JPH0764572A - Filtered reference signal generation system - Google Patents

Abstract

PURPOSE:To decrease the number of times of arithmetic operation to be executed to generate a filtered reference signal. CONSTITUTION:In a noise cancel controller 74 which generates a filtered reference signal (rn) by superimposing the propagating characteristic of a cancel sound propagation system on a reference signal (xn), operates an adaptive signal processing based on a filtered XLMS algorithm by using an error signal (en) at a noise cancel point and the filtered reference signal (rn), and decides the coefficient of an adaptive filter 74b so that a noise at a noise cancel point can be canceled, a filtered reference signal generating filter 74a is constituted of delay characteristic applying parts 27 and 28 and gain varying parts 29 and 30. The delay characteristic applying parts apply a delay characteristic corresponding to a frequency (f) of the reference signal (xn) to the reference signal, and the gain varying parts decide a gain G according to the frequency of the reference signal. Then, the gain is allowed to act on the outputs of the delay characteristic applying parts, and the filtered reference signal (rn) is outputted.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a filtered reference signal generating method in a noise canceling system, and more particularly, it is adapted to cancel noise at a noise canceling point by performing adaptive signal processing based on a filtered X LMS algorithm. The present invention relates to a filtered reference signal generation method in a noise cancellation system that determines a filter coefficient.

[0002]

2. Description of the Related Art One of the techniques used for noise control is active noise control in which a noise canceling sound having a phase opposite to that of noise is emitted from a speaker to reduce the noise. This is ANC (Ac
Also called tive noise control), it is being put to practical use in some indoor spaces such as factories, offices, and automobiles. The noise control system that is currently the mainstream is a feedforward control system using an adaptive filter as shown in FIG. 13, and a filtered X LM is used as an adaptive algorithm.
S (Filtered-X Least Mean Square) is used.

FIG. 13 shows an example in which there is one noise source (primary sound source), one secondary sound source (speaker) that generates cancellation sound, and one noise cancellation point (observation point). 10
Is a primary sound source that is a noise source. When the noise source is an engine, noise generated by engine rotation has a 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 passenger compartment is dominated by the second harmonic of the engine speed, and when the rotation speed is 600 rpm (= 10 rps), the frequency of the noise generated in the passenger compartment is 20 Hz. , The rotation speed is 6000 rpm
At (= 100 rps), the frequency of noise generated in the vehicle compartment is 200 Hz. Reference numeral 11 is a noise cancellation controller that operates to cancel noise (primary sound) xn at the observation point, 12 is a primary sound virtual propagation system indicating a system in which noise propagates from the noise source to the observation point, and 13 is a cancellation sound (secondary sound). Sound is emitted from the speaker, 14 is a secondary sound propagation system (canceling sound propagation system) that indicates the system in which the cancel sound propagates from the speaker to the observation point, including the characteristics of the speaker, and 15 is arranged at the observation point The error microphone detects a synthesized sound of Sn and the cancel sound Sc and outputs the synthesized sound signal as an error signal en.

The noise canceling controller 11 receives the noise xn generated from the noise source as a reference signal and the error signal en at the observation point.
Filtered X LM to minimize signal power
Adaptive noise signal processing based on the S algorithm is performed to output the noise cancellation signal yn. The noise cancellation controller 11 includes an adaptive signal processing unit 11a that performs adaptive signal processing based on the filtered X LMS algorithm.
And an adaptive filter 11b having a digital filter configuration, and a filtered reference signal obtained by convoluting the noise (reference signal) xn generated from the noise source with the propagation characteristic (transfer function C) of the cancel sound propagation system from the speaker to the noise cancellation point. It has a filtered reference signal creation filter 11c for creating rn.

The adaptive signal processing unit 11a receives the error signal en at the noise canceling point and the filtered reference signal rn input through the filter 11c,
Adaptive signal processing based on the filtered X LMS algorithm is performed so as to cancel the noise at the noise canceling point by using these signals, and the adaptive filter 11b.
Determine the coefficient of. The adaptive filter 11b digitally filters the reference signal xn in accordance with the coefficient determined by the adaptive signal processing unit 11a to generate the noise cancellation signal yn.
To cancel the noise. Noise that is not correlated with the reference signal xn is not deleted.

As shown in FIG. 14, the adaptive filter 11b is composed of an FIR type digital filter, and for example, delay elements DL, DL ... Which sequentially delay the input signal xn by one sampling time, and a coefficient for each delay element output. w ₁ (n), w
₂ (n), w ₃ (n) ... Multipliers ML and M that multiply w _N (n)
L, ... And an adder A for sequentially adding the outputs of the respective multipliers
It is realized by D, AD ... That is, the current time n · T
The reference signal in s is xn, and the coefficient of each multiplier at that time is w
Assuming that ₁ (n), w ₂ (n), w ₃ (n) ... w _N (n) and the output (noise cancellation signal) are yn, the adaptive filter 11b is

[0007]

[Equation 1] Then, the noise cancellation signal yn is output.

