ}, and its normalization is [U] [U] ^h = [I] (6).

If the diagonal matrix [S ' _XX ] is an example of the above equation (2), it becomes the following equation (7).

[0039]

[Equation 2]

[0040] Looking at this equation (7), it can be said that there is no correlation between the signals because all the components other than the auto power spectrum, that is, the cross power spectrum, are “0”. Since the elements (principal components) λ _nn of this diagonal matrix [S ′ _XX ] are uncorrelated with each other, it can be assumed that they are noise sources independent of each other. _XX ] elements that are not "0" (elements that are sufficiently large compared to other elements because they do not become "0" in the actual calculation) are not true noise sources but are independent of each other and affect noise. It can be considered as a "virtual noise source" as a noise source. However, since it is an eigenvalue analysis, the diagonal matrix [S ' _XX ] does not necessarily have 12 diagonal elements λ even if the original cross power matrix [S _XX ] is 12 rows and 12 columns. . That is, 12 is the maximum number and may be less than that.

What is important here is that the virtual noise sources λ _nn are irrelevant to each other. Therefore, if the correlation between the virtual noise sources and the noise generation state signal is obtained, which noise generation state signal is selected. This makes it possible to determine whether or not the information of all virtual noise sources can be incorporated into the control of the active control means. Specifically, in the above formula (7), only the rows and columns in which the diagonal components exist (that is, the rows and columns in which the diagonal components are "0" are removed) are used as the subscript "NS". When expressed using, the above equation (5) becomes the following equation (8).

[S _XX ] = [U] _NS [S ′ _XX ] _NS [U] _NS ^h (8) On the other hand, the spectrum matrix [X] created from the measured signal in the physical coordinate system and the above (7) Principal component [X 'after coordinate conversion corresponding to eigenvalue analysis of expression
] The relationship with _NS is as shown in the following expression (9). [X] = [U] _NS [X ′] _NS ∴ [X ′] _NS = [U] _NS ^h [X] (9) Then, the virtual vibration mode is the spectrum matrix [X] and the principal component [ X '] _NS is represented by a column of a cross power matrix [S _XX' ]. At this time, from the above formula (9), [S _{XX '} ] = [X] [X'] _NS ^h = [X] [X] ^h [U] _NS = [S _XX ] [U] _NS (10), and the cross power matrix [S _{XX '} ] is the spectrum matrix [X] created from the measured signal and the eigenvector [U ] And that it can be calculated.

Further, the coherence (virtual coherence) which is the correlation between the spectrum matrix [X] and the principal component [X '] is
It is expressed by the formula 1).

[0044]

(Equation 3)

[0045] Here, S _ii is (i, i) of the cross power matrix [S _XX ].
The term S ′ _jj is the (j, j) term of the diagonal matrix [S ′ _XX ], S
_{ij '} is the (i, j) term of the cross power matrix [S _XX' ]. Then, the virtual
The coherence reveals the correlation between the virtual noise source and the noise generation status signal.

That is, in the invention according to claim 4,
The virtual noise source calculation means calculates the virtual noise source, the correlation calculation means calculates the correlation between the virtual noise source and the noise generation state signal, and based on the correlation, the noise generation state which is the main factor of noise If a signal is selected as a reference signal, the selected reference signals are a set of reference signals that cover the information of the virtual noise source. Therefore, the number of reference signals is set in order to perform good noise reduction control. It will be the minimum required.

In this case, as in the invention according to claim 8,
If the noise generation state signal is selected as the reference signal in the descending order of the coherence (correlation value) obtained by the above equation (11), the noise generation state signal which is the main factor of noise can be easily and accurately selected as the reference signal. . Next, theoretically all the components other than the eigenvalue in the above equation (7) are "0", but in the actual calculation, all the components other than the eigenvalue are "0".
There is no guarantee that Therefore, the correlation between the virtual noise source, which is the diagonal component of the equation (7) above, and the actually observed noise (residual noise) is calculated, and based on these correlations,
From the calculated virtual noise sources, the main virtual noise source that affects the actual noise is identified. Specifically, since the virtual noise sources are independent of each other, the coherence between the virtual noise sources and the residual noise can be analyzed by multiple coherence by simply adding them. The coherence between the virtual noise source and the residual noise can also be calculated by the same formula as the above formula (11), and each matrix required for the calculation of the formula (11) is also calculated by the virtual noise source and the noise. Similar to the case of obtaining the coherence with the generation state signal, it is obtained by the same procedure as the above equations (1) to (10).

That is, according to the fifth aspect of the present invention, even if the true eigenvalue cannot be specified by the actual calculation, the true eigenvalue is calculated based on the calculated correlation between the virtual noise source and the residual noise signal. The main virtual noise sources consisting of are identified. Then, the coherence between the calculated virtual noise source and the residual noise signal is added, for example, in descending order of the value of the virtual noise source, and the virtual noise source with which the sum is sufficiently large (saturated) is found and The main virtual noise source.
That is, according to the invention of claim 7, the main virtual noise source that affects the residual noise is specified by obtaining the sum of the correlation values between the virtual noise source and the residual noise.

On the other hand, the virtual noise source is [V], the transfer function between the virtual noise source [V] and the actually observed noise (residual noise) [D] is [A], the virtual noise source [V] If the transfer function between the reference signal [X] and the reference signal [X] is [B], and the transfer function of the feedforward control system based on the reference signal [X] is [W], then [D] = [ A] [V] ... (12) [X] = [B] [V] ... (13). Then, in order to cancel the noise by the feedforward control system, the result obtained by processing the reference signal [X] with the transfer function [W] should have the same magnitude as the noise and the opposite phase. Therefore, [W] [ X] = − [D] (14) It suffices to create a transfer function [W] satisfying (14).

That is, from the above equations (12) and (14), [A] [V] =-[W] [X] = [W] [B] [V] (15) and both sides are obtained. If the virtual noise source [V] in is deleted and rearranged, the characteristic required for the transfer function [W] becomes [W] = − [A] [B] ⁻¹ (16). That is, the existence of the inverse matrix of the transfer function [B] is a condition for the existence of the transfer function [W] for canceling noise.

Therefore, when the number of reference signals to be selected is the same as the number of main virtual noise sources specified by the virtual noise source specifying means as in the invention according to claim 6, the transfer function [B ] Is a square matrix, the inverse matrix of the transfer function [B] exists except for a special case, and the existence of the transfer function [W] for canceling noise is guaranteed.
In particular, when the adaptive processing according to the adaptive algorithm is executed as in the invention according to claim 3, the convergence of the adaptive digital filter to the optimum value is substantially guaranteed.

