This application is based upon and claims the benefit of priority of the prior Japanese Patent Application No. 2008-108690, filed on Apr. 18, 2008, the entire contents of which are incorporated herein by reference.
The embodiments discussed herein are related to an active noise control apparatus which makes a sonic wave having the same amplitude and opposite phase as those of a noise interfere with the noise, thereby actively controlling the noise.
There is known a technique called an active noise control (ANC) for making a sonic wave (control sound) having the same amplitude and opposite phase as those of a noise interfere with the noise, thereby controlling the noise by the interference effect. In recent years, there is proposed an active noise control apparatus for an air conditioning noise and an indoor noise in a factory or an automobile and the like.
As illustrated in
Many algorithms such as LMS and RLS have been proposed as adaptive algorithm used here heretofore, but since a control sound is required to be produced in real time, Filtered-X LMS (Least Mean Square) algorithm is frequently used in view of a small calculation amount (see B. Widrow and S. Stearns, “Adaptive Signal Processing” (Prentice-Hall, Englewood, Cliffs, N.J., 1985) and “Active Noise Control”, Corona written by Seiji NISHIMURA, Takeshi USAGAWA, and Shirou ISE). The basic principle is for renewing a filter coefficient based on a steepest-descent method so that the residual noise is reduced in consideration of a transfer function from the control sound generating section to the residual noise detecting section. As illustrated in
t
is defined as
x(t)
, the reference signal is vectorized to obtain
x(t)=[x(t), x(t−1), . . . , x(t−Nw+1)]
, to which a transfer function of an error path from the control sound generating section to the residual noise detecting section
ĉ=[ĉ(1), ĉ(2), . . . , ĉ(Nw)]
(wherein,
Nw
is the number of taps of filters of the error path is convoluted to obtain a signal (filter reference signal), the signal is given as illustrated in the equation (1).
r(t)=ĉ*x(t) (1)
(* represents a convolution calculation of vector)
For a renewal equation of filter coefficient, this signal is vectorized to obtain
r(t)=[r(t), r(t−1), . . . , r(t−Nh+1)]
Using this, the renewal equation can be formulated as follows.
h(t+1)=h(t)+μ·e(t)·r(t) (2)
Wherein,
e(t)
represents, at time
t
a residual noise signal,
μ
represents a step size parameter,
h(t)=[h(1, t), h(2, t), . . . , h(Nh, t)]
(wherein,
Nh
represents the number of taps of the adaptive filter), at time
t
represents filter coefficient of the adaptive filter.
In the conventional technique 1 explained with reference to
Even if a control signal which is input to the control sound generating section is an undistorted signal as illustrated with a solid line in
The conventional technique 2 illustrated in
Here, the conventional technique 1 explained with reference to
The conventional technique 2 illustrated in
Here, it is indicated that noise is generated in two frequency bands before the ANC operation, and the noise in one of the frequency bands is cancelled after the ANC operation, but the noise in the other band corresponding to a harmonic can not controlled sufficiently. When the band where noise is not sufficiently controlled is more important subjectively, the problem is serious.
An active noise control apparatus that controls by a control sound a noise which is output from a noise source, includes:
a control sound generating section which inputs a control signal, and produce the control sound;
a residual noise detecting section which detects, as a residual noise signal, a noise remaining after the noise control by the control sound;
a control signal generating section which inputs, as a reference signal, a signal concerning the noise or the generation state of the noise, and generates the control signal; and
a controlling section which inputs the control signal and the residual noise signal, detects the components that cannot be identified in the control signal generating section, and controls the generation of the control signal in the control signal generating section.
It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention, as claimed.
Preferred embodiments of the present invention will be explained with reference to accompanying drawings.
The active noise control apparatus of the first embodiment illustrated in
According to the active noise control apparatus illustrated in
The reference signal detecting section 10 detects a signal (reference signal) concerning the generation state of noise,
x(t)
, and divides the detected reference signal by six band-pass filters 101_1, 101_2, . . . , 101_6 which divides a band into predetermined six bands.
The control sound generating section 30 is arranged to direct to a region where it is desired to control a noise, and outputs a control sound which interferes with a noise.
The residual noise detecting section 40 detects a residual noise which remains after a control sound generated by the control sound generating section 30 interferes with the noise
e(t)
, and divides the detected residual noise signal by the band-pass filters 201_1, 201_2, . . . , 201_6 which divides a band into six bands.