As shown in FIG. 15, the filtered reference signal generating filter 11c is composed of an FIR type digital filter, and for example, delay elements DL, DL ... Which delay the input signal sequentially by one sampling time and each delay. Multiplier _M that multiplies the element output by the coefficients c ₁ , c ₂ , c ₃ ... c M
Are realized by L, ML, ... And addition units AD, AD, ... Which sequentially add the outputs of the respective multiplication units. Coefficients c ₁ , c ₂ , c ₃ ...
・ C _M is the secondary sound propagation system (system from the speaker to the observation point) 1
4 is determined so as to simulate the propagation characteristic of No. 4. Time n
If the reference signal at Ts is xn and the output (filtered reference signal) is r (n), the filter 11c is

[0009]

[Equation 2]

## EQU1 ## The filtered reference signal r (n) is output by executing the operation of. The adaptive signal processing unit 11a has 1
Coefficients w ₁ (n + 1), w ₂ (n + 1), w ₃ (n +) of the adaptive filter 11b at the next time (n + 1) · Ts after the sampling time Ts
1) ... w _N (n + 1) is the coefficient w at the current time n · Ts
₁ (n), w ₂ (n), w ₃ (n) ... w _N (n) and error signal en
And the filtered reference signal rn, the following equation (coefficient update equation) is used.

[0011]

[Equation 3]

However, the j-th filter coefficient updating formula is given by w _j (n + 1) = w _j (n) + μ · r (n-j + 1) · en (4). In equations (3) and (4), (n) is the value at the current sampling time, (n + 1) is the value one sampling time later, and (n-
1) means the value one sampling time before, (n-2) means the value two sampling times before, ... Further, μ is a constant (step size parameter) that determines the step of updating the coefficient of the adaptive filter, and is set to an appropriate value according to the noise cancellation system.

The noise xn generated from the noise source 10 propagates through the primary sound propagation system 12 having a predetermined transfer function H, reaches the noise cancellation point, and becomes the noise Sn. Further, the noise xn is input to the noise canceling controller 11 as a reference signal. The error microphone 15 outputs a synthesized sound of the noise Sn at the noise canceling point and the noise canceling sound Sc output from the speaker 13 as an error signal en. The filtered reference signal generation filter 11c outputs the filtered reference signal rn by executing the operation of Expression (2). The adaptive signal processing unit 11a receives the filtered reference signal rn and the error signal en detected by the error microphone 15 and performs adaptive signal processing based on the equation (3) to determine the coefficient of the adaptive filter 11b. The adaptive filter 11b filters the reference signal xn according to the equation (1) to generate the noise cancellation signal yn and inputs it to the speaker 13. The speaker 13 outputs the noise cancellation signal y
A noise canceling sound is output based on n. The noise canceling sound reaches the noise canceling point via the secondary sound propagation system 14 and cancels the noise Sc. After that, the above control is performed every sampling time Ts, and the error signal en gradually becomes smaller and the noise is canceled.

The above is the case where there is one primary sound source (noise source), one secondary sound source (speaker), and one observation point.
A noise canceling system having K noise sources, M speakers, and L observation points is shown in FIG.
The same parts as those in FIG. 13 are designated by the same reference numerals, the primary sound source is omitted, and the speaker is included in the secondary sound propagation system 14. xain to xaKn are noises generated from K noise sources,
yhin to yhLn are noises to be canceled at each observation point, r111n to rLMKn are filtered reference signals,
ya1n to yaMn are M noise cancel signals output from the adaptive filter 11b, and e1n to eLn are error signals at each observation point. Reference numeral 16 is a signal synthesizing unit expressing the function of the microphone at each observation point, and adding units 16 ₁ to 1
6 ₁ ′ corresponds to the microphone at the first observation point, and the addition unit 1
6 _{2 to} 16 ₂ ′ correspond to the microphone at the second observation point,
..., an adder 16 _L ~ 16 _L 'corresponds to the microphone in the L observation point. ddin to ddLn are external noises that are not to be canceled at each observation point.

[0015]

In the conventional noise canceling system, even if each of the primary sound source (noise source), the secondary sound source (speaker) and the observation point is (1),
The calculation of equations (2) and (3) must be performed within one sampling time. In the case of periodic noise (muffled sound) such as engine sound, the number of taps of the adaptive filter 11b (see FIG. 14).
The number of multipliers ML in N = N) is relatively small at 10 to 20 taps. However, the number of taps of the filtered reference signal generation filter 11c (the number of multipliers ML in FIG. 15 = M) requires many taps because the propagation characteristic of the secondary sound propagation system is complicated, and is usually about 100 taps. You will need it. Therefore, when the sampling frequency is 3 KHZ, the amount of calculation for each sampling time becomes extremely large.
That is, when the filtered reference signal is generated by the expression (2), if the number of taps of the filter 11C is M, M times of multiplication and M-1 times of addition are required. on the other hand,
In the convolution of the equation (1), assuming that the number of taps of the adaptive filter 11b is N, it is necessary to perform N multiplications and N−1 additions, and in the coefficient update by the equation (3), N + 1 multiplications and N multiplications are required. Must be added. Therefore, every 1 sampling time
2N + M + 1 multiplications and 2N + M-2 additions must be performed. Therefore, if M is larger than N, the total amount of calculation greatly depends on the number M of taps of the filter 11c. Also, the above is the case where the noise source, the speaker, and the observation point are each one, but if these are two or more, the amount of calculation will increase more and more.