The special case where the transfer function [B] is a square matrix but its inverse does not exist is when the determinant is "0". Therefore, in the invention according to claim 9, the determinant of the transfer function matrix [B] between the virtual noise source and the noise generation state signal is calculated by the determinant calculation means, and the calculation result is considered ( For example, when the value of the arithmetic expression is “0” or extremely small, the noise generation state signal of the combination at that time is not directly selected as the reference signal,
Since the reference signal is selected, it is ensured that the inverse matrix of the transfer function [B] exists, and the existence of the transfer function [W] for canceling noise is guaranteed. In particular, when the adaptive processing according to the adaptive algorithm is executed as in the invention according to claim 3, the convergence of the adaptive digital filter to the optimum value is guaranteed.

Each element B of the transfer function matrix [B]
_ij can be calculated by the following equation (17). B _ij = S _ij / S _ii (17) In addition, S _ij is the cross power spectrum of the i-th virtual noise source and the j-th noise generation state signal, and is calculated | required by said Formula (10), S _ii is the auto power spectrum of the i-th virtual noise source, and is obtained from [S _vv ] = [V] _NS [V] _NS ^h (18), but in reality, this S _ii is crossed. It is the principal component (eigenvalue) matrix of the power matrix [S _XX ] itself, and is the diagonal matrix [S ′ _XX ] in the above equation (5).

Further, the selected reference signal may differ depending on the frequency band. In such a case, the frequency band having a high level of the residual noise signal is prioritized as in the invention according to claim 10. When the reference signal is selected in this way, the reference signal with which the noise reduction effect is large is selected.
When determining the level of the residual noise signal, the measured residual noise signal is compared with the human hearing characteristic (for example, A characteristic).
By multiplying by, the frequency band in which the level of the residual noise signal in a state where a human can hear is high is determined, so that the reference signal with which the noise reduction effect is greater for humans is selected.

The inventions according to claims 1 to 10 are all directed to noise, whereas the inventions according to claims 11 to 20 are intended for vibration. Therefore, the operations of the inventions according to claims 11 to 20 are substantially the same as the inventions according to claims 1 to 10 above, although there is a difference between sound and vibration.

[0056]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a diagram showing the overall configuration of a first embodiment of the present invention. In this embodiment, an active noise control system according to the present invention is used to reduce noise in a vehicle interior. This is applied to the device 1.

First, the structure will be described. This vehicle active noise control system 1 includes a road surface as a noise source and each wheel 2
a to 2d (only the wheels 2a and 2c on the left side of the vehicle are shown in FIG. 1) is a device for reducing road noise as noise transmitted into the vehicle interior 6. The road noise is noise generated due to the vertical movement of the wheels when the wheels pass through the unevenness on the road surface, and is usually random noise including many frequency components.

Then, the suspensions 10a to 10d (only the suspensions 10a and 10c are shown in FIG. 1) interposed between the vehicle body 9 and the wheels 2a to 2d are in proper places. The acceleration sensor units 5a to 5d (only the acceleration sensor units 5a and 5c are shown in FIG. 1) having a plurality of built-in acceleration sensors for detecting the vertical vibration input are attached.

Each of the acceleration sensor units 5a to 5d has three built-in acceleration sensors (not shown), and accelerates in three directions (x direction, y direction and z direction) of the vehicle body in the front, rear, left and right and up and down directions. The acceleration detection signal v _t (t = 1 to T: T is the number of acceleration sensors, and in this example, 4 (rings) × 3 (directions) = 12, T = 12) is supplied to the controller 20. These acceleration detection signals v _t
Corresponds to the noise generation state detection signal.

On the other hand, in the passenger compartment 6, a plurality of loudspeakers 7 as control sound sources for generating control sounds in the passenger compartment 6 are provided.
a to 7d (only the loudspeakers 7a and 7c are shown in FIG. 1) are arranged, for example, at the foot position of the front seat and at the rear of the headrest of the rear seat, and these loudspeakers 7a to 7d are drive signal supplied from the controller _{20 y m (m = 1~M:} M is a loudspeaker 7a
˜7d, and M = 4 in this embodiment. ) To drive a control sound.

Further, a plurality of microphones 8a to 8d are arranged on the ceiling of each seat in the passenger compartment 6 so as to be as close as possible to the occupant's ear position.
8d measure the sound pressure of the noise remaining at the installation position, and the measured value is the residual noise signal e ₁ (l = _{1 to} L: L is the number of microphones 8a to 8d, and in the present embodiment, L =
It is 4. ) Is supplied to the controller 20.

Here, the controller 20 is composed of a necessary interface circuit such as an A / D converter and a D / A converter, a microcomputer, etc., and the supplied acceleration detection signal v _t and residual noise signal e _l. The loudspeakers 7a to 7d perform control processing based on the above so that the loudspeakers 7a to 7d emit control sounds such that the road noise transmitted to the vehicle interior 6 is canceled.
and supplies the drive signal y _m and (y ₁ ~y ₄₎ to A～7d.

The controller 20 supplies each acceleration supplied from the twelve acceleration sensors at a stage before the actual noise reduction control is executed (for example, immediately after the engine is started and then started, or at the time of regular inspection). From the detection signals v _t , a signal that is the main factor of the road noise transmitted to the vehicle interior 6 is selected, and thereafter, the selected acceleration detection signal v _t is the reference signal x _k (k = 1 to 1). K: K is the reference signal x
, And K <T. ) Is used for noise reduction control.

In the noise reduction control in the controller 20, basically, each reference signal x _k is filtered by the adaptive digital filter W _km with a variable filter coefficient (actually, convolution operation), and the result is added as a subscript. A drive signal y _m is generated by adding it every _m , and the drive signal y _m is supplied to each of the loudspeakers 7a to 7d, while the filter coefficient W _kmi of the adaptive digital filter W _km (i =
0, 1, 2, ..., I-1: I is the number of taps of the adaptive digital filter W _km ), and processing for sequentially updating according to the Filtered-X LMS algorithm is executed.

The arithmetic processing of the drive signal y _m is as shown in the following equation (19), and the adaptive digital filter W
The update calculation process of the filter coefficient W _kmi of _km is
Since the red-X LMS algorithm is followed, the equations (20) and (21) are obtained. However, α is a coefficient called a convergence coefficient, and is involved in the speed and stability of convergence, and the subscripts (n), (n-1),
(N-i) and (n-j) represent values at sampling times n, n + 1, n-i, and n-j, respectively.

