JPH07253790A - Noise cancelling method - Google Patents

Abstract

PURPOSE:To improve the noise cancelling system in performance by identifying the transfer function of the cancel sound propagation system in parallel with the noise cancelling operation and setting the transfer function on the signal processing filter. CONSTITUTION:The cancel sound propagation system from the speaker 23 to the observation point (error microphone 24) is assumed to exist between the reference signal generating section 21 and the adaptive filter 22b to approximate the noise sound cancel system 20. When noise cancellation operates, the transfer function arithnaetic processing section 30, based on such approximated system, uses the reference signal xn, error signal en and adaptive filter coefficient wn to compute the transfer function of the cancel sound propagation system in the noise cancel system and cancels noises by adaptive signal processing based on setting the transfer function on the signal processing filter 22c.

Description

[Detailed description of the invention]

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to LMS (Least Mean Square)
re) Noise cancellation method that cancels noise by adaptive signal processing, especially in a signal processing filter that generates a reference signal for signal processing (filtered reference signal) by convolving the transfer characteristics of the cancellation sound propagation system with a reference signal The present invention relates to a noise canceling method for canceling noise by updating a transfer function in real time.

[0002]

2. Description of the Related Art A method of reducing noise (active control) by radiating a noise-canceling sound having a phase opposite to that of noise from a speaker is in the spotlight, and is being put to practical use in some indoor spaces such as factories and offices. be. A method of reducing noise by active control has also been proposed in the passenger compartment of an automobile.

[0003] FIG. 11 is a diagram showing the configuration of a conventional noise canceling system that reduces noise by active control, in the case of reducing the engine noise of an automobile. Reference _numeral 11 denotes an engine, which is a noise source; 12, a rotation speed sensor for detecting the engine rotation speed R; Department. When the noise source is an engine, the noise generated by engine rotation has periodicity, and its frequency depends on the engine speed. For example, in the case of a 4-cylinder engine, the periodic noise generated in the cabin is dominated by the second harmonic of the engine speed.
At rpm (10 rps), the frequency of noise generated in the cabin is 20 Hz, and the number of revolutions is 6000 rpm (100 r
ps), the frequency of noise generated in the cabin is 200
Hz. The reference signal generating unit 13 stores second harmonic sine wave data in a ROM, and reads out and outputs the data as necessary to generate a reference signal x _n . The timing of reading/outputting this data is controlled according to the engine speed R, so that the reference signal having the frequency of the periodic noise generated according to the engine speed R is output.

Numeral 14 denotes a noise canceling controller which receives the reference signal _xn generated from the reference signal generating section 13 and controls noise S at a noise canceling point (observation point, for example near the driver's ear) in the passenger compartment. A synthesized sound signal of _n and the canceling sound Sc _n is input as an error signal e _n , adaptive signal processing is performed so as to minimize the error signal, and a noise canceling signal y _n is output. The noise canceling controller 14 includes an adaptive signal processing unit 14a, an adaptive filter 14b having a digital filter structure, and a signal obtained by convolving the propagation characteristic (transfer function) of the canceling sound propagation system from the speaker to the noise canceling point in the reference signal _xn . It has a signal processing filter (filtered X signal generating filter) 14c for generating a processing reference signal (filtered reference signal) r _n . 15 is a DA converter that converts the output of the adaptive filter (noise-cancelling signal y _n ) into an analog noise-canceling signal, 16 is a power amplifier that amplifies the noise-canceling signal, and 17 is a canceling speaker that emits noise-cancelling sound Sc _n . , 18 are located at noise canceling points, detect a synthesized sound of noise S _n and canceling sound Sc _n , and output the synthesized sound signal as an error signal e _n , and 19 amplifies the error signal e _n . an amplifier, 20 a low-pass filter for removing aliases,
20' is an A that converts the low-pass filter output to digital;
D converter.

The adaptive signal processor 14a receives the error signal e _n at the noise cancellation point and the reference signal r _n for signal processing inputted via the signal processing filter 14c, and uses these signals to reduce the noise at the noise cancellation point. Adaptive filter 1 performs adaptive signal processing so as to cancel
Determine the coefficient of 4b. For example, the adaptive signal processing unit 14a follows the well-known filtered XLMS (Least Mean Square) adaptive algorithm so that the error signal en input from the error microphone 18 is minimized.
Determine the coefficient of b. The adaptive filter 14b receives the reference signal x _n according to the coefficients determined by the adaptive signal processing unit 14a.
is digitally filtered to obtain a noise cancellation signal y
Output _n to cancel the noise. Note that the reference signal x
_n must be a signal highly correlated with the noise S _n to be eliminated, and sounds uncorrelated with the reference signal are not eliminated.