The noise canceling controller 11 is a DSP
Although it is composed of a (digital signal processor), most of the computing power of the DSP is used for generating the filtered reference signal. If the DSP has sufficient computing power, it is still good, but if it is insufficient, high speed,
An expensive DSP is required, the number of filter taps must be reduced, or the sampling time must be lengthened. If the number of taps is reduced, the calculation accuracy deteriorates, and if the sampling time is lengthened, the convergence speed of the adaptive filter decreases.

Therefore, a noise canceling method has been proposed in which the amount of calculation is reduced by using a synchronous adaptive filter.
According to the method using the synchronous adaptive filter, the number of calculations can be reduced, but there is a limitation that it cannot be used when the reference signal frequency (frequency of noise signal to be canceled) fluctuates. In particular, since the noise generated in the automobile is dominated by the second harmonic of the engine speed, the reference signal frequency also changes according to the engine speed, which cannot be adopted when canceling the engine sound. There is. Further, the method using the synchronous adaptive filter has a problem that it is not possible to cancel random noise in a narrow band.

From the above, it is an object of the present invention to provide a filtered reference signal generation system capable of reducing the amount of calculation executed in one sampling time in the adaptive signal processing with almost no deterioration in the silencing performance. Another object of the present invention is to provide a filtered reference signal generation method capable of reducing the amount of calculation executed in one sampling time for generating a filtered reference signal with almost no deterioration in silencing performance. Still another object of the present invention is to provide a filtered reference signal generation method capable of canceling narrow band periodic noise and random noise, and reducing the amount of calculation with almost no deterioration of the silencing performance.

[0019]

According to the present invention, there is provided a delay characteristic imparting section for imparting a delay characteristic whose delay time changes according to the frequency of a reference signal to a reference signal, and a frequency characteristic of the reference signal. This is achieved by configuring a filtered reference signal generation filter with a gain varying unit that varies the gain accordingly.

[0020]

The delay characteristic imparting section imparts the delay characteristic corresponding to the frequency of the reference signal to the reference signal, the gain varying section determines the gain in accordance with the reference signal frequency, and the gain is output to the output of the delay characteristic imparting section. It acts and outputs a filtered reference signal. Also, a signal synthesizing unit is provided, the delay characteristic adding unit adds different delay characteristics to the reference signal according to the reference signal frequency and outputs a plurality of delay signals, and the gain changing unit sets each delay signal to the frequency of the reference signal. The signal combining circuit outputs a filtered reference signal by combining the outputs of the gain changing units by applying a predetermined gain corresponding to the signal.

[0021]

【Example】

(a) Filter for generating filtered reference signal of the present invention The present invention focuses on "when the noise signal (reference signal) has a narrow band, the response of the secondary sound propagation system can be approximated mainly by a steady response component" To form a filtered reference signal generating filter. In this way, approximation with the stationary response component can reduce the number of taps of the filtered reference signal generation filter, and can significantly reduce the amount of calculation for generating the filtered reference signal.

Embodiment in which Reference Signal Frequency is Constant FIG. 1 shows a filter when the reference signal frequency is constant.
It is an explanatory view of a filter for generating a reference signal.
In FIG. 1 (a), 11c is a conventional FIR digital filter.
A filter-configured filter for generating a filtered reference signal.
And has the structure shown in FIG. That is,
DL sequentially delays the reference signal xn for one sampling time
Delay element, ML is a coefficient C for each delay element output_R0, C_R1, C
_R2, ・ ・ ・ C_RTr-1Is a multiplier for multiplying by
It is an adder that sequentially adds forces. Coefficient C_R0, C_R1, C_R2,
... C _RTr-1Is the secondary sound propagation system (from the speaker to the
It is decided to simulate the propagation characteristics of the system).

The response when a single frequency signal (sine wave) xn expressed by xn = sin (2πfnT) 0 ≦ n xn = 0 n <0 is input to the filtered reference signal generating filter 11c is The sum of the transient response component and the steady response component can be expressed by the following equation: Response (output) = (transient response component) + (steady response component). Here, the transient state is FI
The R filter continues for the maximum delay time ((the number of taps-1) x 1 sampling time), and the output thereafter is only the steady response component. The steady response component rn is represented by the following equation rn = G · sin (2πfnT + θ), and is a signal having the same frequency as the input but different amplitude and phase from the input.

