JPH0573070A - Noise controller - Google Patents

Abstract

PURPOSE:To automatically set the gains of an input and an output in optimum states according to a noise detection gain and the gain of an electro-acoustic transducer by providing a fuzzy inference part which performs control so that the resolution of an A/D converter and a D/A converter enters optimum states for a noise signal and an error signal. CONSTITUTION:Computing elements 8 and 9 detects the maximum value and minimum value of the short-time mean value of the level values of the noise signal and error signal and outputs the results to the inference part 14 which performs fuzzy inference. The inference part 14 infers a dynamic range and an S/N required for the noise signal and error signal according to the input so that they are adapted to the input ranges of A/D converters 10 and 11. The values of gain controllers 5 and 7 are determined by inferring a necessary gain again from the inference results. Further, the value of a gain controller 6 is determined by inferring sound pressure for suppressing a noise so that a signal gain generated by the electro-acoustic transducer 3 is within the dynamic range of the D/A converter 12.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a noise control device using active noise control in an external noise environment invading a small space.

[0002]

2. Description of the Related Art In recent years, an active control method has been proposed in which environmental noise is controlled by outputting a control sound from a speaker using a digital signal processing technique to suppress the noise at a listening position. Conventionally,
This type of noise control device generally has a structure as disclosed in Japanese Patent Laid-Open No. 61-296392. The configuration will be described below with reference to FIG.

As shown in FIG. 4, the noise detection signal detected by the noise detector 41 is input to the convolution processing unit 43 and the control unit 48 through the A / D converter 42. On the other hand, the error signal detected by the error detector 46 is input to the controller 48 via the A / D converter 47. This signal and A /
Based on the output signal of the D converter 42, the control unit 48 updates the transfer function of the convolution processing unit 43 so that the error signal detected by the error detector 46 decreases. This allows
The output of the convolution processing unit 43 is output from the electroacoustic converter 45 via the D / A converter 44, and noise is suppressed by the error detector 46.

[0004]

However, in the above configuration, when performing noise control, the gain of the input / output signal is A / D converted according to the detection gain of the noise detector 41 and the gain of the electroacoustic converter 45. There is a problem that the resolutions of the devices 42 and 47 and the D / A converter 44 must be adjusted in advance in accordance with the system that suppresses noise so that the resolution is optimum.

The present invention solves the above problems and detects the state of the system according to the resolution of the A / D converter and D / A converter as means for controlling the noise control to the optimum gain state. It is an object to propose a noise control device having a function of adjusting input / output gain.

[0006]

In order to achieve the above object, the noise control device of the present invention is a noise control device for actively performing noise control, and is capable of digitally processing a frequency component for actively canceling noise. A / D converter and D used
An inference unit for controlling the gain of the gain controller for controlling the input level of the / A converter is provided. The inference unit controls the gain according to the output of the noise detector that detects noise, and the gain control according to the output of the error detector that detects error noise of the atmosphere that suppresses noise. The second control means and the third control means for controlling the gain in total according to the outputs of both detectors are provided, and the first control means is provided at the input stage of the adaptive filter for controlling the frequency component. A second gain controller that controls the first gain controller that controls the input of the A / D converter, and the second control means that controls the output of the D / A converter provided at the output stage of the adaptive filter. Control the third
The control means controls the third gain controller which controls the input of the A / D converter provided in the error input stage for controlling the characteristic of the adaptive filter.

Further, the inference unit uses the first, second, and third control means so as to match the dynamic range of the A / D converter and D / A converter. The inference means has a fuzzy inference for the gain of the gain controller.

[0008]

The noise control device of the present invention having the above-mentioned structure is an A / D converter and a D / A converter which are indispensable for the digital signal processing technique for controlling the sound emitted from the electroacoustic converter for active noise control. The inference unit automatically controls the input level to the converter to an optimum state with good resolution within its dynamic range according to the outputs of the noise detector and the error detector. Especially, the A / D converter and D
A / A converter has an inference unit for performing fuzzy inference and a gain controller for performing control as a means for controlling the resolution of the A / A converter to be in an optimum state. The present invention proposes a noise control device having a non-adjustment function that does not require the input / output gain to be optimally set.

Here, the fuzzy inference is an inference method that processes variables according to a specific rule (fuzzy rule) and outputs a controlled variable, and the fuzzy rule is a rule that "if ... If ..." is defined as IF (X1 = A AND X2 = B
…) The expression form is THEN (Y = Z). Here, (X1 = A AND X2 = B ...) is the antecedent part,
(Y = Z) is called the consequent part. X1 = A means the variable X1
Means that the degree to which belongs to the membership function A is calculated. The method of outputting the inference result according to the fuzzy rule is a logical AND operation of the degree of belonging to the antecedent part, that is, the operation of selecting the smallest degree of belonging of all the antecedent parts, and the consequent part. On the other hand, the logical sum operation, that is, the largest degree of belonging of all consequent parts is selected to obtain the membership function, and the value of the position of the center of gravity is used as the inference result. By performing such an inference process, it is possible to realize a noise control device that enables adjustment suitable for the situation even if the noise control device is installed in a state with different conditions.