The adaptive filter 14b, as shown in FIG. 12, is composed of an FIR type digital filter. For example, a delay element D
L, DL . . . and coefficients w _1n , w _2n ,
Multiplication units ML, ML, . _{. .} for multiplying w _3n .
and adders AD, AD . . .
・It is realized by That is, x _n is the reference signal at the current time n·Ts, and w _1n , w _2n ,
_{If w 3n . _ _} ) to output the noise cancellation signal y _n .

As shown in FIG. 13, the signal processing filter 14c is composed of an FIR type digital filter.
A delay element D that sequentially delays the input signal by one sampling time
L, DL . . . and coefficients c ₁ , c ₂ , c ₃ ·
. . are implemented by multiplication units _ML , ML, . . . that multiply cM, and addition units AD, AD, . The coefficients c ₁ , _{c 2} , c ₃ . . Assuming that the reference signal at time _n ·Ts is _xn and the output (reference signal for signal processing) is rn, the signal processing filter 14c is expressed by the following equation: rn= _Σci · _xn-i+1 (i=1 to M ) Executes the operation of (2) and outputs the signal processing reference signal r _n .

The adaptive signal processor 14a calculates the coefficients w _1n+1 , w _2n+1 , w _3n+1 , . _Nn+1 ,
Using the coefficient at the current time nT, the error signal e _n , and the signal processing reference signal r _n , the following formula (coefficient update formula) w _jn+1 =w _jn +μ·rn _-j+1 ·e _n (3 ) (where j=1, 2, . . . N). (3)
In the formula, the suffix n is the value at the current sampling time, the suffix (n+1) is the value after one sampling time,
The suffix (n-1) means the value one sampling time before, the suffix (n-2) means the value two sampling times before, and so on. Also, μ is a constant of 1 or less (step size parameter) that determines the step for updating the coefficients of the adaptive filter, and is set to an appropriate value according to the noise canceling system.

When the engine 11 rotates, its rotation speed R
is detected by the rotational speed sensor 12, and the reference signal generator 1
3 generates a reference signal _xn having a frequency corresponding to the engine speed R, and a signal processing filter 14c and an adaptive filter 14b
to enter. At this time, the periodic engine sound generated from the engine 11 propagates through the air having a noise propagation system (primary sound propagation system) having a predetermined transfer function and reaches a noise cancellation point. The signal processing filter 14c receives the reference signal x _n
is convoluted with the transfer characteristic C _M of the cancellation sound propagation system to generate a signal processing reference signal r _n and input it to the adaptive signal processing unit 14c. In parallel with the above, a synthesized sound (error signal) e _n of noise and canceled sound at the noise canceling point is detected by the error microphone 18 and input to the adaptive signal processing unit 14a via an amplifier, a low-pass filter, and an AD converter. . The adaptive signal processing unit 14a performs LMS adaptive signal processing according to equation (3) using the error signal e _n and the signal processing reference signal r _n to determine the coefficients of the adaptive filter 14b. The adaptive filter 14b applies digital filtering to the reference signal _xn in accordance with the coefficients determined by the adaptive signal processor 14a to generate a noise cancellation signal _yn .
Input to the speaker 17 via a converter and a power amplifier. As a result, the noise canceling sound is output from the speaker 17, reaches the noise canceling point via the canceling sound propagation system, and acts to cancel the noise. After that, the above operation is repeated and the noise is quickly canceled.

[0010]

[Problems to be Solved by the Invention] By the way, conventional noise
In the cancellation method, the transmission of the cancellation sound propagation system
Characteristic C_M.is identified prior to noise canceling operation, and the transmission
The reach characteristic is set in the signal processing filter 14c. and,
During noise cancellation operation, the signal processing filter 14c is referred to
The transfer function C of the cancellation sound propagation system is added to the signal_M.convoluted with
The reference signal for signal processing r_nand the signal processing unit 14a
is the signal processing reference signal r_nand error signal e _nsuitable using
Adaptive signal processing is performed to determine the coefficients of the adaptive filter 14b.
be. Such conventional noise cancellation methods are
Transfer function and noise of canceling sound propagation system
If it matches the transfer function identified before the cancellation operation,
There is no problem. However, during noise cancellation operation
The transfer function of the cancellation sound propagation system in
Affected by temperature, etc., it changes from moment to moment.
It will be different from the transfer function identified before the cancel operation.
Such deviation of the transfer function (called modeling error) occurs.
noise will degrade the performance of the noise cancellation system and
In the bad case, the system becomes unstable and cancels the noise
On the contrary, there is a risk of causing a sound increase phenomenon.
be.