Therefore, if the transient response component is ignored and the system response is approximated only by the steady response component, the input / output relationship of the filtered reference signal generating filter 11c is as shown in FIG. 1 (b). Can be modeled with a simple system. That is, it can be modeled by the delay characteristic imparting unit 21 that imparts the delay characteristic of the delay time d to the input signal and the gain imparting unit (amplifier) 22 that applies the gain G to the delay signal.
According to the filter for generating the filtered reference signal of FIG. 1B, by realizing the delay characteristic adding unit with the tap delay line (memory), the gain G is added to the delay signal.
The filtered reference signal rn can be generated by only one product operation of multiplying by. That is, by generating the filtered reference signal using the configuration of FIG. 1B, it is possible to significantly reduce the amount of calculation as compared with the conventional method. Incidentally, in the conventional ANC system, there are many cases where there are a plurality of observation points (control points) as shown in FIG. 16, and the amount of calculation required to generate the filtered reference signal becomes enormous, so this effect Is big.

Embodiment in which the frequency of the reference signal changes When the frequency f of the reference signal xn changes, the amplitude G and the phase θ of the steady response signal rn change according to the frequency f. Therefore, when the frequency f of the reference signal xn changes, the delay time d and the gain G are changed according to the change of the frequency f as shown in FIG. 2, and the transfer function of the secondary sound propagation system at each frequency is changed. It suffices to approximate C (f). Figure 2
1, (a) is a configuration diagram of a filter for generating a filtered reference signal when the frequency f is constant (corresponding to FIG. 1 (b)), (b) is a configuration diagram in which the delay time d and the gain G are variable, (c) is a principle configuration diagram for varying the delay time d and the gain G based on the frequency f of the reference signal xn, and (d) is the delay time d and the gain G based on the frequency f of the reference signal xn.
It is a concrete block diagram which makes variable.

In FIGS. 2 (b) and 2 (c), 23 is a delay time characteristic imparting section for varying the delay time, 24 is a variable gain amplifier, and 2 is a variable gain amplifier.
Reference numeral 5 is a delay time calculation unit that determines the delay time d according to the frequency f, and reference numeral 26 is a gain calculation unit that determines the gain G according to the frequency f. The delay time calculator 25 calculates the delay time d based on the frequency f of the reference signal xn and inputs the delay time d to the delay characteristic imparting unit 23. The delay time imparting unit 23 delays the reference signal xn by the input time d. Output. On the other hand, the gain calculator 26 calculates the gain G based on the frequency f of the reference signal xn and inputs it to the variable gain amplifier 24. The variable gain amplifier 24 amplifies the delay signal _xnd with a gain G and outputs a filtered reference signal rn.

FIG. 3 shows the transmission of the secondary sound propagation system in the passenger compartment.
It is a function explanatory diagram, (a) is a gain characteristic, (b) is a phase characteristic
Yes, and each horizontal axis is frequency. In the present invention, the frequency
Several areas are shown below, for example, f ≦ f₁  f₁<F ≦ f₂  f₂<F ≦ f₃  f₃<F ≦ f_Four  f_Four<F ・ ・ ・ (5) Divide into 5 areas, and gain characteristic and position
Linearly approximate the phase characteristics.

The gain calculation unit 26 calculates the following equation G (f) = a ₁ f + b ₁ (f ≦ f ₁ ) G (f) = a ₂ f + b ₂ (f ₁ <f ≦ f ₂ ) G (f) = a ₃ f + b ₃ (f ₂ <f ≦ f ₃ ) G (f) = a ₄ f + b ₄ (f ₃ <f ≦ f ₄ ) G (f) = a ₅ f + b ₅ (f ₄ <f) (6) The delay time calculator 25 calculates the gain G according to the frequency by the following equation: d (f) = d ₁ (f ≦ f ₁ ) d (f) = d ₂ (f ₁ <f ≦ f ₂ ) d ( f) = d ₃ (f ₂ <f ≦ f ₃ ) d (f) = d ₄ (f ₃ <f ≦ f ₄ ) d (f) = d ₅ (f ₄ <f) Calculate d.

FIG. 2 (d) is a concrete configuration diagram in which the delay time d and the gain G are made variable on the basis of the frequency f of the reference signal xn. 27 is a signal delay unit, which is realized by a tap delay line memory. That is, the signal delay unit 27 includes tap delay lines (memory) for the number of taps (= M), and each storage unit has the reference signal x at the current time.
n, the reference signal xn-1 one sampling time before, the reference signal xn-2 two sampling time before, ... (M-1) The reference signal xn _{-M + 1} before the sampling time is stored. ing. 28 is a data selector unit, and a memory address for storing the reference signal x _nd a calculation unit, which is delayed for the delay time d and said time for calculating a delay time d corresponding to the frequency f of the reference signal xn from (6) It has a correspondence table. The data selection unit 28 delays the delay time d according to the frequency f of the reference signal.
Is calculated, and the reference signal x _{n− d} corresponding to the delay time is read from the signal delay unit 27 and output. 29 is the reference signal xn
Gain calculating unit that the gain G (5) is calculated by equation outputs according to the frequency f of 30 delayed signal x _nd the gain G
Is a multiplication unit that multiplies by and outputs the filtered reference signal rn.