C ^ _lmj is a transfer function filter between the l-th loudspeaker and the m-th microphone. J of the transfer function filter C ^ _lm (j = 0, 1, 2, ..., J-1: J is transmitted). The tap number of the function filter) -th filter coefficient. From the above, as shown in FIG. 2 which is a block diagram showing the functional configuration, the controller 20 of the present embodiment selects the reference signal x _k from the acceleration detection signal v _t and outputs it. 21 and each filter coefficient W of the reference signal x _k and the adaptive digital filter W _km according to the above equation (19).
_Based on the drive signal generation unit 22 that _{convolves kmi} and adds for each subscript m to generate and output the drive signal y _m , and the reference signal x _k and the transfer function filter C _lm , for updating according to the above equation (22). and updating the reference signal calculation unit 23 for calculating a reference signal r _klm, updating the reference signal r _klm and the residual noise signal e _l
And a filter coefficient updating unit 24 that updates the filter coefficient W _kmi based on the above equation (20).

FIGS. 3 and 4 are flowcharts showing the outline of the processing executed in the controller 20.
The operation of this embodiment will be described with reference to FIGS. That is,
FIG. 3 shows the content of the processing for determining which of the acceleration detection signals v _t supplied by the reference signal selection unit 21 is to be the reference signal x _k, and noise reduction as described above. It is executed before the control is executed. Further, FIG. 4 shows the content of the noise reduction control based on the reference signal x _k and the residual noise signal e _l .

When the processing shown in FIG. 3 is executed, first in step 101, the acceleration detection signal v _t and the residual noise signal e _l supplied from each acceleration sensor are subjected to FFT.
Read enough numbers to process. The residual noise signal e _l read here is transmitted from the true noise source into the passenger compartment 6 because no control sound is generated from the loudspeakers 7a to 7d in the passenger compartment 6. The noise itself. When the process of step 101 is completed, the process proceeds to step 102, and the data block including the acceleration detection signal v _t and the data block including the residual noise signal e _l are FFT-processed to obtain the spectrum matrix [X].
And [E].

Next, in step 103, the cross power matrix [S _XX ] of the spectrum matrix [X] is calculated according to the above equation (1). And step 104
And shift to the diagonal matrix [S 'based on the above equation (5).
_XX ] is calculated. Further, in this step 104, the principal component [X ′] _NS required in the following calculation is also calculated according to the above equation (9).

Next, in step 105, the residual noise signal e _l and the calculated virtual noise sources (diagonal matrix [S ') are calculated based on the spectrum matrix [E] and the principal component [X'] _{NS. XX} ] diagonal component) and coherence (virtual coherence) E _COH
Is calculated. Specifically, the coherence E _COH is calculated for each of the combinations of the microphones 7a to 7d that output the residual noise signal e ₁ and each virtual noise source.
The coherence E _COH is calculated according to the same formula as in the above (11), but the cross power matrices [S _EE ] and [S _{EE '} ] required for the calculation are also expressed in the above formulas (9) and (10). It is calculated according to the same formula as.

Then, the process proceeds to step 106 and the step
Coherence E found in page 105_COHOn the basis of,
Virtual noise source affecting the noise level in the passenger compartment 6
Determine (main virtual noise source). This is step 10
The diagonal matrix [S ' _XX] Is the diagonal component of
It consists of a component that is a value and a component that is not an eigenvalue.
The latter will be "0".
Because it is. Therefore, is it a component that is not "0"?
It is necessary to select components that are not eigenvalues from
Whether to identify the component as a true eigenvalue (main virtual noise source)
The standard of can not be decided unconditionally.

Therefore, in this embodiment, the coherence E
_COH is added for each residual noise signal e _{l in} descending order of virtual noise source (eigenvalue), and the virtual noise source corresponding to the added coherence E _COH when the added value becomes a predetermined value or more,
The main virtual noise source. The reason is that it is possible to predict that a diagonal component having a larger value than others is an eigenvalue. In addition, the principal required as a virtual noise source
Since the components are independent from each other, even if the coherence E _COH of the virtual noise source and the residual noise signal e _l is added for each residual noise signal e _l , the coherences do not overlap, so that the added coherence This is because if the value is sufficiently large, the virtual noise source that affects the road noise in the vehicle interior 6 can be identified and specified as the main virtual noise source. That is, the main virtual noise source can be determined from the sum of its coherence E _COH .

Next, in step 107, the coherence (virtual coherence) V _COH between each acceleration detection signal v _t and the main virtual noise source (true eigenvalue) specified in step 106 is calculated. Calculate Specifically, the coherence V _COH is calculated for each combination of the acceleration sensor that outputs the acceleration detection signal v _t and each main virtual noise source. The coherence V _COH is calculated according to the above (11).

Then, in step 108, the reference signal x is selected on the basis of the coherence V _COH . Specifically, the acceleration detection signal v _t is selected as the reference signal x _{k in the} descending order of the coherence V _COH .
However, the number of reference signals x _k selected here is the same as the number of main virtual noise sources specified in step 106.

The selection of the reference signal x _k means that the acceleration sensor that outputs the acceleration detection signal v _t selected as the reference signal x _k is selected, and the noise after that is selected. In the reduction control, the acceleration detection signal v supplied from the selected acceleration sensor
_This means that _t is used as the reference signal x _k . When the process of step 108 is finished, the process of FIG. 3 is finished, and thereafter, the noise reduction control of FIG. 4 is executed.

That is, the processing shown in FIG. 4 is executed at each sampling timing in synchronization with the sampling clock. First, the acceleration detection supplied from the acceleration sensor selected in step 201 is detected. The signal v _t is read as the reference signal x _{k and} stored as the reference signal x (n) at the sampling time n. Next, in step 202, the reference signal x
And the adaptive digital filter W _km and above (1
The drive signal y _m is calculated according to the equation (9).

Then, the routine proceeds to step 203, where the drive signal y _m is output to each of the loudspeakers 7a to 7d. Then, the process proceeds to step 204, where each microphone 8a-
The residual noise signal e _l supplied from 8d is read, and the residual noise signal e _l (n) at the sampling time n is read.
Memorize as.

Then, the process proceeds to step 205 and the above (2
The update reference signal r _klm is calculated according to the equation (1), and the update reference signal r _klm (n) at the sampling time n is calculated.
Memorize as. Then, the _{process proceeds} to step 206, and the filter coefficient W _kmi of the adaptive digital filter W _km is updated according to the above equation (1). When the process of step 206 is completed, this process is completed and the next sampling /
After waiting until the timing comes, the above step 20
It returns to 1 and performs the above-mentioned processing again.