In view of the above, an object of the present invention is to improve the performance of a noise canceling system by identifying the actual transfer function of the canceling sound propagation system in parallel with the noise canceling operation and setting the transfer function in the signal processing filter. It is an object of the present invention to provide a noise canceling method that can be improved.

[0012]

According to a first aspect of the present invention, there are provided means for approximating a noise cancellation system assuming that a cancellation sound propagation system exists between a reference signal generator and an adaptive filter; means for computing the correlation between the reference signal and the error signal to obtain a combination of the correlation and the adaptive filter coefficient; Canceled sound in a noise cancellation system using means for calculating a correlation, means for calculating an autocorrelation of a reference signal, a correlation value between the reference signal and a canceled sound propagation system output, and an autocorrelation value of the reference signal This is achieved by means of calculating the transfer function of the propagation system and means of setting the transfer function to the signal processing filter of the noise cancellation system.

According to a second aspect of the present invention, there are provided means for approximating a noise cancellation system assuming that a cancellation sound propagation system exists between a reference signal generator and an adaptive filter; means for obtaining a combination of the adaptive filter coefficients, means for calculating the gradient of the error mean square curve using the mean square value and the adaptive filter coefficient value, means for calculating the correlation between the error signal and the reference signal, means for calculating a transfer function of a cancellation sound propagation system in a noise cancellation system using the gradient and the correlation value (the correlation value between the error signal and the reference signal); and setting the transfer function to a signal processing filter of the noise cancellation system. and by means of

[0014]

[action]

The noise cancellation system is approximated assuming that the noise propagation system of the first invention exists between the reference signal generator and the adaptive filter. During noise cancellation operation, the correlation between the reference signal and the error signal is calculated, the combination of the correlation and the adaptive filter coefficient is obtained, and the combination of the correlation value and the adaptive filter coefficient value is used to cancel the noise in the reference signal and the approximation system. Calculate the correlation with the propagation system output. It also computes the autocorrelation of the reference signal. Next, using the correlation value between the reference signal and the output of the cancellation sound propagation system and the autocorrelation value of the reference signal, the transfer function of the cancellation sound propagation system in the noise cancellation system is calculated, and the transfer function is used in the noise cancellation system. set the signal processing filter to cancel the noise.

(Second Invention) A noise cancellation system is approximated assuming that a cancellation sound propagation system exists between the reference signal generator and the adaptive filter. During the noise canceling operation, a combination of the mean square value of the error signal and the adaptive filter coefficient is obtained, and the gradient of the error mean square curve is calculated using the combination of the mean square value and the adaptive filter coefficient value. Also, the correlation between the error signal and the reference signal is calculated. Then, using the gradient and the correlation value (the correlation value between the error signal and the reference signal), the transfer function of the cancellation sound propagation system in the noise cancellation system is calculated, and the transfer function is set in the signal processing filter of the noise cancellation system. to cancel the noise. The first above,
According to the second invention, the transfer function of the cancellation sound propagation system is identified in parallel with the noise cancellation operation, and the transfer function is set in the signal processing filter. , and the performance of the noise canceling system can be improved.

[0016]