In the embodiment of outputting a plurality of data depending on the frequency, 1 is set based on the frequency f of the reference signal xn.
This is a case where one data (delayed signal x _nd ) is read to create the filtered reference signal rn, but there are the following problems. That is, when the frequency of the reference signal xn frequently changes across the boundary fi (i = 1, 2, ... 5) of the frequency domain, the gain and the delay time change frequently and discontinuously. As a result, the transfer function simulating the secondary sound propagation system becomes discontinuous, the filtered reference signal waveform does not become smooth, discontinuities occur, and noise canceling performance deteriorates. In particular, the problem of discontinuity in delay time is great.

From the above, even if the frequency of the reference signal xn frequently changes across the frequency domain boundary fi (i = 1, 2, ... 5), the delay time and the gain do not change discontinuously. Need to In other words, at any frequency f, it is necessary to generate the filtered reference signal so that the transfer function changes smoothly with the frequency change. In order to satisfy such a requirement, a plurality of delay signals are output based on the frequency f of the reference signal xn, a plurality of gains are calculated and the corresponding delay signals are multiplied, and the multiplication results are combined to produce a filtered reference. The signal rn is output.

FIG. 4 is a block diagram of an embodiment in such a case.
Two delayed signals x based on the frequency f of the reference signal xn
_nd and x _nd-1 (x _nd-1 is a reference signal one sampling time before the reference signal x _nd ) are generated, and two gains G ₀ and G ₁ are generated, and G ₀ · x _nd + G ₁ · Output x _nd-1 as the filtered reference signal rn.
In addition, (a) is a principle configuration diagram, (b) is the frequency f of the reference signal xn
It is a concrete block diagram which makes delay time d and gain G variable based on.

In FIG. 4 (a), 31 is a delay characteristic imparting section for varying the delay time, 32 ₁ and 32 ₂ are variable gain amplifiers, 3
Reference numeral 3 is a delay time calculation unit that determines the delay time d according to the frequency f, 34 is a gain calculation unit that determines the gains G ₀ and G ₁ according to the frequency f, 35 is a delay unit that delays one sampling time signal, 36 is an adder. Delay time calculation unit 33
It calculates a delay time d based on the same way the reference signal frequency f and embodiments of the tooth input to the delay characteristic applying unit 31, the delay characteristic applying unit 31 delays the reference signal xn after a delay time d is the signal x _nd The delay unit 35 further delays the delay signal x _nd by one sampling time and outputs the delay signal x _nd-1 . On the other hand, the gain calculation unit 34 calculates the gains G ₀ and G ₁ based on the reference signal frequency f to obtain the variable gain amplifier 3
Input to 2 ₀ and 32 ₁ . Variable gain amplifier 32 ₀ , 32 _1
Performs amplification of the gains G ₀ and G _{1 on} the delay signals x _nd and x _nd−1 , and the adder 36 adds the outputs of the variable gain amplifiers to obtain a filtered reference signal rn (= G ₀ · x _nd +
G ₁ · x _nd-1 ) is output.

In FIG. 4B, 37 is a signal delay section, which is realized by a tap delay line memory. That is, the signal delay unit 37 includes tap delay lines (memory) for the number of taps (= M), and each storage unit has the reference signal xn at the current time, the reference signal xn−1 before the sampling time, and the two samplings. Reference signal xn-2 before time,
... (M-1) The reference signal _{xd-M + 1} before the sampling time is stored. Reference numeral 38 denotes a data selection unit, which is provided with a calculation unit that calculates the delay time d according to the reference signal frequency f, and a correspondence table of the delay time d and the memory addresses that store the delay signals x _nd and x _nd-1. There is. The data selection unit 38 calculates the delay time d corresponding to the reference signal frequency f, reads the delay signals x _nd and x _nd−1 corresponding to the delay time from the signal delay unit (memory) 37, and outputs the read signals.

Reference numeral 39 denotes a gain calculation unit for calculating and outputting gains G ₀ and G ₁ according to the frequency f of the reference signal xn, and 40,
41 is a multiplication unit that multiplies the delayed signals x _nd and x _nd-1 by the gains G ₀ and G ₁ , and 42 is a filtered reference signal rn (= G ₀ · x _nd + G ₁ · x _{nd- This} is an adder that outputs ₁ ). Multipliers 40 and 41 and adder 4
Reference numeral 2 constitutes a 2-tap filter, which acts so that the gain and phase of the filtered reference signal rn, which is an output, change smoothly. The gains G ₀ and G ₁ are such that the gain and phase of the filtered reference signal rn change smoothly according to the frequency f, and match the actual gain characteristics and phase characteristics (see the solid line) shown in FIG. 3 as much as possible. To decide.