[0079] When the process shown in FIG. 4 in the controller 20 is executed, since the drive signal y _m is supplied successively to each loudspeaker 7a to 7d, the passenger compartment 6 to the drive signal y _m Depending on the control sound will be generated,
Immediately after the control is started, the filter coefficient W _kmi of the adaptive digital filter W _km does not always converge to the optimum value.
It cannot be said that the road noise is reduced by the control sounds emitted from the loudspeakers 7a to 7d.

However, when the processing shown in FIG. 4 is repeatedly executed, the filter coefficient updating unit 24 follows each filter coefficient W of the adaptive digital filter W _km according to the LMS algorithm.
_{Since kmi} is updated, the filter coefficient W _kmi converges toward the optimum value, and the road noise transmitted into the passenger compartment 6 is canceled by the control sound.
The noise level in the passenger compartment 6 is reduced.

Moreover, in the present embodiment, the reference signal x _k is selected from the plurality of acceleration detection signals v _t by the process of FIG. 3, so that a sufficient noise reduction effect is obtained while the reference signal x _k is obtained. The number of x _k can be minimized, whereby the calculation load on the controller 20 can be minimized, and a cheaper microcomputer or the like can be used, and the cost can be reduced.

The advantage of executing the processing of FIG. 3 will be described in more detail. That is, as shown in FIG. 5, two true noise sources N ₁ and N ₂ , one evaluation point E, and acceleration detection signals v _{1 to} v output from four reference signal measurement points.
_4, consider a filter W to output them acceleration detection signal v ₁ to v ₄ toward the filter to evaluation point E, the system comprising. It should be noted that arrows in FIG. 5 represent propagation paths of noise and signals, respectively, and vibrations and noises are not transmitted to places without arrows (for example, between the noise source N ₁ and the acceleration detection signal v ₃ ). It means that. The numbers above the arrows indicate the gains of the respective transmission systems, and the numbers below the circles of the noise sources N ₁ and N ₂ represent their noise (vibration) levels.

Simulation in such a system
Evaluation point E and each acceleration detection signal v₁~ V_FourBetween
As for the coherence of, evaluation point E and acceleration detection
Signal v₁, V₂The coherence between and is CH in Figure 6.₁,
CH₂It becomes comparatively small like
Outgoing signal v₃, V_FourThe coherence between and is C in Figure 6.
H ₃, CH_FourIt becomes relatively large like. This is acceleration
Degree detection signal v₃, V_FourSource of noise N₂Level of
Is the acceleration detection signal v₁, V₂Source of noise N₁of
Since it is twice the level, at the evaluation point E,
Noise source N₂Whether the component transmitted from is dominant
It is. However, because of the large coherence
Acceleration detection signal v₃, V_FourAs the reference signal
If the sound reduction control is executed, the noise source N₁Information about
Information is not used for feedforward control.
Therefore, a sufficient noise reduction effect cannot be expected.

On the other hand, when the processing as shown in FIG. 3 is executed, the eigenvalue analysis is first performed by the processing of steps 101 to 104 to calculate the virtual noise source. In this case, as shown in FIG. 7, four virtual noise sources (virtual noise source candidates) are calculated, but by the process of step 105, between the calculated virtual noise source and the evaluation point E. Four coherences are required. And step 1
By the processing of 06, the sum of these four coherences is sequentially calculated. In this case, the sum of the two large coherences E _COH1 and E _COH2 is approximately 1, as shown in FIG. From this, it can be seen that in the system of FIG. 5, the number of main virtual noise sources is two. That is, as shown in FIG. 7, the diagonal components PC ₁ and PC _{2 which} are sufficiently larger than the others are specified as the main virtual noise sources.

Then, the coherence V _COH between the diagonal components PC ₁ and PC ₂ which are the main virtual noise sources and the respective vibration detection signals v _{1 to} v ₄ is calculated in step 107,
In step 108, the same number of reference signals x _{k as} the number of main virtual noise sources (two in the case of the system of FIG. 5) are selected based on the magnitude of the coherence V _COH . The number of the reference signals x _k selected here is set to be the same as the number of main virtual noise sources, as described with reference to the above equations (12) to (16). This is because the transfer function [B] between the signals x _k is a square matrix, which increases the possibility that the inverse matrix [B] ⁻¹ exists.

FIG. 12 shows a system different from that of FIG. 5, but the meanings of the arrows, symbols and numerals are the same as in the case of FIG. In the system of FIG. 12, the acceleration detection signals v ₂ and v ₃ include both components of the noise sources N ₁ and N ₂ .
A simulation was performed on this system, and the evaluation point E and
When the coherence between the acceleration detection signals v _{1 to} v ₄ is calculated, the coherence between the evaluation point E and the acceleration detection signal v ₁ becomes relatively large as CH ₁ in FIG. The coherence between E and the acceleration detection signals v ₂ and v ₃ becomes medium like CH ₂ and CH _{3 in} FIG. 13, and the coherence between the evaluation point E and the acceleration detection signal v ₄ is shown in FIG. It becomes relatively small like CH ₄ . Therefore, when the reference signal x _k is determined on the basis of coherence,
The acceleration detection signal v ₄ is more likely to be discarded, but the acceleration detection signal v ₄ is the acceleration detection signal v 4.
_Since the information of the noise source N ₂ is purely represented as compared with ₂ and v ₃ , even if the acceleration detection signal v ₄ is used, efficient noise reduction control is executed, Efficient control cannot be expected.

On the other hand, when the processing of FIG. 3 is executed,
First, the main virtual noise source is specified by the processing of steps 101 to 106. Specifically, if the virtual noise source calculated in step 104 is the system shown in FIG.
As shown in FIG. 14, when there are four, the coherence between each virtual noise source and the evaluation point E is calculated in step 105, and the sum of these coherences is sequentially calculated in step 106, which is shown in FIG. Like two coherence E
It can be seen that most of the information is occupied by _COH1 and E _COH2 . Therefore, as shown in FIG. 7, two diagonal components PC ₁ and PC _{2 that} are sufficiently larger than the others are identified as the main virtual noise sources. After that, it is similar to the case of the system of FIG.

As described above, according to the present embodiment, by executing the process of FIG. 3 before executing the noise reduction control as shown in FIG. Find the acceleration detection signal v _t , which is the main factor of noise,
The found acceleration detection signal v _t is used as a reference signal x _k.
The reference signal x _k used for noise reduction control is irrespective of the structure of the applied system.
It is possible to minimize the number of Then, as the number of reference signals x _k decreases, the calculation load on the controller 20 decreases, and an inexpensive calculation processor having a relatively low capacity can be used.