【Example】

(a) Overall Configuration FIG. 1 is a diagram showing the configuration of a system embodying the noise canceling method of the present invention. In the figure, 20 is a noise canceling system, which generates a noise canceling sound so that the power of the error signal is minimized. 21 is a reference signal generator for generating a reference signal _xn , and 22 is a noise cancellation controller which receives the reference signal _xn generated from the reference signal generator 21 and detects the noise _Sn at the noise cancellation point in the vehicle interior. and a synthesized sound signal of the canceling sound Sc _n is inputted as an error signal e _n , adaptive signal processing is performed so that the error signal is minimized, and a noise canceling signal y _n is output. noise cancellation controller 22
is an adaptive signal processing unit 22a, an adaptive filter 22b having a digital filter configuration, and a signal processing reference signal (filtered reference signal) by _convolving a reference signal xn with a propagation characteristic (transfer function) CM of a cancellation sound propagation system. It has a signal processing filter 22c that produces r _n . In the signal processing filter 22c, the actual transfer function c of the cancellation sound propagation system calculated during the noise cancellation operation is successively set as will be described later. Therefore, even if the transfer function of the canceling sound propagation system fluctuates due to factors such as the number of passengers, the temperature in the passenger compartment, and other factors, the signal processing filter 22c is set to an actual transfer function, and the reference signal _xn is convoluted with the transfer function. Create a filtered reference signal r _n .

Numeral 23 is a canceling speaker for emitting noise canceling sound _Scn , and numeral 24 is arranged at a noise canceling point, detects a synthesized sound of noise Sn and canceling sound _Scn , and _converts the synthesized sound signal into an error signal e. It is an error microphone output as _n . A DA converter that converts the output signal (noise cancellation signal _yn ) of the adaptive filter 22b into an analog noise cancellation signal, a power amplifier that amplifies the noise cancellation signal and inputs it to a speaker, an amplifier that amplifies the error signal en, A low-pass filter that removes aliases from the error signal, an AD that converts the low-pass filter output to digital and inputs it to the noise cancellation controller 22
Although there are converters and the like, illustration thereof is omitted. Reference numeral 30 denotes a transfer function calculation processing unit for calculating the transfer function of the cancellation sound propagation system, which calculates the transfer function of the cancellation sound propagation system using the reference signal _xn , the error signal _en , and the coefficient _wn of the adaptive filter 22b. are calculated and sequentially set in the signal processing filter 22c.

(b) Transfer function calculation processing unit (b-1) First transfer function identification method A primary sound virtual propagation system (noise propagation system) H from the noise source to the observation point FIG. 2 shows the noise canceling system 20 expressed using the next sound propagation system (canceling sound propagation system) C _R .
In the figure, 25 is a noise propagation system H, 26 is a cancellation sound propagation system C _R , and the cancellation sound propagation system 26 also includes speaker characteristics. If adaptation occurs relatively slowly,
The noise cancellation system 20 can be approximated as shown in FIG. That is, even if adaptive signal processing is performed assuming that the cancellation sound propagation system 26 exists between the reference signal generator 21 and the adaptive filter 22b, adaptive filter coefficients equivalent to those of the noise cancellation system 20 in FIG. 2 can be obtained. By approximating the noise cancellation system as described above, the following equations (4) and (5) hold. In the formula, the underlined is a one-row matrix (vector), and T means a transposed matrix. The suffix n means the value at the current sampling time, and ni means the value i sampling before the current time.

E[ _xn e _n ] ₌ E[ _xn ( _dn + _unTw )] ₌ E[ _xndn ]+E[ ^xnunT ] ^w (4) _where w = [ w ₀ w ₁ w ₂ ... w _I-1 ] ^T u _n = [u _n u _n-1 un _-2 ... u _n-I+1 ] ^T x _n = [x _n x _n-1 x n _-2 ... _xn-J+1 ] _TE [ ^xnjuni ]=E[ _xnjxniT ^] c (5) where _i =0 to (I-1), _j =0 to J ₋₁ c _{= [ c 0 c 1} ^c ₂ . can be approximated by the average sum of the products of the reference signal x _n and the error signal e _n at N sampling times (=Σ x _n e _n /N). Note that N is 100, for example. Other E[ ] have similar meanings, for example, E[ x _n u _n ^T ] is the correlation between the reference signal x _n and the cancellation sound propagation system output signal u _n ^T . Also, in equation (5), E[x _nj x _{n− i} ^T ] is the autocorrelation of the reference signal.

In equation (4), the correlation E[ x _n e _n ] between the reference signal and the error signal and the adaptive filter coefficient w are known,
E[ x _n d _n ] and E[ x _n u _n ^T ] are unknown. Therefore,
Let the combination of the correlation E[ x _n e _n ] and the adaptive filter coefficient w be 2
A set is obtained, and simultaneous equations are created based on the equation (4) from these two sets of data, and E[ x _n d _n ] and E[ x _n u _n ^T ] are obtained by solving the simultaneous equations. On the other hand, the autocorrelation E[x _nj x _ni ^T ] of the reference signal, which is the right side of Equation (5), is calculated, and using this and E [ x _n u _n ^T ] from Equation (5), cancellation sound propagation The transfer function c of system 26 can be determined.