FIG. 5 shows the gain and position of the secondary sound propagation system in the vehicle interior.
It is explanatory drawing of a phase characteristic (solid line) and its approximate characteristic (dotted line).
It Three delayed signals based on the frequency f of the reference signal xn
x_{n- d}, X_nd-1, X_nd-2Along with three
Gain G₀, G₁, G₂Occurs, G₀・ X_nd+ G₁・ X_nd-1+ G₁・ X_nd-2 Output as filtered reference signal rn
Gain G₀, G₁, G ₂By appropriately determining
Gain and phase characteristics of filtered reference signal rn
Can be approximated to the actual gain and phase characteristics (point
See line). For example, the frequency domain f₁<F ≦ f₂Then G₀(f) = a₀f + b₀  G₁(f) = a₁f + b₁  G₂(f) = a₂f + b₂ Based on 3 gain G₀, G₁, G₂Decide
The gain and phase characteristics of the filtered reference signal rn are
As indicated by the dotted line. In addition, in other frequency regions
However, the gain of 2 or more is similarly determined.

General Structure of Filter for Generating Filtered Signal of the Present Invention When transient phenomenon cannot be ignored (when transient phenomenon period is long, frequency fluctuation of reference signal and convergence speed of adaptive filter are fast), transient response component is also to some extent. It is necessary to configure a simulated system (filter for generating a filtered reference signal) in consideration. Therefore, generally, the filtered reference signal rn is generated by using a plurality (a small number) of gains (coefficients) G _{0 to} Gm as shown in FIG. With this configuration, it is possible to reduce the amount of calculation in generating the filtered reference signal and configure a noise canceling system having excellent silencing performance.

FIG. 6 (a) is a principle configuration diagram, and FIG. 6 (b) is a specific configuration diagram. In FIG. 6A, reference numerals 51 _{1 to} 51 m denote delay time variable delay characteristic imparting units, 52 _{1 to} 52 m denote variable gain amplifiers, 53 denotes a delay time calculating unit that determines the delay times d _{1 to} dm according to the frequency f. , 54 are gains G depending on the frequency f
Gain calculation units for determining _{1 to} Gm, and 55 _{1 to} 55m are addition units. In FIG. 6 (b), 56 is a signal delay unit,
It is realized by the tap delay line memory. 5
7 is a data select section, the delay signal delayed by a time d ₁ to dm to response to the reference signal frequency _{f x n-d1 ~x n-} dm
Is read from the signal delay unit 56 and output. Reference numeral 58 is a gain calculation unit that calculates and outputs gains G _{1 to} Gm corresponding to the frequency f of the reference signal xn, and 59 _{1 to} 59m are delay signals x _n-d1.
˜x _n-dm multiplied by gains G ₁ to G m, 60 ₁ ˜
60m is an addition unit.

(B) Noise Canceling System FIG. 7 is a block diagram of a noise canceling system employing the filter for generating the filtered reference signal shown in FIG. 2 (d), which is an example of canceling the engine sound. Reference numeral 71 denotes a waveform shaping unit that shapes the waveform of an ignition pulse IGP generated according to the rotation of the engine, and 72 is a reference signal that generates a second harmonic of the engine speed as a reference signal xn by the PLL method based on the ignition pulse IGP. It is a generator (RG).

FIG. 8 is a block diagram of the reference signal generator, and FIG. 9 is a waveform diagram for explaining the reference signal generation control. Reference signal generator 7
2, 72a is a phase comparator, 72b is a phase voltage converter, 72c is a voltage controlled oscillator (VCO) that outputs a sine wave signal y having a frequency proportional to the input voltage, and the output signal y of the voltage controlled oscillator 72c is It becomes the reference signal xn. The phase comparator 72a outputs a signal (a signal that is at a high level from the rise of u to the zero crossing point of y) v indicating the phase difference between the waveform shaping output u of the ignition pulse and the output y of the voltage controlled oscillator. The phase voltage converter 72b generates a voltage proportional to the phase difference and inputs it to the voltage controlled oscillator 72c, and the voltage controlled oscillator outputs a sine wave signal y having a frequency proportional to the input voltage. By such PLL control, the voltage-controlled oscillator output y coincides with the second harmonic of the engine rotation and is output as the reference signal xn.

Reference numeral 73 is a frequency counter for measuring the frequency of the reference signal, and counts the high-speed clock pulse from the rising edge of the waveform shaping output of the ignition pulse to the next rising edge to measure and output the frequency f of the reference signal. . 74
Is a noise canceling controller that operates to cancel the noise (primary sound) Sn at the observation point, and 75 is a secondary sound propagation system (cancellation that indicates the system in which the cancellation sound propagates from the speaker to the observation point, including the characteristics of the speaker. Sound propagation system), 76 is an error microphone arranged at the observation point, which detects a synthesized sound of the noise Sn and the cancel sound Sc and outputs the synthesized sound signal as an error signal en.

The noise canceling controller 74 receives the second harmonic of the engine sound as the reference signal xn and the error signal en at the observation point.
Filtered X LM to minimize signal power
Adaptive noise signal processing based on the S algorithm is performed to output the noise cancellation signal yn. The noise cancellation controller 74 includes an adaptive signal processing unit 74a that performs adaptive signal processing based on a filtered X LMS algorithm,
An adaptive filter 74b having a digital filter configuration, and a reference signal xn for creating a filtered reference signal rn by convolving a propagation characteristic (transfer function C) of a cancel sound propagation system from a speaker to a noise cancellation point It has a filter 74c.
The filtered reference signal generating filter 74c has the same configuration as the filtered reference signal generating filter shown in FIG. 2D, and the same portions are denoted by the same reference numerals.