Moreover, the main virtual noise source that affects the road noise in the passenger compartment 6 is specified, and the number K of the reference signals x _k selected later is set to be the same as the number of the main virtual noise sources. , Transfer function [B] except in special cases described later
Since there is an inverse matrix of, there is an advantage that it is guaranteed that there is an optimum value of the filter coefficient W _kmi of the adaptive digital filter W _km for canceling the road noise, which results in good noise reduction. You can expect an effect.

Here, in this embodiment, the loudspeaker 7 is used.
a to 7d correspond to control sound sources, and microphones 8a to 8
d corresponds to the residual noise detecting means, each acceleration sensor in the acceleration sensor units 5a to 5d corresponds to the noise generating state detecting means, the reference signal selecting section 21 and the processing of FIG. 3 correspond to the reference signal selecting means, The drive signal generation unit 22, the update reference signal generation unit 23, the filter coefficient update unit 24 and the processing shown in FIG. 4 correspond to the active control means, and the update reference signal generation unit 23, the filter coefficient update unit 24 and steps 204 to. Two
The processing of 06 corresponds to the adaptive processing means, the processing of step 103 corresponds to the virtual noise source calculation means, the processing of step 107 corresponds to the correlation calculation means, and steps 105 to 106.
Processing corresponds to the virtual noise source identification means.

FIG. 19 is a diagram showing the second embodiment of the present invention, and is a flow chart showing the outline of the processing for selecting the reference signal executed in the controller as in FIG. 3 of the first embodiment. Is. Since the overall configuration and the content of the noise reduction processing are the same as those in the first embodiment, their illustration and description will be omitted. Moreover, the same reference numerals are given to the steps that execute the same processing as in the case of FIG. 3, and the duplicated description will be omitted.

That is, in the present embodiment, the transfer function [B] is used even though the number K of the reference signals x _k described at the end of the first embodiment is the same as the number of main virtual noise sources. We are trying to prevent it from being a special case where the inverse matrix of does not exist. Specifically, when the coherence V _COH between the main virtual noise source and the acceleration detection signal v _t is obtained in step 107, the process _proceeds to step 301, where the candidate of the reference signal x _k is its coherence V _COH. Select in descending order of. However, the reference signal x _k selected here
The number of candidates of is set to a value that is one more than the number of main virtual noise sources specified in step 106. For example, when the number of main virtual noise sources is “6”, step 30
The candidate of the reference signal x _k selected in 1 is the coherence V
Only seven are selected in descending order of _COH .

Next, in step 302, the same number of signals as the number of main virtual noise sources are selected from the acceleration detection signals v _t that are candidates for the reference signal x _k selected in step 301, and In step 303, the transfer function matrix [B] between the virtual noise source and the selected acceleration detection signal v _t is created. Each element of the transfer function matrix [B] is obtained according to the above equation (17).

Then, the process proceeds to step 304, the determinant d of the transfer function matrix [B] is calculated, and then step 30
5, the process determines whether the determinant d is "0" or a sufficiently small value. If this determination is “NO”, the process returns to step 302 and a new acceleration detection signal v _t
Is selected and the processes of steps 303 to 305 are executed. After that, the determination in step 305 is “YES”.
Until the following, the processes of steps 302 to 305 are repeatedly executed while selecting the combination of the acceleration detection signals v _t .

Then, the determination in step 305 is "YE
When a S ", the process proceeds to step 306, the acceleration detection signal v _t that is elected at that time, is selected as a reference signal x _k, which is terminated the process of this FIG. With such a configuration, it is possible to avoid selecting a combination of the acceleration detection signals v _t that has a transfer function matrix [B] with a sufficiently small determinant d as the reference signal x _k. ], It is ensured that there is an optimum value of the filter coefficient W _kmi of the adaptive digital filter W _km for canceling the road noise, which is even better than in the first embodiment. The noise reduction effect can be expected.

For example, in the system shown in FIG. 5, since the number of main virtual noise sources is two, the acceleration detection signal v ₁
2 to ₄ are selected as the reference signal x _k , and the determinant d is as shown in FIGS. 9 to 11, respectively. The subscript of d in FIGS. 9 to 11 corresponds to the combination of the acceleration detection signals v _t selected in the process of step 302. For example, d ₁₂ is a determinant when the acceleration detection signals v ₁ and v ₂ are selected.

According to this, the determinants d ₁₂ and d ₃₄ are "0".
Therefore, the acceleration detection signals v ₁ and v ₂ are simultaneously selected as the reference signal x _k , and the acceleration detection signals v ₃ and
It is avoided that v ₄ is simultaneously selected as the reference signal x _k . If selected as the reference signal x _{k which} is a combination of these, in the former case, the information of the noise source N ₂ is not taken into the feedforward control system,
In the latter case, the information of the noise source N ₁ is not taken into the feedforward control system, and in any case, the optimum filter coefficient W _i that can cancel the noise
Is uncertain.

Similarly, even in the system of FIG. 12, since the number of main virtual noise sources is two, the acceleration detection signals v ₁ to
Two signals out of v ₄ are selected as the reference signal x _k , and the determinant d becomes as shown in FIGS. 16 to 18, respectively. The meaning of the subscript of d is the same as above. According to this, since the determinant d ₂₃ is “0”, it is avoided that the acceleration detection signals v ₂ and v ₃ are simultaneously selected as the reference signal x _k . This combination is the reference signal x
_{If k} is selected, the ratio of the information of the noise sources N ₁ and N ₂ included in the acceleration detection signals v ₂ and v ₃ is 50%.
Therefore, although the level is different, substantially only one acceleration detection signal is taken into the feedforward control system, and the lowest filter coefficient W _i is still undetermined.

The operation and effect of the second embodiment will be described in more detail according to the result of the experiment conducted on the configuration of FIG. First, when the acceleration detection signals v _t are supplied from a total of 12 acceleration sensors in three directions of four wheels, steps 101 to 1 are performed.
The virtual noise source is calculated by the processing of 04. The calculation result is shown in FIG. According to this, the diagonal matrix [S '
_Since it cannot be confirmed that the diagonal component of [ _XX ] is exactly “0”, theoretically, there are 12 eigenvalues and 12 virtual noise sources that are uncorrelated with each other. Become. However, not all the calculated 12 virtual noise sources affect the road noise in the vehicle interior 6.