(b-2) First Configuration of Transfer Function Operation Processing Section FIG. 4 shows the transfer function c calculated based on the first transfer function identification method.
31 is a configuration diagram of a transfer function arithmetic processing unit 30 that calculates
reference signal x_nand error signal e_nCorrelation E[x _n e_n]
A first correlation calculation unit 32 for calculating the reference signal and the error signal
Correlation E[x _n e_n] and adaptive filter coefficientswtwo unions with
E[x _n d_n] and E[x _n u _n ^T.] respectively
Create a system of equations with unknowns and solve the system
so that E[x _n u _n ^T.] to output the second correlation operation
Department. In addition, E[x _n u _n ] is the reference signalx _n and the approximation
Output signal of the cancellation sound propagation system 26 in the systemu
_n ^T. It is a correlation with 33 is the autocorrelation E[x
_njx_ni ^T.], 34 is the correlation E
[x _n u _n ^T.] and the autocorrelation E[x_njx_ni ^T. ]
is input, and the transmission
achievement functioncis a transfer function calculation unit that calculates and outputs
It should be noted that the second correlation calculator 32 needs to solve the simultaneous equations.
However, this solution will be discussed later.

(b-3) Second transfer function identification method When the noise canceling system is approximated as shown in FIG. 3, equations (6) and (7) hold. In the formula, the underline ( ) is a one-row matrix (vector). However, only R is a matrix. Also, T means a transposed matrix. The suffix n is the value at the current sampling time, n−i
means the value i samples before the current time. E[e _n ² ]=E[d _n ² ]+2 w ^T p+ w ^T R w (6) where: p = E [ u _n d _n ] R = E [ u _n u _n ^T ] ∇ = 2 p + 2 R w (7) where ∇ = [ ∇ ₀ ∇ ₁ ∇ ₂ . . . ∇ _I-1 ] ^T In equation (6), E[e _n ² ] is the autocorrelation of the error signal, which is the average sum of the squares of the error signal e _n at N sampling times (= Σe _n ² /N). Note that N is 100, for example. The square of the error signal e _n is a quadratic function (error characteristic curved surface) of the adaptive filter coefficient w as shown in FIG. ^In the adaptive signal processing, the adaptive filter coefficient w
(=*w) is determined.

Also, ∇ means the gradient of the error characteristic curve, which is obtained by differentiating the equation (6) with respect to w. on the other hand,
The gradient ∇ _i of the error characteristic surface can also be expressed as in equation (8). ∇ _i =E[e _n x _ni ^T ] c (8) In equation (6), the root mean square value E[e _n ² ] of the error signal e _n
and adaptive filter coefficients w are known, E[d _n ² ], p and R are unknowns. Therefore, the mean square value E of the error signal e _n
Obtaining combinations of [e _n ² ] and adaptive filter coefficients w, creating simultaneous equations based on each of these combinations and equation (6),
By solving the simultaneous equations, p and R can be obtained. This p and R are
By substituting into equation (7), the gradient ∇ of the error characteristic surface can be obtained. Therefore, the correlation E[ _en x
_ni ^T ], and using this correlation and the gradient ∇ obtained from the equation (7), the transfer function c of the cancellation sound propagation system 26 can be obtained from the equation (8).

(b-4) Second configuration of transfer function calculation processing section FIG. 6 is another configuration diagram of transfer function calculation processing section 30 for calculating transfer function c by the second transfer function identification method. _is an error mean square calculator for calculating the mean ^square value E [ _en ² ] of the error signal e _n ; , R as unknowns, and outputs p and R by solving the simultaneous equations.
Note that p is the correlation matrix of u _n and d _n , and R is the autocorrelation matrix of u _n . 43 is a gradient calculator that calculates the gradient ∇ by equation (7); 44 is a correlation calculator that calculates the correlation between the error signal e _n and the reference signal x _ni ^T ; 45 is the gradient ∇ and the correlation E[e
_n x _ni ^T ], and calculates and outputs the transfer function c of the cancellation sound propagation system from the equation (8).