The controlling section 300 includes six harmonic component calculating sections 301_1, 301_2, . . . , 301_6 which calculate harmonic components with respect to outputs of the six adaptive filters 102_1, 102_2, . . . , 102_6 for the respective divided bands of the control signal generating section 100; error path correction filters 302_1, 302_2, . . . , 302_6 which convolute transmission characteristics of the error path from the control sound generating section 30 to the residual noise detecting section 40 into each of the harmonic components, thereby correcting each of the harmonic components; six band-pass filters 303_1, 303_2, . . . , 303_6 which divide a residual noise signal detected by the residual noise detecting section 40 into six bands respectively corresponding to bands of the harmonic components; and six correlation calculating sections 304_1, 304_2, . . . , 304_6 which calculate correlations between the residual noise signals divided by the band-pass filters 303_1, 303_2, . . . , 303_6 and the harmonic components.
The control signal generating section 100 includes six adaptive filters 102_1, 102_2, . . . , 102_6 which perform filtering operations for reference signals in each of the bands divided by the reference signal detecting section 10, and an adder 103 which adds outputs of the six adaptive filters 102_1, 102_2, . . . , 102_6. Further, the control signal generating section 100 includes a threshold value storing section 202 which stores a threshold value, and
a switch group 203 which compares correlation values calculated by the correlation calculating sections 304_1, 304_2, . . . , 304_6 of the distortion evaluating section 300
corr1(t), corr2(t), . . . corr6(t)
with corresponding threshold values of the multiple threshold values TH1 to TH6 stored in the threshold value storing section 202 respectively, thereby selecting a band of the divided bans which is to be used for renewing a filter coefficient.
The operations of the active noise control apparatus of the first embodiment will be explained with reference to block diagrams in
In the active noise control apparatus of the first embodiment, an operation of processing both the residual noise signal and reference signal corresponding to a noise detected by the reference signal detecting section 10 by the control signal generating section 100, and an operation of processing both the control signal and residual error signal by the Controlling section 300 are executed in parallel. However, if a filtering coefficient is renewed in the adaptive filters 102_1, 102_2, . . . , 102_6, corresponding frequency components of the reference signal and the residual noise signal detected at the same time are used for the calculation.
In the block diagrams in
t
and the following processes (1) to (12) are carried out repeatedly.
(Reference Signal Detecting Section)
(1) The reference signal detecting section detects
x(t)
a reference signal.
(2) The band-pass filters 101_1, 101_2, . . . , 101_6 are applied to
x(t)
the detected reference signals, and the reference signals divided into the six bands
x
i(t) (i=1, 2, . . . ,6)
are calculated.
x
i(t)=bpfi*x(t) (i=1, 2, . . . ,6)
(The control signal generating section)
(3) Filtering coefficient of adaptive filter
h
i(t) (i=1, 2, . . . ,6)
Using the equation 23, from the divided reference signals
x
i(t) (i=1, 2, . . . ,6)
control signals in the respective bands
y
i(t) (i=1, 2, . . . ,6)
are produced.
y
i(t)=hi(t)*xi(t) (i=1, 2, . . . ,6)
(4) Control signal in respective bands
y
i(t) (i=1, 2, . . . ,6)
are added, the control signal,
y(t)
is produced and is output as a control sound from the control sound generating section 30.
(Controlling section)
(5) A residual noise signal
e(t)
is detected by the residual noise detecting section.
(6) For outputs of the adaptive filters in the respective divided bands
y
i(t) (i=1, 2, . . . ,6)
harmonic components
y
i(t)3 (i=1, 2, . . . ,6)
are calculated. Odd-order (third, fifth, . . . ) harmonics are generated due to an excessive large input to a speaker, but since the influence of third component specifically is relatively large, the fifth or higher order harmonics are omitted here.
(7) For respective harmonic components
y
i(t)3 (=1, 2, . . . ,6)
error path corrections are performed, and corrected harmonic components
hm
i(t) (i=1, 2, . . . ,6)
are calculated.
hm
i(t)=ĉ*yi(t)3 (i=1,2, . . . ,6)
(wherein
ĉ
represents a transfer function of an error path from the control sound generating section 30 to the residual noise detecting section 40)
(8) A residual noise signal
e(t)
is divided into six bands corresponding to the respective bands of the harmonic components.
e′
i(t)=bpf′i*e(t) (i=1, 2, . . . ,6)
(9) For harmonic components for the individual divided bands
hm
i(t) (i=1,2, . . . ,6)
and the residual noise signals,
e′
i(t) (i=1, 2, . . . ,6)
harmonic distortions
corr
i(t) (i=1, 2, . . . ,6)
are calculated.