Therefore, in step 105, the coherences E _{COH1 to} E _COH12 of the calculated 12 virtual noise sources and the residual noise signal e _l are calculated for each residual noise signal e _l , and these are calculated in step 106. The coherences E _{COH1 to} E _COH12 are added in order from the largest virtual noise source. FIG. 21 shows the result of adding the coherences E _COH for the six largest virtual noise sources, and FIG. 22 shows the coherence E for the seven largest virtual noise sources.
_This is the result of adding _COH . In addition, coherence E _COH
The simple addition of is allowed because the virtual noise sources are uncorrelated with each other, as described above.

From these FIGS. 21 and 22, it is possible to judge that most of the road noise in the passenger compartment 6 can be represented by using 6 or 7 virtual noise sources, but as it is, Accurate judgment is difficult. Therefore, in order to make a more reliable determination, the sum of coherences (overall value) at the control target frequency (here, 50 to 250 Hz) in the noise reduction control is calculated for each addition number of coherence E _COH. , As shown in FIG. According to this, it can be seen that the total value of coherence is saturated at about 6 or 7. Therefore, in step 106, 6 or 7 virtual noise sources (main virtual noise sources) may be specified in descending order. 6 or 7
The number of virtual noise sources specified is preferably as large as possible within the range permitted by the computing capacity of the controller 20, although it is arbitrary.

Next, in step 107, the coherence V _COH between the specified virtual noise source and the acceleration detection signal v _t is calculated, and then the _{routine proceeds} to step 301, in the descending order of the coherence V _COH. , The number of acceleration detection signals v _t that is one more than the number of main virtual noise sources is selected. Here, it is assumed that seven acceleration detection signals v _t are selected. FIG. 24 shows the coherence V
The coherence between six virtual noise sources with large _COH and one acceleration detection signal v ₁ is shown. even here,
Since the virtual noise sources are uncorrelated with each other, the multiple coherence between the plurality of virtual noise sources and one acceleration detection signal is obtained by adding the coherences V _COH .

Then, the processes of steps 302 to 305 are executed, and the acceleration detection signal v _t having a combination in which the determinant d does not become "0" is selected. FIG. 25 shows the calculation of the determinant of the transfer function matrix [B] between the virtual noise sources when the number of the main virtual noise sources is 6, and the six acceleration detection signals v _t with a large coherence V _COH. FIG. 26 is a result, and FIG. 26 shows a transfer function matrix [B when the main virtual noise sources are seven and the seven acceleration detection signals v _t with a large coherence V _COH. ] Is a calculation result of the determinant. Since the determinant of these is not “0” at the controlled frequency, the determination in step 305 is “Y”.
ES ”, and the acceleration detection signal v _t at that time is identified as the reference signal x _k in step 306. In the determinant calculation, the values of the acceleration detection signal and the virtual noise source are each multiplied by 1000 in order to prevent cancellation of digits.

FIGS. 27 and 28 are frequency characteristic diagrams showing the noise reduction effect when the noise reduction control of FIG. 4 is executed using the reference signal x _k selected by the above procedure. Represents the noise level when the noise reduction control is not executed, and the solid line represents the residual noise level when the noise reduction control is executed. 27 shows a case where 6 virtual noise sources are specified, and FIG. 28 shows a case where 7 virtual noise sources are specified. In any case, 5 to 7 dB
The noise reduction effect to some extent is obtained.

Thus, even with the configuration of this embodiment,
The same effect as that of the first embodiment can be obtained.
In particular, in this embodiment, the determinant d of the transfer function matrix [B] is calculated, and the acceleration detection signal v _t in a combination in which the determinant d is “0” is not selected as the reference signal x _k . From this, it is ensured that the inverse matrix [B] ^{−1 of} the transfer function [B] exists, and a better noise reduction effect can be obtained more reliably than in the first embodiment.

Here, in this embodiment, steps 303,
The processing of 304 corresponds to the determinant calculation means. FIG. 29 is a diagram showing a third embodiment of the present invention, which is for selecting a reference signal executed in the controller as in FIG. 3 of the first embodiment and FIG. 19 of the second embodiment. It is a flow chart which shows an outline of processing. Since the overall configuration and the content of the noise reduction processing are the same as those in the first embodiment, their illustration and description will be omitted. Also, the steps that execute the same processing as in the cases of FIGS. 3 and 19 will be assigned the same reference numerals, and overlapping description thereof will be omitted.

That is, in the first and second embodiments, the difference due to the frequency band is not taken into consideration, but the reference signal x _k to be selected may differ depending on the frequency band in some cases. Conceivable. Therefore, in the third embodiment, the level of the residual noise signal e _l (more specifically,
By using the reference signal x _k selected in a high frequency band of the residual noise signal e _l multiplied by the A characteristic which is one of human hearing characteristics) preferentially, the result can be obtained for each frequency band. Even if they are different, good noise reduction control is executed.

That is, in step 105, the coherence E
_{When COH} is calculated, the process _proceeds to step 401, and the characteristic E _A of the spectrum matrix [E] which is the result of frequency analysis of the residual noise signal e _l is calculated. Specifically, the characteristic E _{A is} obtained by multiplying the spectrum matrix [E] by the A characteristic.
Is calculated. Next, the process proceeds to step 402, the characteristic E _A and the coherence E _COH are multiplied and integrated for each frequency, and based on the integrated result, step 106
To identify the main virtual noise source.

Next, in step 107, the coherence V _COH is calculated, and then in step 403, the coherence V _COH and the characteristic E _A are multiplied and integrated for each frequency, and based on the integrated result. In step 301, the number of candidates for the reference signal x _k is specified by one more than the number of main virtual noise sources. The subsequent process is the same as the process in the second embodiment.

With such a configuration, the coherence E
_The main virtual noise source is specified and the reference signal x _k is selected based on the value obtained by multiplying _COH and V _COH by the characteristic E _A obtained by modifying the residual noise signal e _l with the A characteristic for each frequency. To be done. Therefore, the noise in the passenger compartment 6 in the frequency band that the occupant feels the highest can be favorably reduced, and a comfortable passenger compartment space can be realized. Other functions and effects are similar to those of the first and second embodiments.

In each of the above embodiments, the noise to be controlled is the road noise, but the noise that can be reduced is not limited to this. For example, the air conditioning noise generated by the air conditioner, the intake pipe, etc. Intake and exhaust noise generated by the exhaust pipe, wind noise generated at the door mirror position, etc. may be targeted. However, in that case, it is necessary to appropriately detect a signal indicating the noise generation state. For example, in the case of air conditioning noise, a signal synchronized with the rotation of the air compressor may be used as the noise generation state signal. If there is a wind noise, a signal synchronized with the rotation of the engine crankshaft may be used as the noise generation state signal.If there is wind noise, the vibration pickup is fixed to the door mirror and the output is the noise generation state signal. And it is sufficient.