(c) Overall operation For example, when canceling the engine sound in the passenger compartment, the previously measured transfer function of the cancellation sound propagation system is initially set in the signal processing filter 22c. In this state, when the noise canceling operation starts, the reference signal generator 21 generates a reference signal x _n having a frequency corresponding to the engine speed,
It is input to the signal processing filter 22c and the adaptive filter 22b. At this time, the engine sound generated from the engine propagates through the air having a noise propagation system (primary sound propagation system) having a predetermined transfer function and reaches a noise cancellation point. The signal processing filter 22c convolves the initially set transfer characteristic C _M of the cancellation sound propagation system with the reference signal x _n to generate a signal processing reference signal r _n and inputs it to the adaptive signal processing unit 22 a.
In parallel with the above, the synthesized sound (error signal) e _n of the noise and the cancellation sound at the noise cancellation point is generated by the error microphone 2
4 and input to the adaptive signal processing unit 22a.

The adaptive signal processing section 22a performs LMS adaptive signal processing according to the equation (3) using the error signal e _n and the signal processing reference signal r _n to determine the coefficient w _n of the digital filter of the adaptive filter 22b. . adaptive filter 22
b performs digital filter processing on the reference signal xn according to the coefficients determined by the adaptive signal processing unit 22a to generate a noise cancellation signal _yn , which is input to the speaker 23 via a DA converter and power amplifier (not shown). As a result, the noise canceling sound is output from the speaker 23, reaches the noise canceling point via the canceling sound propagation system, and acts to cancel the noise. After that, the above operation is repeated. On the other hand, in parallel with the noise cancellation control by the adaptive signal processing, the transfer function arithmetic processing unit 30 performs the reference signal x _n , the error signal e _n , the adaptive filter coefficient w
Using _n , the transfer function c of the cancellation sound propagation system in the current state is calculated, and the signal processing filter 22c
set to When a new transfer function is set, the signal processing filter 22c convolves the transfer function with the reference signal x _n to generate the signal processing reference signal r _n , and the adaptive signal processing unit 22
Enter in a. Thereafter, similarly, the transfer function of the canceling sound propagation system in the current state is set to the signal processing filter 22c at all times, adaptive signal processing is performed using the transfer function, and noise is quickly canceled.

(d) Adaptive Solution of Equations The first transfer function identification method requires solving simultaneous linear equations. In the following, the method for solving the simultaneous linear equations with two variables will be described as an example. First, a method for solving a linear equation with one element will be described, and then a method for solving a simultaneous linear equation with two variables will be described. (d-1) Adaptive solution of the linear equation with one variable The solution of the linear equation with one variable ax=b (x is an unknown number) is transformed into e=b−ax (9) and adapted so that the error e becomes zero. when x
is the value of A block diagram of equation (9) is shown in FIG. However, 51 is a multiplier and 52 is an adder.

From the equation (9), the square of the error e is e²=b²−2abx+a²x² , which is a quadratic function of x (error characteristic curve). above formula
Differentiating with respect to x and obtaining the gradient ∇ of the error characteristic curve, the following equation ∇ = de²/dx=-2ab+2a²x becomes. If the error e is controlled to be the minimum (zero),
In other words, the current time value x_nin the gradient direction by a predetermined value μ (μ
is the step size parameter and is less than or equal to 1)
so that the next value x_n+1and then x_n+1the current time value x_n
sequentially as x_n+1in the same way
do. i.e. x_n+1the following formula x_n+1= x_n−μ∇_n = x_n-μ(-2ab+2a²x_n) = x_n+2 μa(b−ax_n) = x_n+2 μae_n(Ten) By successively obtaining ∇_n= 0 (E
minimum) and x at that time_n(=*x_n) requires
It is a solution of the equation ax=b.

By the way, equation (10) has the same structure as equation (3) for obtaining coefficients of an adaptive filter in adaptive signal processing. Therefore, b in FIG. 7 is a target value, e is an error signal, ax is a cancel signal for canceling the target value,
a is a reference signal, x is an adaptive filter coefficient obtained by adaptive signal processing so that e=0, 51 is an adaptive filter,
52 corresponds to an error detection sensor. From the above, the solution of the one-dimensional linear equation ax=b (where x is an unknown number) can be obtained as x _n when the error e _n converges (zero) by the configuration shown in FIG. 8, 51 and 53 are multipliers, 52 and 54 are adders, 55 is a multiplier for multiplying by a coefficient of 2μ, and 56 is a delay section (register) for delaying _xn+1 by one sampling time.