The noise generated from the engine propagates through a predetermined primary sound propagation system to reach the noise cancellation point and becomes noise Sn. The second harmonic of the engine speed is output from the reference signal generator 72, and the frequency f of the reference signal xn is measured by the frequency counter and output. The error microphone 15 generates a synthesized sound of the noise Sn at the noise canceling point and the noise canceling sound Sc output from the speaker as the error signal en.
Output as. The signal delay unit 27 of the filtered reference signal generation filter 74c includes tap delay line memories corresponding to the number of taps, and each storage unit has a reference signal xn at the current time, a reference signal xn−1 before one sampling time, and two sampling times. Previous reference signal xn-2, ... (M-1) The reference signal _{xd-M + 1} before the sampling time is stored. The data selection unit 28 calculates the delay time d corresponding to the frequency f of the reference signal xn, reads the reference signal x _nd corresponding to the delay time from the signal delay unit 27, and outputs it. Gain calculator 2
9 calculates and outputs a gain G according to the frequency f of the reference signal xn, and the multiplication unit 30 multiplies the delay signal x _nd by the gain G and outputs the filtered reference signal rn.

The adaptive signal processing unit 74a receives the filtered reference signal rn and the error signal en detected by the error microphone 76, performs adaptive signal processing based on the equation (3), and determines the coefficient of the adaptive filter 74b. . The adaptive filter 74b performs a filtering process on the reference signal xn according to the equation (1) to generate a noise cancellation signal yn and inputs it to a speaker (not shown). The speaker outputs a noise cancel sound based on the noise cancel signal yn. The noise canceling sound reaches the noise canceling point via the secondary sound propagation system 75 and cancels the noise Sc. After that, the above control is performed every sampling time Ts, and the error signal en gradually becomes smaller and the noise is canceled. FIG. 10 is a configuration diagram of a noise canceling system that employs the filtered reference signal generating filter shown in FIG. 6B, and the same parts as those in FIGS. 6 and 7 are designated by the same reference numerals. Since the operation of FIG. 10 can be easily understood from the above description, the description thereof will be omitted.

(C) Modification In the above description, narrow-band periodic noise such as engine sound is assumed as noise, but the present invention can also be applied to the case of canceling narrow-band random noise. FIG. 11 is an explanatory diagram of the configuration of a filter for generating a filtered reference signal of narrow band random noise, where (a) is a conventional FI.
FIG. 3B is a configuration diagram of a filtered reference signal generation filter having an R-type digital filter configuration, and FIG. 3B is a configuration diagram of a filtered reference signal generation filter according to the present invention.

In (a), DL is a delay element that sequentially delays the reference signal (random noise) xn for one sampling time, and ML is a coefficient C _R0 , C _R1 , C _R2 , ...
A multiplier that multiplies C _{RT r−1} , and AD is an adder that sequentially adds the outputs of the multipliers. Coefficients C _R0 , C _R1 , C _R2 , ...
C _RTr-1 is determined so as to simulate the propagation characteristic of the secondary sound propagation system (system from the speaker to the observation point). The stationary response when the narrow band stationary random noise x _n is input to the FIR digital filter can be approximately expressed as x _nd . Therefore, the input / output relationship of the filtered reference signal generating filter can be modeled by a simple system as shown in FIG. 11B, as in the case of periodic noise. That is, it can be modeled by the delay characteristic imparting unit 81 that imparts the delay characteristic of the delay time d to the input signal and the gain imparting unit 82 that applies the gain G to the delay signal.

FIG. 12 is a block diagram of a noise canceling system for canceling narrow band random noise.
The same reference numerals are given to the same portions as. 10 is different from FIG. 10 in that narrow band random noise is input as the reference signal xn and the configuration for measuring the frequency of the reference signal is different. Reference numeral 91 is a bandpass filter unit for dividing the narrow band random noise xn into a plurality of frequency components, and 92 is a frequency for comparing the levels of the respective frequency components and outputting the center frequency of the frequency component having the maximum level as the frequency f of the random noise. It is the decision unit. The data selection unit 57 calculates the delay time d according to the frequency f, and the signal delay unit 56 delays signals x _nd , x _nd-1 , ..., X.
_{The ndm} is read and output, and the gain calculation unit 58 outputs the frequency f.
Determining the gain G ₁ ~Gm according to the multiplication unit 59 ₁ ~59m
Delay signal in an adder 60 ₁ ~60m x _nd,
The gains G _{1 to} Gm are convoluted with x _nd-1 , ..., X _ndm and the filtered reference signal rn is output. Although the present invention has been described above with reference to the embodiments, the present invention can be variously modified according to the gist of the present invention described in the claims, and the present invention does not exclude these.