In each of the above embodiments, the object to be controlled is noise, but the object to be reduced is not limited to noise. For example, active control force is generated between the suspensions 10a to 10d and the vehicle body. A control actuator is interposed, and an acceleration sensor (residual vibration detection means) for detecting residual vibration is arranged on the vehicle body side, and the control actuator has the same acceleration detection signal v _t and residual vibration detection as in the above embodiment. If the control is performed based on the output signal (residual vibration signal) of the acceleration sensor as the means, the vehicle active vibration control device can reduce the vibration transmitted from the suspensions 10a to 10d to the vehicle body side.

The object to which the present invention is applied is not limited to a vehicle. For example, an active noise control device for reducing noise in the interior of an aircraft or a building, vibration propagating in a vehicle body, or transmission from a machine tool to a floor. It may be applied to an active vibration control device that reduces the generated vibration. In each of the above embodiments, the main virtual noise source is specified based on the coherence E _COH ,
The reference signal x _k is selected on the basis of the coherence V _COH between the identified virtual noise source and the acceleration detection signal v _t , but if it can be expected to improve to some extent as compared with the conventional one, at step 104. The coherence V _COH between the calculated virtual noise source and each acceleration detection signal v _t is calculated, and the maximum number of reference signals x _k are selected in the descending order of the coherence V _COH , for example, within the range permitted by the calculation capability of the controller 20. You may

The number of reference signals x _k to be selected does not necessarily have to be the same as the number of main virtual noise sources. For example, the maximum number of reference signals x _k within the range allowed by the computing capacity of the controller 20. You may make it select.
Further, in each of the above embodiments, the drive signal y is set to the Filter.
The drive signal y is generated according to the ed-X LMS algorithm, but the present invention is not limited to this, and the drive signal y may be generated according to another adaptive algorithm (for example, a frequency domain LMS algorithm). In some cases, it may be an active noise control device or an active vibration control device that executes only feedforward control (for example, control for generating the drive signal y by processing the reference signal x with a filter having a fixed filter coefficient). .

[0115]

As described above, according to the present invention,
Since the noise generation state signal and the vibration generation state signal, which are the main factors of noise and vibration, are selected as reference signals, the minimum required reference signal is selected and used for noise reduction control and vibration reduction control. It is possible to reduce the load to the minimum, which leads to cost reduction and good noise reduction control and vibration reduction control.

Particularly, in the case of the invention according to claim 8 or 18, since it is guaranteed that the inverse matrix of the transfer function matrix exists except for a special case, good noise reduction control and vibration reduction can be performed with high probability. The effect is that the control is executed. Further, according to the invention of claim 9 or 19, since it is assured that the inverse matrix of the transfer function matrix exists, excellent noise reduction control and vibration reduction control are executed with a higher probability. effective.

The invention according to claim 10 or 20 has the effect that good noise reduction control and vibration reduction control are executed even if the reference signals selected for each frequency band differ.

[Brief description of drawings]

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

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

FIG. 3 is a flowchart showing an outline of a reference signal selection process executed in the controller of the first embodiment.

FIG. 4 is a flowchart showing an outline of noise reduction processing executed in the controller of the first embodiment.

FIG. 5 is a configuration diagram of a system for which a simulation is performed.

6 is a diagram showing coherence between an acceleration detection signal and an evaluation point in the system of FIG.

7 is a diagram showing, for each frequency, a virtual noise source calculated in the system of FIG.

8 is a diagram showing coherence between a virtual noise source and an evaluation point calculated in the system of FIG.

9 is a diagram showing a determinant of a transfer function matrix in the system of FIG.

10 is a diagram showing a determinant of a transfer function matrix in the system of FIG.

11 is a diagram showing a determinant of a transfer function matrix in the system of FIG.

FIG. 12 is a configuration diagram of another system for which simulation is performed.

13 is a diagram showing coherence between an acceleration detection signal and an evaluation point in the system of FIG.

14 is a diagram showing, for each frequency, a virtual noise source calculated in the system of FIG.

15 is a diagram showing coherence between a virtual noise source and an evaluation point calculated in the system of FIG.

16 is a diagram showing a determinant of a transfer function matrix in the system of FIG.

17 is a diagram showing a determinant of a transfer function matrix in the system of FIG.

18 is a diagram showing a determinant of a transfer function matrix in the system of FIG.

FIG. 19 is a flowchart showing an outline of reference signal selection processing executed in the controller of the second embodiment.

FIG. 20 is a diagram showing an example of a calculated virtual noise source.

FIG. 21 is a diagram showing coherence between a calculated virtual noise source and an evaluation point.

FIG. 22 is a diagram showing coherence between a calculated virtual noise source and an evaluation point.

FIG. 23 is a graph showing the relationship between the number of virtual noise sources and the overall coherence value.

FIG. 24 is a diagram showing coherence between six virtual noise sources and one acceleration detection signal.

FIG. 25 is a diagram showing a calculation result of a determinant of a transfer function matrix when there are six main virtual noise sources.

FIG. 26 is a diagram showing a calculation result of a determinant of a transfer function matrix when there are seven main virtual noise sources.

FIG. 27 is a frequency characteristic diagram showing a noise reduction effect according to the second embodiment.

FIG. 28 is a frequency characteristic diagram showing another noise reduction effect according to the second embodiment.

FIG. 29 is a flowchart showing an outline of reference signal selection processing executed in the controller of the third embodiment.