[0010]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 shows an embodiment of the present invention. As shown in FIG. 1, as its constituent elements, 1 is a noise detector, 2 is an error detector, 3 is an electro-acoustic converter for silencing the vicinity of the error detector 2, 13 is an amplifier, and 4 is a noise detector. An adaptive filter for suppressing the noise detected in 1 by the electroacoustic converter 3, 10 is a first A / D converter for converting the noise signal output from the noise detector 1 into a digital signal, and 11 is an error signal Second A / D converter for converting into digital, 12 is a D / A converter for converting the output of the adaptive filter 4 into analog, and 8 is a computing unit for detecting the signal level of the noise signal detected by the noise detector 1. , 9 is a calculator for detecting the signal level of the error signal detected by the error detector 2, and 5 is a first unit for controlling the gain of the noise signal detected by the noise detector 1 and inputting it to the A / D converter 10. Gain controller, 7 is detected by error detector 2 A third gain controller for controlling the gain of the error signal and inputting it to the A / D converter 11, and 6 for controlling the gain of the output signal of the D / A converted adaptive filter 4 and inputting it to the amplifier 13. 2 is a gain controller, and 14 is an inference unit which performs fuzzy inference to determine the optimum gain state of the gain controllers 5, 6, 7 based on the outputs of the arithmetic units 8, 9.

The noise signal detected by the noise detector 1 is input to the signal input of the adaptive filter 4 via the gain controller 5 and the A / D converter 10. On the other hand, the error signal detected by the error detector 2 is input to the error input of the adaptive filter 4 via the gain controller 7 and the A / D converter 11.
The adaptive filter 4 updates the coefficient so as to reduce the error signal input to this error input, and filters the noise signal. That is, in a region where noise is suppressed, a signal that has an opposite phase to the noise that the signal from the electroacoustic transducer 3 has an opposite phase is output. As a result, the adaptive filter 4
Is output from the electroacoustic converter 3 through the D / A converter 12, and noise is suppressed in the vicinity of the error detector 2.

In the above operation, the A / D converter 10,
11 and the D / A converter 12, the dynamic range of conversion is determined by the number of digital bits, and the resolution is determined by the S / N ratio of the converter itself. That is, when the absolute sound pressure of noise cannot be assumed, the A / D converter 10,
It is necessary to set the gain of the noise detector according to the input range of 11. As the method, the inference unit 14 which detects the maximum and minimum values of the short-time average and the maximum and minimum values of the long-term average of the level values of the noise signal and the error signal by the arithmetic units 8 and 9 and performs fuzzy inference Estimates the dynamic range and S / N ratio required for the noise signal and the error signal so as to match the input range of the A / D converters 10 and 11 based on their inputs. The necessary gain is again inferred from those inference results, and the values of the gain controllers 5 and 7 are determined based on the result. Further, the signal gain by which the electroacoustic converter 3 generates a sound pressure for suppressing noise is D / A converter 1.
The value of the gain controller 6 is determined based on the result of inference so as to be within the dynamic range of 2.

An example of the configuration of the inference unit 14 which performs fuzzy inference will be described with reference to FIG. The inference unit 14 has an inference means including a plurality of fuzzy operation units 20 and a plurality of fuzzy inference units 30. The fuzzy operation unit 20 has the same number of membership function circuits (hereinafter abbreviated as MFC) 21, the conditional logical product unit 22, the MFG 23 that generates a membership function corresponding to the consequent unit, and the conclusive logical product unit. 24
have. Variables are input to the MFC 21 of the antecedent part in order to calculate the respective terms of the antecedent part. This operation output is input to the conditional logical product unit 22 and the logical product unit 24. The MFG 23 generates the membership function of the consequent part for inferring the control amounts of the gain controllers 5, 6, and 7,
The result is input to the logical product section 24. Conclusion In the logical product part 24, M
The membership function value input from the FG 23 is cut by the value input from the conditional AND unit 22, and the result is output to the inference unit 30 as an operation value of the fuzzy operation unit 20. The fuzzy arithmetic unit 20 is independently provided by the number of rules provided for controlling the gain controllers 5, 6 and 7, respectively. The inference unit 30 includes an OR unit 31 and a fixed value calculation unit 32.
The arithmetic value input from the fuzzy arithmetic unit 20 is combined by the logical sum unit 31 to determine an ambiguous control amount. This ambiguous control amount is input to the definite calculation unit 32, the center of gravity thereof is obtained, and this value is used as a control amount for the control inputs of the gain controllers 5, 6 and 7 for the first, second and third inference units 14.
It is output as the output of the control means.