[0048]

As described above, according to the present invention, the filter for generating the filtered reference signal is composed of the delay characteristic imparting section and the gain varying section, and the delay characteristic imparting section provides the reference signal with the delay characteristic corresponding to the frequency of the reference signal. Since the gain variable unit determines the gain according to the reference signal frequency and outputs the filtered reference signal by applying the gain to the output of the delay characteristic adding unit, the filtered reference signal is generated. It is possible to reduce the amount of calculation executed in one sampling time for performing the above, and thus it is possible to reduce the amount of calculation executed in one sampling time in the adaptive signal processing. Further, by applying the filtered reference signal generating filter of the present invention to a noise canceling system, the amount of calculation can be reduced, and moreover, narrow band periodic noise and random noise can be canceled.

Further, according to the present invention, the filter for generating the filtered reference signal is composed of the delay characteristic giving section, the gain varying section and the signal synthesizing section, and the delay characteristic giving section is different from the reference signal according to the reference signal frequency. A delay characteristic is added to output a plurality of delay signals, the gain variable unit applies a predetermined gain according to the frequency of the reference signal to each delay signal, and the signal synthesis circuit synthesizes the output of each gain variable unit. Since it is configured to output the filtered reference signal, the delay time and gain can be changed smoothly according to the frequency change, and as a result, the filtered reference signal can be changed continuously, and noise cancellation Performance can be improved.

[Brief description of drawings]

FIG. 1 is an explanatory diagram of a filtered reference signal generation filter of the present invention when a reference signal frequency is constant.

FIG. 2 is a configuration diagram of a filter for generating a filtered reference signal of the present invention when the reference signal frequency changes.

FIG. 3 is an explanatory diagram of a gain, a phase characteristic, and an approximate characteristic of a secondary sound propagation system in a vehicle compartment.

FIG. 4 is a configuration diagram of a filtered reference signal generation filter of the present invention when a plurality of delay signals are used.

FIG. 5 is an explanatory diagram of a gain, a phase characteristic and another approximate characteristic of the secondary sound propagation system in the vehicle compartment.

FIG. 6 is a general configuration diagram of a filter for generating a filtered reference signal according to the present invention.

FIG. 7 is a configuration diagram of a noise canceling system of the present invention.

FIG. 8 is a configuration diagram of a reference signal generator.

FIG. 9 is a waveform diagram for explaining reference signal generation control.

FIG. 10 is another configuration diagram of the noise canceling system of the present invention.

FIG. 11 is an explanatory diagram in the case of narrow band random noise.

FIG. 12 is a configuration diagram of a noise canceling system that cancels narrow band random noise.

FIG. 13 is a configuration diagram of a conventional noise canceling device.

FIG. 14 is a configuration diagram of an adaptive filter.

FIG. 15 is a configuration diagram of a filter for generating a filtered reference signal.

FIG. 16 is a configuration diagram of a conventional noise canceling device when there are a plurality of noise sources, speakers, and observation points.

[Explanation of reference numerals] 72 ·· Reference signal generation unit 73 · · Frequency counter 74 · · Noise cancellation controller 74a · · Adaptive signal processing unit 74b · · Adaptive filter 74c · · Filtered reference signal generation filter

Claims (2)

[Claims] 1. A speaker that outputs a cancellation sound that cancels noise at a noise cancellation point, a microphone that detects an error signal that is a combined sound of the noise at the noise cancellation point and the cancellation sound, and a reference according to noise generated from a noise source. A reference signal generator that generates a signal, a filter that generates a filtered reference signal by convolving the reference signal with the propagation characteristics of a cancel sound propagation system from each speaker to a microphone, and noise cancellation by performing a filtering process on the reference signal. An adaptive filter for generating a signal and inputting it to a speaker, and an adaptive signal processing based on a filtered X LMS algorithm using an error signal and a filtered reference signal at a noise canceling point, cancel noise at the noise canceling point. As adaptive filter A noise canceling controller having an adaptive signal processing unit for determining a coefficient, in a filtered reference signal generation method in a noise canceling system including, a filter for generating the filtered reference signal, delay time according to the reference signal frequency The delay characteristic imparting unit that imparts a variable delay characteristic to the reference signal and the gain varying unit that varies the gain according to the reference signal frequency are provided, and the delay characteristic imparting unit applies the delay characteristic corresponding to the reference signal frequency to the reference signal. The filtered reference signal generation method, wherein the gain varying unit applies a gain according to the reference signal frequency to the output of the delay characteristic adding unit and outputs a filtered reference signal. 2. A filter for generating a filtered reference signal further includes a signal synthesizing unit, wherein the delay characteristic adding unit adds different delay characteristics to a reference signal according to a reference signal frequency and outputs a plurality of delay signals. ,
The filter according to claim 1, wherein the gain varying unit applies a predetermined gain corresponding to the frequency of the reference signal to each delayed signal, and the signal synthesizing circuit synthesizes outputs of the gain varying units and outputs a filtered reference signal. Dreference signal generation method.