[Explanation of symbols]

 1 Vehicle active noise control device 5a, 5c Acceleration sensor unit 7a, 7c Loudspeaker (control sound source) 8a-8d Microphone (residual noise detection means) 20 Controller 21 Reference signal selection unit (reference signal selection means) 22 Drive signal generation Part 23 Reference signal generating part for updating 24 Filter coefficient updating part

Continuation of the front page (51) Int.Cl. ⁶ Identification code Office reference number FI Technical display location G01H 3/00 A G10K 11/16 H03H 21/00 8842-5J

Claims (20)

[Claims] 1. A control sound source capable of generating a control sound that interferes with noise emitted from a noise source, and a plurality of noise generation state detection means for detecting the generation state of the noise and outputting it as a noise generation state signal. Reference signal selection means for selecting a part of the plurality of noise generation state signals output from the plurality of noise generation state detection means as a reference signal, and the noise after the interference is reduced based on the reference signal. An active control unit for controlling a control sound source, wherein the reference signal selection unit selects the noise generation state signal, which is a main factor of the noise, as the reference signal. 2. A control sound source capable of generating a control sound that interferes with noise emitted from a noise source, residual noise detection means for detecting residual noise after the interference and outputting it as a residual noise signal, and generation of the noise. A plurality of noise generation state detection means for detecting a state and outputting as a noise generation state signal, and a reference signal for selecting a part of the plurality of noise generation state signals output from the plurality of noise generation state detection means as a reference signal The reference signal selecting means and the active controlling means for controlling the control sound source so as to reduce the noise after the interference based on the reference signal and the residual noise signal. An active noise control device characterized in that the noise generation state signal, which is a factor, is selected as the reference signal. 3. The active control means includes an adaptive digital filter for filtering the reference signal to generate a drive signal for driving the control sound source, and the post-interference based on the reference signal and the residual noise signal. 3. The active noise control device according to claim 2, further comprising adaptive processing means for updating the filter coefficient of the adaptive digital filter according to an adaptive algorithm so as to reduce noise. 4. The virtual noise source calculation means for calculating a virtual noise source based on the noise generation state signal, and the correlation between the virtual noise source and the plurality of noise generation state signals. 4. The active component according to claim 1, wherein the noise generation state signal, which is the main factor, is selected as the reference signal based on the calculation result of the correlation calculation unit. Type noise control device. 5. The virtual noise source calculation means for calculating a virtual noise source based on the noise generation state signal, and the reference signal selection means mainly based on a correlation between the virtual noise source and the residual noise signal. A virtual noise source specifying unit for specifying a virtual noise source; and a correlation calculating unit for calculating a correlation between the specified virtual noise source and the plurality of noise generation state signals,
The active noise control device according to claim 2 or 3, wherein the noise generation state signal, which is the main factor, is specified as the reference signal based on a calculation result of the correlation calculation means. 6. The active noise control device according to claim 5, wherein the reference signal selection means sets the number of the selected reference signals to be the same as the number of the specified virtual noise sources. 7. The virtual noise source identification means identifies, as the main virtual noise source, a combination of virtual noise sources in which the sum of correlation values between the virtual noise source and the residual noise signal is a predetermined value or more. The active noise control device according to claim 5 or claim 6. 8. The method according to claim 4, wherein the reference signal selection unit selects the noise generation state signal as the reference signal in descending order of the correlation value calculated by the correlation calculation unit. Active noise control device. 9. The reference signal selection means includes determinant calculation means for calculating a determinant of a transfer function matrix between the virtual noise source and the noise generation state signal, and the determinant is taken into consideration in consideration of the determinant. 9. The active noise control device according to claim 4, wherein the noise generation state signal which is a main factor is selected as the reference signal. 10. The active noise control according to claim 2, wherein the reference signal selection unit selects the reference signal by giving priority to a frequency band having a high level of the residual noise signal. apparatus. 11. A control vibration source capable of generating a control vibration that interferes with a vibration generated from a vibration source, and a plurality of vibration generation state detection means for detecting a generation state of the vibration and outputting it as a vibration generation state signal. Reference signal selection means for selecting a part of the plurality of vibration generation state signals output from the plurality of vibration generation state detection means as a reference signal, and vibration after the interference is reduced based on the reference signal. Active control means for controlling the controlled vibration source, wherein the reference signal selection means selects the vibration generation state signal, which is a main factor of the vibration, as the reference signal. apparatus. 12. A control vibration source capable of generating a control vibration that interferes with a vibration emitted from a vibration source, residual vibration detection means for detecting the residual vibration after the interference and outputting it as a residual vibration signal, and A plurality of vibration generation state detection means for detecting the generation state and outputting as a vibration generation state signal, and a reference for selecting a part of the plurality of vibration generation state signals output from the plurality of vibration generation state detection means as a reference signal Signal control means, and active control means for controlling the controlled vibration source based on the reference signal and the residual vibration signal so as to reduce the vibration after the interference, the reference signal selection means, The active vibration control device is characterized in that the vibration generation state signal, which is the main factor of, is selected as the reference signal. 13. The active control means includes an adaptive digital filter for filtering the reference signal to generate a drive signal for driving the controlled vibration source, and the post-interference based on the reference signal and the residual vibration signal. 13. The active vibration control apparatus according to claim 12, further comprising: adaptive processing means for updating the filter coefficient of the adaptive digital filter according to an adaptive algorithm so that the vibration of the vibration is reduced. 14. The reference signal selection means calculates a virtual vibration source based on the vibration generation state signal, and calculates a correlation between the virtual vibration source and the plurality of vibration generation state signals. 14. The active component according to any one of claims 11 to 13, further comprising: a correlation calculation unit for selecting the vibration generation state signal, which is the main factor, as the reference signal based on a calculation result of the correlation calculation unit. Type vibration control device. 15. The reference signal selection means is based on a virtual vibration source calculation means for calculating a virtual vibration source based on the vibration generation state signal and a correlation between the virtual vibration source and the residual vibration signal. A virtual vibration source specifying unit that specifies a virtual vibration source, and a correlation calculation unit that calculates a correlation between the virtual vibration source and the plurality of vibration generation state signals, the calculation unit of the correlation calculation unit based on a calculation result of the correlation calculation unit. The vibration generation state signal, which is a main factor, is selected as the reference signal.
The active vibration control device according to claim 2 or claim 13. 16. The active vibration control device according to claim 15, wherein the reference signal selection means sets the number of the selected reference signals to be the same as the number of the specified virtual vibration sources. 17. The virtual vibration source identification means identifies, as the main virtual vibration source, a combination of virtual vibration sources in which a sum of correlation values between the virtual vibration source and the residual vibration signal is a predetermined value or more. The active vibration control device according to claim 15 or 16. 18. The reference signal selecting means selects the vibration generation state signal as the reference signal in the descending order of the correlation value calculated by the correlation calculating means.
An active vibration control device according to claim 17. 19. The reference signal selecting means includes determinant calculating means for calculating a determinant of a transfer function matrix between the virtual vibration source and the vibration generation state signal, and the determinant is taken into consideration in consideration of the determinant. 19. The active vibration control device according to claim 14, wherein the vibration generation state signal that is a main factor is selected as the reference signal. 20. The active vibration control according to claim 12, wherein the reference signal selecting unit selects the reference signal by giving priority to a frequency band having a high level of the residual vibration signal. apparatus.