(d-2) Adaptive solution of binary simultaneous linear equations binary system of linear equations a₁₁x₁+a₁₂x₂=b₁ a_{twenty one}x₁+a_{twenty two}x₂=b₂ The solution of the above system of equations is e₁=b₁-(a₁₁x₁+a₁₂x₂) e₂=b₂-(a_{twenty one}x₁+a_{twenty two}x₂) (11) and the error (e₁ ²+ e₂ ²) is minimized (zero)
x when adapted to₁, x₂is the value of (11)
When shown in a block diagram, it becomes as shown in FIG. However, 61-6
4 is a multiplier, 65 to 68 are adders, 71 and 72 are sum-of-products circuits.
is the road.

error e₁, e₂Each square of e₁ ²=b₁ ²-2b₁a₁₁x₁-2b₁a₁₂x₂+a₁₁ ²x₁ ²+2a₁₁a₁₂x₁x₂+a
₁₂ ²x₂ ² e₂ ²=b₂ ²-2b₂a_{twenty one}x₁-2b₂a_{twenty two}x₂+a_{twenty one} ²x₁ ²+2a_{twenty one}a_{twenty two}x₁x₂+a
_{twenty two} ²x₂ ² and the above equations are x₁, x₂Differentiating by ∂e₁ ²/∂x₁= -2b₁a₁₁+2a₁₁ ²x₁+2a₁₁a₁₂x₂ ∂e₁ ²/∂x₂= -2b₁a₁₂+2a₁₂ ²x₂+2a₁₁a₁₂x₁ ∂e₂ ²/∂x₁= -2b₂a_{twenty one}+2a_{twenty one} ²x₁+2a_{twenty one}a_{twenty two}x₂ ∂e₂ ²/∂x₂= -2b₂a_{twenty two}+2a_{twenty two} ²x₂+2a_{twenty one}a_{twenty two}x₁ becomes. Therefore, the gradients of the error characteristic surfaces are given by
∇ to be represented₁, ∇₂given by the sum of

[0032]

[Number 1] Total power of error e₁ ²+ e₂ ²is the minimum (zero)
The current time value x_1n, x_2nof
The next value x
_1n+1, x_2n+1and then x_1n+1, x_2n+1the current time
the value of x_1n, x_2nsequentially as x_1n+1, x_2n+1similarly sought
It means that you should go ahead. i.e. x_1n+1，
x_2n+1are sequentially obtained by the following formulas,
Finally, ∇₁＋∇₂= 0 (minimum error) and
x_1n, x_2nis the solution of the simultaneous linear equation with two variables.

[0033]

[Number 2] By the way, the formula (12) has the same structure as the coefficient update formula for obtaining the coefficients of the adaptive filter in the adaptive signal processing.
Therefore, b ₁ and b ₂ in FIG. 9 are target values at two observation points, e ₁ and e ₂ are error signals at two observation points,
(a ₁₁ x ₁ +a _{1 2} x _2n ), (a ₂₁ x ₁ +a ₂₂ x ₂ ) are cancel signals for canceling the target values, a _{1 1} , a ₁₂ ,
a ₂₁ and a ₂₂ are reference signals, x ₁ and x ₂ are adaptive filter coefficients obtained by adaptive signal processing so that e ₁ =0 and e ₂ =0, 71 and 72 are adaptive filters, and 67 and 68 are Corresponds to the error detection sensor.

From the above, the solution of the two-dimensional simultaneous linear equations can be obtained as x _1n and x _2n when the errors e _1n and e _2n converge (zero) by the configuration shown in FIG. In FIG. 10, 71, 72, 73 and 74 are sum-of-products circuits;
5 and 76 are multipliers for multiplying the coefficient 2μ, 77 and 78 are 1
Delay units (registers) 79 and 80 for delaying x _{1 n+1} and x _{2 n+1} respectively by the sampling time are adders.
By the way, in the transfer function identification method of the present invention, it is generally necessary to solve simultaneous linear equations, which can be obtained by an adaptive solution method as in the case of binary simultaneous linear equations,
The details are omitted. Although the present invention has been described above with reference to the embodiments, the present invention can be modified in various ways according to the gist of the invention described in the claims, and the present invention does not exclude these modifications.