FIG. 3 shows the membership function for each variable. FIG. 3A shows inputs from the arithmetic units 8 and 9, NSM.
AX, NLMAX, NSMIN, NLMIN, ESMA
It is a membership function relating to X, ELMAX, ESMIN, and ELMIN and indicates a change in the state of a signal.

On the other hand, FIG. 3 (b) shows the membership function of the consequent part for obtaining the gain control amount. The symbols NL to PL and ENL to EPL for the outputs of the arithmetic units 8 and 9 corresponding to the level values of the noise signal and the error signal respectively indicate the following contents. Noise signal: Short-time average: Maximum value (NSMAX) NL: Small NS: Somewhat small ZE: Normal PS: Somewhat large PL: Large Noise signal: Long-term average: Maximum value (NLMAX) NL: Small NS: Little small ZE: Normal PS: Somewhat large PL: Large Noise signal: Short-time average: Minimum value (NSMIN) NL: Small NS: Somewhat small ZE: Normal PS: Somewhat large PL: Large Noise signal: Long-term average: Minimum value (NLMIN) NL : Small NS: Somewhat small ZE: Normal PS: Somewhat large PL: Large Error signal: Short-time average: Maximum value (ESMAX) NL: Small NS: Somewhat small ZE: Normal PS: Somewhat large PL: Large error signal: Long time Average: Maximum value (ELMAX) NL: Small NS: Small ZE: Normal PS: Large PL: Threshold error signal: Short-time average: Minimum value (ESMIN) NL: Small NS: Somewhat small ZE: Normal PS: Somewhat large PL: Large Error signal: Long-term average: Minimum value (ELMIN) NL: Small NS: Little small ZE : Normal PS: Slightly large PL: Large In addition, the sign of the membership function for the gain controllers G1, G2, G3 of the gain controllers 5, 6, 7 by the first, second, and third control means is as follows. become that way. Gain control amount G1 ENL: very small NL: small NM: slightly small NS: slightly small ZE: as it is PS: slightly large PM: slightly large PL: large EPL: very large gain control amount G2 ENL: very small NL: Small NM: Slightly small NS: Slightly small ZE: As it is PS: Slightly large PM: Slightly large PL: Large EPL: Very large Gain control amount G3 ENL: Very small NL: Small NM: Slightly small ZE: Little small ZE: As it is PS: A little bigger PM: A little bigger PL: Greater EPL: Very bigger As a fuzzy rule for making a fuzzy inference using this membership function, for example,

[0016]

[Table 1]

The following is conceivable. As for the fuzzy rules, other than those shown in this table, various combinations of all variables and inferences can be considered.

[0018]

As described above, the noise control device of the present invention has the resolution of the A / D converter for each of the noise signal and the error signal,
By providing an inference unit for controlling the input level of the D / A converter so that the dynamic range of the D / A converter is in an optimum state, the input / output gain is adjusted according to the noise detection gain, the electroacoustic converter gain, and the like. Therefore, even if the absolute sound pressure level of noise changes significantly and it is necessary to adjust the input level of the A / D converter, there is no need to adjust the gain of the signal. A noise canceller having an adjusting function can be provided.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of a noise control device that is an embodiment of the present invention.

FIG. 2 is a block diagram showing a configuration of an inference unit of the noise control device of the embodiment.

FIG. 3 is a graph showing a membership function used in the inference unit of the embodiment.

FIG. 4 is a block diagram showing a configuration of a conventional noise control device.

[Explanation of symbols]

 1 noise detector 2 error detector 3 electroacoustic transducer 4 adaptive filter 5 gain controller (first gain controller) 6 gain controller (second gain controller) 7 gain controller (third gain control) Device) 8,9 arithmetic unit 10,11 A / D converter 12 D / A converter 14 inference unit

Claims (2)

[Claims] 1. A noise detector for detecting noise, a first gain controller for controlling the signal level detected by the noise detector, and an analog-to-digital conversion for the output of the first gain controller. One A / D converter and the first A / D
An adaptive filter that controls the frequency component of the output of the converter,
D for converting the output of the adaptive filter to digital-analog
/ A converter, a second gain controller that controls the level of the output of the D / A converter, an electroacoustic converter that performs electroacoustic conversion of the output of the second gain controller, and the electroacoustic An error detector installed in a region where the noise is suppressed by the output of the converter, a third gain controller that controls the level of the output of the error detector, and an output of the third gain controller. A second A / D converter that performs analog-digital conversion and controls the characteristics of the adaptive filter; and the first, second, and third
And an inference unit for controlling the gain of the gain controller, wherein the inference unit controls the gain of the first gain controller according to the output of the noise detector, and the error detection unit. A second gain controller for controlling the gain of the second gain controller according to the output of the second controller
And a third control means for controlling the gain of the third gain controller based on the outputs of the noise detector and the error detector. 2. The inference unit comprises a first and second A / D converter and a D
2. An inference means for performing fuzzy inference so that the input to the / A converter is within a range in which the respective resolutions of the first and second A / D converters and the D / A converter are optimum. Noise control device.