(wherein
T
represents a correlation calculation range, and
L
represents a correlation calculation length)
(Residual Noise Detecting Section)
(10) For the detected residual noise signal
e(t)
a band-pass filter
bpf
i (i=1, 2, . . . ,6)
is applied,
thereby dividing the band into six, and the residual noise signal after dividing
e
i(t) (i=1, 2, . . . ,6)
are calculated.
e
i(t)=bpfi*e(t) (i=1, 2, . . . ,6)
(Control Signal Generating Section)
(11) For a band where harmonic distortions
corr
i(t) (i=1, 2, . . . ,6)
become greater than predetermined threshold values for the adaptive learning control,
TH
i (i=1, 2, . . . ,6)
, the band-divided residual noise signals are set to 0, thereby selecting a band to be used for renewing a filtering coefficient of the adaptive filter.
(Renewal of Filtering Coefficient of Adaptive Filter)
(12) By the reference signals after the band is divided
xi(t)
and the residual noise signals,
e″i(t)
the filtering coefficient of the adaptive filter
h
i(t) (i=1, 2, . . . ,6)
are renewed.
h
i(t+1)=hi(t)+μ·e″i(t)·ĉ*xi(t) (i=1, 2, . . . ,6)
(wherein
μ
represents a step size parameter, and
ĉ
represents a transfer function of an error path from the control sound generating section to the residual noise detecting section.)
The active noise control apparatus of the first embodiment is operated as described above, evaluates a generation state of a harmonic distortion in each of multiple divided bands to control the learning operation of the filtering coefficient, so that it is possible to avoid a deterioration of the noise control performance by a harmonic distortion, and to enhance the sound control effect.
In
In
The band-pass filters 401_1, 401_2, . . . , 401_6 are the same as the band-pass filters 303_1, 303_2, . . . , 303_6 of the controlling section 300 illustrated in
e(t)
into six bands corresponding to harmonic components.
The level calculating sections 402_1, 402_2, . . . , 402_6 input band components e1′ (t), . . . , e6′ (t) of the residual noise signal, respectively, calculate mean values for a predetermined time (Te) for respective band components, and obtains mean values of sound pressure levels of the respective bands.
A level calculating section i which processes the i-th (i=1, . . . , 6) band component e1′ (t) carries out, for example, the following action.
The square of e1′ (t) (e1′ (t))2 is calculated from the input ei′ (t). A total sum of values of each time of the current time and a past time which are latched in delaying devices (not illustrated), i.e., {ei′ (t)}2, {ei′ (t−1)}2, . . . , {ei′ (t−Te)}2, thereby obtaining outputs b1i of the level calculating sections 402—i by the following equation.
The threshold value estimating sections 403_1, 403_2, . . . , 403_6 input outputs b11, . . . , b16 of the six level calculating sections 402_1, 402_2, . . . , 402_6 as sound pressure levels of the respective bands, change the threshold values TH1, TH2, . . . for controlling adaptive learning operation, and output the same to the threshold value storing section 202 (see
Next, two methods of changing threshold value by the threshold value estimating sections 403_1, 403_2, . . . , 403_6 will be explained.
According to a first method of changing threshold value, a threshold value is changed in the following manner.
The control based on the first method of changing threshold value is carried out, so that it is possible to enhance the noise control performance without generating a harmonic distortion (unusual sound), even when a spectrum after sound control is changed due to a surrounding noise or an environment of the active noise control apparatus.
According to a second method of changing threshold value, a threshold value is changed in the following manner.
When a band corresponding to a harmonic component is a band where a sensitivity of a near is high, the adaptive learning operation control threshold value is set to a small value. With this, when a high harmonic distortion is easily sensed, it is possible to control such that a noise control performance is enhanced without generating a high harmonic distortion (unusual sound).
Although Filtered-X LMS algorithm is used as the adaptive algorithm in the embodiments described above, another adaptive algorithm may be used.
According to the present invention, a generating state of a harmonic distortion is evaluated and learning of a filtering coefficient in the control sound generating section is controlled so that deterioration of the noise control performance caused by the harmonic distortion can be avoided, and the sound control effect can be enhanced.
All examples and conditional language recited herein are intended for pedagogical purposes to aid the reader in understanding the principles of the invention and the concepts contributed by the inventor to furthering the art, and are to be construed as being without limitation to such specifically recited examples and conditions, nor does the organization of such examples in the specification relate to a showing of the superiority and inferiority of the invention. Although the embodiment(s) of the present invention(s) has(have) been described in detail, it should be understood that the various changes, substitutions, and alterations could be made hereto without departing from the spirit and scope of the invention.
Number | Date | Country | Kind |
---|---|---|---|
2008-108690 | Apr 2008 | JP | national |