[0035]

As described above, according to the present invention, the transfer function of the cancellation sound propagation system is identified in parallel with the noise cancellation operation,
Since the transfer function is set in the signal processing filter, the transfer function of the actual cancellation propagation system can be set in the signal processing filter, and the performance of the noise cancellation system can be improved.

[Brief description of the drawing]

FIG. 1 is an explanatory diagram of a system embodying a noise cancellation method of the present invention;

FIG. 2 is an explanatory diagram of a noise cancellation system expressed using a primary sound and secondary sound propagation system;

FIG. 3 is an explanatory diagram of an approximation system;

FIG. 4 is a configuration diagram of a first transfer function calculator;

FIG. 5 is an explanatory diagram of an error characteristic curve;

FIG. 6 is a configuration diagram of a second transfer function calculator;

FIG. 7 is an explanatory diagram of adaptive solution of a linear equation with one variable;

FIG. 8 is a configuration diagram of adaptive solution of a linear equation with one element;

FIG. 9 is an explanatory diagram of adaptive solution of a binary system of linear equations;

FIG. 10 is a configuration diagram of an adaptive method for solving a system of binary linear equations;

FIG. 11 is a configuration diagram of a conventional noise cancellation device;

FIG. 12 is a configuration diagram of an adaptive filter;

FIG. 13 is a configuration diagram of a signal processing filter;

[Description of symbols]

20... noise cancellation system 21 reference signal generator 22a... adaptive signal processing unit 22b... adaptive filter 22c signal processing filter 23 speaker 24... error microphone 30 .. Transfer function calculation processing unit

Claims (2)

[Claims] 1. A canceling sound generator for outputting a canceling sound for canceling noise at a noise canceling point,
A sensor that detects an error signal that is a synthesized sound of noise and canceled sound at the noise cancellation point, a reference signal generator that generates a reference signal corresponding to the noise generated from the noise source, and propagation of the canceled sound from the canceled sound generator to the sensor. A signal processing filter that generates a reference signal for signal processing by convolving the propagation characteristics in the system with the reference signal, and a noise cancellation signal that is generated and canceled by performing a predetermined filtering process on the reference signal output from the reference signal generation unit. Noise in a noise cancellation system comprising an adaptive filter input to a sound generator and an adaptive signal processor that determines coefficients of the adaptive filter so as to cancel noise at a noise cancellation point using an error signal and a reference signal for signal processing In the cancellation method, the noise cancellation system is approximated assuming that a cancellation sound propagation system exists between the reference signal generator and the adaptive filter, the correlation between the reference signal and the error signal is calculated, and the combination of the correlation and the adaptive filter coefficient is calculated. using these correlation values and adaptive filter coefficient values to calculate the correlation between the reference signal and the output of the cancellation sound propagation system in the approximation system; a correlation value with the propagation system output;
A noise cancellation method comprising calculating a transfer function of a cancellation sound propagation system in a noise cancellation system using the autocorrelation value of the reference signal and setting the transfer function in a signal processing filter in the noise cancellation system to cancel noise. 2. A canceling sound generator for outputting a canceling sound for canceling noise at a noise canceling point,
A sensor that detects an error signal that is a synthesized sound of noise and canceled sound at the noise cancellation point, a reference signal generator that generates a reference signal corresponding to the noise generated from the noise source, and propagation of the canceled sound from the canceled sound generator to the sensor. A signal processing filter that generates a reference signal for signal processing by convolving the propagation characteristics in the system with the reference signal, and a noise cancellation signal that is generated and canceled by performing a predetermined filtering process on the reference signal output from the reference signal generation unit. Noise in a noise canceling system comprising an adaptive filter input to a sound generator and an adaptive signal processor that determines coefficients of the adaptive filter so as to cancel noise at a noise cancellation point using an error signal and a reference signal for signal processing In the cancellation method, a noise cancellation system is approximated assuming that a cancellation sound propagation system exists between a reference signal generator and an adaptive filter, a combination of the mean square value of the error signal and the adaptive filter coefficient is obtained, and these mean square values and computing the slope of the error mean square curve using the adaptive filter coefficient values, computing the correlation between the error signal and the reference signal, and computing the noise using the gradient and the correlation value between the error signal and the reference signal A noise canceling method of calculating a transfer function of a canceling sound propagation system in a canceling system and setting the transfer function in a signal processing filter in the noise canceling system to cancel noise.